docs: write the changelog on dev, not on feature branches

Reverses the "entry goes in with the work" rule from two commits ago. That was wrong on the evidence: of the docs(changelog) commits in history, 3 of 4 in seismo-relay and 2 of 4 in Terra-View were made directly on dev. The rule was generalized from one unrepresentative commit rather than from the pattern. It also caused the exact problem it was supposed to avoid. With four worktrees in flight, every branch edits the same few lines at the top of CHANGELOG.md; feat/ach-rescue-on-connect and feat/sensor-check-h5 collide on that file and nothing else. Writing the entry once, on dev, after the merge removes the whole conflict class. The second benefit is accuracy: an entry written after the merge describes what actually landed, including anything that changed during conflict resolution. The sensor-check branch is a live example — its Unreleased preamble describes a release that no longer looks like that. The failure mode of writing it later is forgetting, so the merge is explicitly not finished until Unreleased is updated — same sitting, reconstructed from the branch commit messages. Unchanged: no preamble under Unreleased, the mandatory operational consequence, and cutting the version on dev when ready to ship to main. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Qcu9ByJfuKBQxmrWb8rSrN
docs: make the release cadence explicit
2026-09-18 18:52:17 +00:00 · 2026-09-18 18:04:54 +00:00 · 2026-09-18 17:39:56 +00:00 · 2026-09-17 06:10:03 +00:00 · 2026-09-17 05:45:26 +00:00 · 2026-09-16 20:54:46 +00:00
56 changed files with 6033 additions and 247 deletions
@@ -4,7 +4,391 @@ All notable changes to seismo-relay are documented here.
 ---
-## [Unreleased]
+## Unreleased
 ### Added
 - **Rescue-on-connect for `bridges/ach_server.py`** — `--stop-monitoring`
  (SUB 0x97), `--disable-ach` (SUB 0x2C read → 0x7E write → 0x7F confirm) and
  `--rescue` (both).  They fire immediately after the startup handshake and
  **before** the event walk, so a unit that is recording back-to-back on a
  stuck-triggered geophone is quieted as early in the session as possible.
  Each action is independently guarded — a failure does not abort the download
  — and the outcome is written to `rescue.json` in the session directory.
  This inverts the `docs/runbooks/wedged_unit_recovery.md` approach.  That
  runbook reaches the unit *inbound* and clears the modem's Destination Address
  to stop it dialing.  When the device is instead wedged mid-modem-init — ALEOS
  logs `tcpmode trying to send to invalid socket` and re-runs `Initialize Auto
  answer` every ~75 s, orphaning any held inbound session — inbound cannot win.
  Pointing the modem's Destination at an `ach_server` and letting the unit call
  *us* gives a device-initiated session the modem bridges properly.
  ⚠ Prefer `--stop-monitoring` alone on first contact.  `--disable-ach` stops
  the unit calling, which is the only channel to a unit in this state; stopping
  the recording ends the call-home loop on its own when ACH is
  "after event recorded".
 - **Blastware-compatible channel FFT (`waveform_fft`).** Reproduces Blastware's
  FFT Report: DC-removed, no window, zero-padded to 4096 (0.25 Hz bins at
  1024 sps), single-sided `2/N` amplitude. Matches Blastware's dominant
  frequency to the exact bin and the amplitude to report precision across all
  28 channels of the 7-event BE12844 oracle set. `channel_spectrum()` /
  `dominant_frequency()`; tests in `tests/test_waveform_fft.py`.
 - **USBM RI8507 / OSMRE compliance chart on the event-report PDF
  (`sfm/compliance.py`).** The velocity-vs-frequency blasting-compliance
  scatter Blastware draws in the upper-right of its Event Report: each channel's
  significant cycles as `(frequency, peak velocity)` points (zero-crossing
  method, so each channel's cloud tops out at its PPV) plotted against the
  RI8507 Drywall (0.75 in/s) and plaster (0.50 in/s) limit curves, drawn
  continuous (constant-displacement bounds meeting the plateaus — no vertical
  steps). Sized and positioned to match a Blastware report, measured off the
  reference PDF. A technical breakdown of the curve is in
  `docs/ri8507_compliance_curve.md`.
 - **Sensor self-check waveforms decoded and drawn (`minimateplus.sensor_check`).**
  The "Sensor Check" traces Blastware shows to the right of the waveform panel
  live in the series-3 binary's trailing block as four length-prefixed records
  (`0x3c`–`0x3f`) using the same delta-block codec as the main waveform:
  Tran/Vert/Long geophone ring-downs (the transducer's damped impulse response —
  resonant frequency + overswing/damping) and a MicL pulse train (the mic's
  known-signal gain check). `gather_report_data` decodes them from the retained
  BW binary at report time; the report renders them as a strip flush against the
  waveform panel plus the **Sensor Check → Frequency / Overswing Ratio** sub-rows
  in the stats table. Verified against the reports on all 7 oracle events (mic
  zero-crossing frequency = 20.1 Hz exact; geophone ring-downs consistent
  ~7.5 Hz with overswing ~3.5). Tests in `tests/test_sensor_check.py`.
 - **Inspector tab in `seismo_lab.py` — annotated hex reader for series-3
  binaries (`minimateplus/binary_annotate.py`).** Tiles a raw Blastware file
  into labeled spans (header / STRT / body record-chain / trailing metadata +
  calibration + sensor-check records / footer) so a binary can be combed by eye.
 ### Fixed
 - **Event-report waveform panel — stacked-lane y-tick collision.** The lanes
  touch, so each lane's bottom `-1.0` overprinted the next lane's top `1.0` at
  the shared boundary. Prune the extreme ticks so each lane shows clean interior
  ticks only.
 - **Event-report header — serial+firmware line ran off the page.** The long
  `BE##### V ##.##-#.## MiniMate Plus` string overflowed the right margin;
  tighter right-column indent + BW's slightly smaller header size so it fits.
 ---
 ### Migration
 **None.**  Every change here is additive and reads from data already on disk —
 the `.h5` samples and the retained raw BW binary.  No `.h5`/DB change, no
 schema change, no migration, no backfill, and **no `TOOL_VERSION` bump**: a
 report regenerated for an existing event simply gains the new panels, and the
 `ach_server` rescue flags don't touch the codec.
 ---
 ## v0.30.0 — 2026-09-12
 **The series-4 correctness release** — the Thor / Micromate counterpart to
 v0.26.0's series-3 work.  The decoder is now verified per-sample against
 Thor's own CSV exports: **459 waveform files, 3,807,158 / 3,807,165 samples
 exact** across three independent ground-truth corpora, and production IDFW is
 **575/575** with zero truncations and zero decode failures.  Series-3
 re-verified **unchanged at 14,338/14,338** after every shared-codec change.
 ⚠ **This release owes the prod store a Thor backfill.**  Every stored
 series-4 geophone value is **3.3% low**, and histogram peaks from monitoring
 runs longer than ~4 hours can be far worse (the interval cap discarded the
 tail, frequently the part holding the peak).  Run
 `scripts/backfill_thor_events.py` — `TOOL_VERSION` is bumped to `0.30.0`, so
 regeneration is gated correctly and **no `--force` is needed**.  DB backup
 first.  Series-3 events are untouched by this release and do not need
 re-running.
 ⚠ **Terra-View displays these values.**  Series-4 geophone readings will rise
 ~3.3% after the backfill, and some histogram PPVs will rise a great deal more.
 That is a correction, not a regression.
 ### Fixed — event-report PDF used a per-trace geo Y scale
 The waveform plot scaled each geo lane to its own peak, so a small channel
 filled its lane and looked as large as a big one, and the `Geo: X in/s/div`
 footer reflected only whichever channel was measured first — wrong for the
 other two.  All three geo lanes now share one symmetric scale (max |sample|
 across them, padded, 0.05 in/s floor), matching the event modal and BW's
 single amp/div; the footer reflects that shared scale.  Mic keeps its own psi
 scale.  Large events are unchanged.
 ### Fixed — series-4 (Thor / Micromate) decoder is now per-sample exact
 Verified against **Thor's own CSV exports**, which carry a per-sample
 four-column block beside every binary (`CSV/<name>.IDFW.csv`) — 1,012 paired
 files that had been sitting in the corpus unused.  Previous notes asserted
 "Thor has no ASCII ground truth", which is why the decoder stayed pinned to a
 superseded walker with an unverifiable scale factor.
 | metric | before | after |
 |---|---|---|
 | IDFW per-sample exact | 39.1% | **100.000%** (1,057,536/1,057,536) |
 | IDFW files fully exact | 0/153 | **153/153** |
 | IDFW PPV median error | −3.32% | **−0.002%** |
 | IDFH within 2% of Thor PPV | 51.1% | **100.0%** (858/858) |
 | prod IDFW PPV median error (8 units) | −3.3% | **−0.001%** |
 | decode cost | — | 6 ms/file |
 Four independent root causes:
 - **Geo LSB was `0.0003`, should be `0.000310308`** — the old value was Thor's
  4-decimal *display rounding* of the LSB mistaken for the LSB, so every
  series-4 geophone sample read **3.3% low**.  Pinned to ±6e-11 by
  intersecting 991,415 rounding constraints; corroborated by the ±full-scale
  seed (`±32226`) in unwritten IDFH slots.  Applies to IDFH too, which had a
  separate (also wrong) `10.0/32768`.
 - **IDFH histograms were capped at 250 intervals** — the segment validator
  required the interval counter's high byte to be zero, but the counter is a
  uint16 cumulative index, so every segment past interval 255 was rejected.
  Any run over ~4 hours lost its tail, often the part holding the peak.
  540/858 corpus files affected.
 - **Record mode `00 00` (raw int16, 10-byte header) was unhandled** — the
  record fell through the dispatch, silently dropping each channel's first
  512 samples.  This produced the long-standing "loud events truncate"
  symptom.  `MODE_ABSOLUTE` is now also accepted as a segment-0 preamble.
 - **Body-offset search matched `00 02 00` inside record headers** — picking a
  candidate part-way down the chain, which decodes a rotation-shifted body
  that drops each channel's segment 0.  The search now anchors on record
  headers and takes the chain head.
 Also fixes the separately-tracked "UM-series decodes ~1000× low" bug
 (`UM11402_20260406130113.IDFW` now matches its device report exactly).
 Series-3 re-verified **unchanged at 14,338/14,338 exact** after the shared
 `waveform_codec` change.
 ⚠ **This is a codec change: the Thor store owes a regeneration.**  Run
 `scripts/backfill_thor_events.py` (bump `TOOL_VERSION` first, or pass
 `--force`), DB backup first.  All stored series-4 `.h5`/sidecar peaks are
 currently ~3.3% low, and histogram peaks for runs over ~4 hours may be
 badly low.
 ⚠ **Thor's histogram PPV has a 0.0050 in/s display floor** — 41.4% of prod
 IDFH sidecars report a component PPV larger than their own vector sum.  On
 quiet files the decoder is now *more* accurate than that reference.
 New: `scratch/verify_thor_against_csv.py`, `tests/test_idf_binary_codec.py`
 (10 tests, fixtures under `tests/fixtures/thor-idf/`).
 ### Fixed — mic-disabled (3-channel) units
 Verified on a second corpus (`9-10-26-csv-req`: UM11402, UM12947, UM20147) —
 **139/139 waveforms per-sample exact (1,273,380 samples), 877/877 histograms
 within 2%** (was 66.9% and 56.6%).
 - **Waveform body head sat below the scan floor.** A 3-channel unit's shorter
  header puts the record chain head at `0x0dba`, under the old
  `_BODY_SCAN_FLOOR` of `0x0E00`.  The scan couldn't see it and fell through
  to the Vert segment-0 record, decoding a body shifted one position around
  the channel rotation — Vert came up exactly 512 samples short.  Floor
  lowered to `0x0C00`; body-offset scoring now accepts 3 channels as "equal"
  instead of demanding 4.
 - **Histogram interval record is 56 bytes, not 72.**  It is
  `16 × n_channels + 8`, so mic-disabled units pack 56.  Assuming 72 read 7
  intervals out of every 10-interval segment then walked off alignment into
  garbage decoding as ~10 in/s peaks (errors up to +191,000%).  The interval
  count now comes from the segment's cumulative counter and the stride is
  derived from it; also recovers 4 files that decoded no intervals at all.
 Combined across both corpora: **292/292 waveform files, 2,330,916/2,330,916
 samples exact.**  Production IDFW truncations 41 → 22.
 ### Fixed — `40 NN` int16 blocks with NN > 8
 `data_block_len()` rejected any `40 NN` block with `NN > 0x08`.  The cap had
 no evidence behind it: every corpus available when it was written used only
 NN ∈ {1,2,3,4,8}, so it was never exercised.  Loud UM12947 events use NN of
 12, 16, 20 … up to 196, and because the block walker stops at the first
 unrecognised tag rather than raising, rejecting them surfaced as **silently
 short channels** (e.g. Tran 1812 / Vert 2132 / Long 2324 on a file whose
 export has 2324 for all three).  The bound is the buffer, not a constant.
 Verified against Thor exports for UM12947 (2025-07-14 … 09-25, 167
 waveforms): length mismatches **22 → 0**, **1,476,242/1,476,249** samples
 exact.  These are not truncated recordings — the exports carry full sample
 counts.
 `tests/test_waveform_codec.py` asserted the cap as intended behaviour; that
 assertion was wrong and has been replaced with one pinning the opposite,
 carrying the evidence.
 ### Result across all three ground-truth corpora
 **459 waveform files, 3,807,158 / 3,807,165 samples exact.**  Production
 IDFW: **575/575**, zero truncations, zero decode failures, median PPV error
 −0.0007% across 8 units.  Series-3 re-verified **unchanged at 14,338/14,338**
 after every shared-codec change.
 The 7 residual samples each differ by one 4th-decimal tick and are **Thor's
 own rounding**: intersecting the per-sample rounding constraints over that
 corpus is infeasible (the binding pair contradict by 2.3e-11, 7e-5 relative),
 so no single linear LSB reproduces every printed value.  `_GEO_LSB_IPS` is
 already pinned to ~1e-11 — do not retune it to chase these.
 ---
 ## v0.29.0 — 2026-09-04
 First release to reach prod since **v0.27.0**, so it ships **both** the
 `false_trigger_reason` column below *and* the v0.28.0 offset (DC-baseline)
 detector: v0.28.0 was version-bumped in-tree (`TOOL_VERSION`, CHANGELOG) but
 never tagged or deployed, so 0.29.0 is the first build to carry either to prod.
 Pairs with Terra-View ≥ 0.24.0. The `false_trigger_reason` column auto-migrates
 on startup; the offset detector still needs the shape backfill on the prod store
 (`scripts/backfill_event_shape.py`) to populate `shape_offset*` on existing rows.
 ### Added
 - **`events.false_trigger_reason` — optional FT cause.** A nullable `TEXT`
  column recording *why* an event is a false trigger (e.g. `"offset"`), as a
  subtype of the FT flag: setting a reason via the sidecar review PATCH implies
  `false_trigger=1`, and the reason is cleared whenever FT ends up 0
  (confirm-real, clear-FT, `set_false_trigger(false)`). `propagate_review_to_twins`
  carries the reason to the histogram/waveform twin alongside the flag.
  Auto-migrated (`_SCHEMA` + `_migrate` ADD COLUMN — not the Migration-1
  rebuild); exposed via `/db/events`. Terra-View surfaces it as a manual
  "Flag as offset" action + an `FT · offset` badge.
 ### Fixed
 - **BlastMate serials — the family prefix is read from the file, not guessed.**
  The Blastware filename encodes only the serial *number* (`L895…` → 10895);
  the two-letter prefix is not in it. `waveform_store` synthesised `"BE"`, so
  an imported **BlastMate** (serials `BA…`) was filed under a MiniMate Plus
  serial that does not exist — silently, and Terra-View read it straight
  through. `save_imported_bw` now resolves serial as hint → file body →
  filename guess, via a new `_serial_from_bw_bytes` that accepts a candidate
  only when its numeric part matches the filename. `client._decode_0a_partial_header`
  likewise matched a literal `b"BE"` in monitor-log partial records; on a
  BlastMate that returned −1 and skipped the whole block, losing the **geo
  threshold** along with the serial. It now matches any two-letter prefix and
  requires the NUL terminator — stricter than the search it replaces.
  BlastMate is the MiniMate Plus's larger Series III sibling and its files are
  byte-compatible: all 1,493 in the DL2 archive decode through the existing
  codec at 100%, same four channels. **The serial string was the only thing
  blocking BlastMate support in SFM.** Four archive units were affected —
  BA9229, BA10060, BA10895, BA15957.
  **No backfill and no `TOOL_VERSION` bump**: this changes which serial an
  *import* is filed under, not any decoded value, so existing sidecars and
  `.h5` files are untouched. **No migration either** — prod holds no BlastMate
  events (the archive's BA units last recorded 2018-10 through 2023-11; the
  prod backfill reaches back only to ~May 2025).
 ---
 ## v0.28.0 — 2026-09-02
 **Offset (DC-baseline) false-trigger detector.** Productionizes the validated
 pre-trigger detector: a geophone event whose baseline sits off zero and stays
 flat across the record (sensor bumped / settled / drifted) is now flagged and
 surfaced in Terra-View as an `offset` false-trigger reason — catching offsets the
 crest/near-peak spike rule misses (an offset is low-crest and flat).
 ### Added
 - `shape_metrics.offset_from_samples` / `offset_from_h5`: per geophone channel,
  `|median(pre-trigger)| ≥ 0.025 in/s` AND `pre/mid/end spread ≤ 0.02` → offset;
  the consistency test rejects transients (a real event moves one third). Reads
  the `.h5` samples + the `pretrig_samples` attr, range-aware via the in/s float
  samples. Constants `OFFSET_FLOOR` / `OFFSET_MAX_SPREAD` are tunable.
 - `events.shape_offset` / `shape_offset_axis` / `shape_offset_pre` /
  `shape_offset_spread` columns (auto-migrated: `_SCHEMA` + the `_migrate`
  ADD COLUMN loop), computed at all three ingest paths and by
  `backfill_event_shape.py`, exposed via `/db/events`.
 Requires the shape/offset backfill on the prod store to populate existing events:
 `python scripts/backfill_event_shape.py --db-path … --store-root …`.
 ---
 ## v0.27.0 — 2026-08-28
 **Per-sample decoder verification at scale, plus the offset investigation.**
 The series-3 codec is now verified sample-by-sample against **14,338** preserved
 Blastware ASCII exports — 1,249 waveform and 13,089 histogram, spanning 45 units
 and files back to 2018. That is 11x the ground truth the production store
 carried, and it found one real codec bug (below).
 ### Fixed
 - **Sub-minute histograms with a partial final block decoded to nothing**
  (`histogram_codec.detect_multi_interval_stride`). The stride search confirmed
  itself on a third block header whenever the body was long enough to hold one —
  but a body can exceed two strides and still contain only two real blocks, because
  a *partial* final block leaves trailing padding. BE18193 `T193L0XM.CI0H` (51
  intervals at 2 s = one full 30-interval block plus a 21-interval remainder, in a
  2787-byte body) therefore had its correct stride of 612 discarded and produced an
  empty decode. A missing third header now means end-of-stream rather than
  disqualification; the block-counter check, which is what actually prevents the
  false positives that once mis-dispatched 9,082 files, is unchanged.
  Found by decoding the full DL2 archive against its preserved Blastware ASCII
  exports. Across **63,535 unique** histogram binaries the fix recovers **4 files** —
  `K440HJCN.3C0H` and `K557IF1U.8K0H` (stride 252), `T191HVNP.0S0H` (92) and
  `T193L0XM.CI0H` (612) — with **zero** files regressed. Verification over all
  14,340 archive pairs goes 14,337 → 14,338 exact, the only remainder being two
  series-4 IDF files that belong to a different codec.
  (The DL2 export keeps a byte-identical `Sent/` mirror of its root, so a naive
  walk double-counts every binary — 127,035 paths are 63,535 distinct files. The
  ASCII exports are *not* mirrored, so the 14,340 pair count is already distinct.)
  **No prod backfill is required for this.** Verified after the fact: all four
  recovered files are archive-only — none exists in the production store or the
  events DB — and re-running stride detection over the production store's
  **10,215** histogram binaries shows **0 files whose decode changes**. The fix
  matters for future ingests of sub-minute histograms with a partial final block,
  not for anything already stored.
  (`TOOL_VERSION` moves with the release, so whenever a backfill *is* next run for
  some other reason it will regenerate the whole store rather than skipping. That
  is harmless — the output is byte-identical for every currently-stored file — but
  it means the run takes its full ~2 hours on the NAS.)
 - **Histogram/waveform twin matching is now interval-based** (`find_twins`). A real
  trigger is recorded twice — as a triggered waveform (stamped at the trigger instant)
  and inside the scheduled histogram whose interval contains it (stamped at the 7am/7pm
  interval start) — so the two twins can be **hours apart**. The old ±5-minute window
  silently missed them, which broke review propagation (flagging one twin didn't flag its
  twin). Twins are now matched by same serial + identical `peak_vector_sum` + opposite
  record type + the waveform falling within the histogram's interval (bounded by the next
  same-serial histogram). `window_seconds` is retained but ignored. Fixes terra-view #102
  sub-task 2.
 - **`/health` reported a hard-coded `0.1.0`** instead of the real service version.
  `sfm/server.py` now derives its version from `minimateplus.event_file_io.TOOL_VERSION`,
  making that constant the single source of truth for the service version and the
  sidecar stamp alike — one place to bump at release.
 - **`CLAUDE.md` had 793 NUL bytes appended** after its last line, which made `grep`
  treat the file as binary and silently skip it. Present since at least v0.21.0.
  Stripped.
 ### Added
 - **`docs/offset_investigation.md`** — a dated journal of the "offset" hardware
  fault: base rate, detector design, per-unit case files, ruled-out hypotheses
  (each kept with the evidence that killed it), and Instantel's own autozero
  procedure with its 2027–2069 acceptance window.
 - **`scratch/verify_against_ascii.py`** — decodes a corpus of BW binaries and
  diffs every sample against the paired `_ASCII.TXT`. Includes a saturation
  carve-out: BW clamps clipped events to the range maximum and writes `OORANGE`,
  while the decoder faithfully reports counts past nominal full scale.
 - **`scratch/offset_scan3.py`** — offset detector. Measures the resting floor in
  the *pre-trigger* window (definitionally quiet) and requires it to hold across
  pre / middle / end. Result: **5 of 45 units (11%)**, stable across a 2x
  threshold range. Supersedes `offset_scan.py` and `offset_scan2.py`, both kept
  as the reasoning trail.
 ### Verified
 - **19,244 healthy channel-events sit at a pre-trigger floor of exactly 0.000
  (62.7%), 94.5% within ±1 quantisation unit, median +0.0000.** No systematic
  zero-point bias in the decoder — an independent confirmation of the
  32000-count geo full scale, arrived at from a different direction than the
  ASCII sample comparisons.
 ---
@@ -2,39 +2,134 @@
 Ground-up Python replacement for **Blastware**, Instantel's Windows-only software for
 managing MiniMate Plus seismographs. Connects over direct RS-232 or cellular modem
-(Sierra Wireless RV50 / RV55). Current version: **v0.26.0**.
+(Sierra Wireless RV50 / RV55). Current version: **v0.30.0**.
 Stack-level context — which repo owns what, and how the three project versions
 pair — lives in `../terra-view/docs/tmi-stack.md`, which is also loaded as
 `~/CLAUDE.md`.
 ---
-## Where things stand (updated 2026-08-27)
+## Where things stand (updated 2026-08-28)
 Read this first when picking the project back up.
- **Series-3 decode is correct and verified.** All 11,603 series-3 binaries in
+- **Series-3 decode is verified per-sample at scale (v0.27.0).** The full DL2
-  the prod snapshot pass every check (channel lengths, peaks vs the device's
+  archive decodes **14,338 / 14,338** paired files exactly against their
-  own reported PPV, nothing above full scale, length vs declared record time).
+  preserved Blastware ASCII exports — 1,249 waveform + 13,089 histogram, 45
-  Ground truth: 1,211/1,211 histograms exact per-interval and 75/75 waveform
+  units, files back to 2018. That is 11x the ground truth the prod store
-  sample counts exact against preserved Blastware ASCII exports.
+  carried, and it supersedes the old "per-sample on 11%, peak-only on 89%"
-  ⚠ That is per-sample proof on 11% of files and peak-only consistency on the
+  caveat. Harness: `scratch/verify_against_ascii.py` (note its saturation
-  other 89% — see `docs/instantel_protocol_reference.md` §7.6.1.
+  carve-out — BW clamps clipped events, the decoder reports true counts).
- **Series-4 (Thor / Micromate) is NOT verified.** UM-series sits at ~48%
+  Independent corroboration of the 32000-count scale: 19,244 healthy
-  against device peaks with a ~1.7% systematic bias and a near-zero tail.
+  channel-events sit at a pre-trigger floor of exactly 0.000 (62.7%), 94.5%
-  Thor IDFW is pinned to `decode_waveform_legacy` deliberately.
+  within ±1 quantisation unit, median +0.0000 — no zero-point bias.
 - **Series-4 (Thor / Micromate) is now verified per-sample (2026-09-10).**
  **1,057,536 / 1,057,536** geo samples across all 153 genuine Thor waveform
  files reproduce Thor's own CSV export exactly; IDFH peaks are within 2% on
  858/858 (median -0.004%).  The ground truth was in the corpus all along —
  Thor writes `CSV/<name>.IDFW.csv` beside each binary with a **per-sample**
  four-column block.  Harness: `scratch/verify_thor_against_csv.py`.
  Four bugs, all fixed: geo LSB was `0.0003` (display rounding of the real
  `0.000310308`, so every sample read **3.3% low**); the IDFH segment
  validator required a zero counter high byte, **capping every histogram at
  250 intervals**; record mode `00 00` (raw int16) was unhandled, silently
  dropping each channel's first 512 samples; and the body-offset search
  matched `00 02 00` *inside* record headers, decoding a rotation-shifted
  body.  IDFW is no longer pinned to `decode_waveform_legacy`.
  Series-3 re-verified unchanged at 14,338/14,338 after the shared-codec
  change.
 - **Mic-disabled (3-channel) units are a distinct shape (2026-09-10).**
  Verified on a second corpus (`~/thor-csv-req`, UM11402/UM12947/UM20147):
  **139/139** waveforms per-sample exact, **877/877** histograms within 2%.
  Two structural differences: the shorter header puts the waveform record
  chain head at `0x0dba` (below the old `_BODY_SCAN_FLOOR` of `0x0E00`, so it
  was invisible and Vert came up exactly 512 short), and the histogram
  interval record is **56 bytes, not 72** — `16 × n_channels + 8`, derived per
  segment from the cumulative interval counter, never assumed.
 - **`40 NN` blocks are not capped at NN=8 (2026-09-11).**  `data_block_len()`
  rejected `NN > 0x08`, a guard with no evidence behind it — the corpora
  available when it was written only used NN ∈ {1,2,3,4,8}.  Loud UM12947
  events use NN up to 196, and since the walker stops at the first
  unrecognised tag rather than raising, this surfaced as silently short
  channels.  Verified on 167 UM12947 waveforms: length mismatches 22 → 0,
  1,476,242/1,476,249 samples exact.
 - **Production IDFW is now 575/575** — zero truncations, zero decode
  failures, median PPV error −0.0007% across 8 units (was 41 truncated + 1
  failing, −3.3%).  Across all three ground-truth corpora: **459 files,
  3,807,158/3,807,165 samples exact**; the 7 stragglers differ by one
  4th-decimal tick and are Thor's own rounding — no single linear LSB can
  reproduce every printed value (the constraints are infeasible by 7e-5
  relative), so do NOT retune `_GEO_LSB_IPS`.
 - **Open, not blocking:** 14 sensitive-range files show an exact 8x
  (= 10.0/1.25) units discrepancy; `scripts/backfill_sidecars.py --force` also
  inserts DB rows for store files that have none (one-time per store) and the
  dry-run does not report that count.
- **After any codec change, regenerate the store** — `backfill_sidecars.py
+- **After any codec change, regenerate the store** — `backfill_sidecars.py`
-  --force` then `backfill_event_shape.py`, DB backup first. Stored `.h5` files
+  then `backfill_event_shape.py`, DB backup first. Stored `.h5` files do not
-  do not update themselves.
+  update themselves. No `--force` needed as long as `TOOL_VERSION` was bumped
- **Parked:** the "offset" hardware-fault investigation (Appendix E of the
+  (it gates regeneration). ⚠ On the office NAS this takes **~2 hours**
-  protocol reference) pending the multi-year BW archive.
+  (~1.5 files/sec vs 85/sec on the dev box — gzip-4 in `sfm/event_hdf5.py`
  against a Synology CPU). Budget it up front.
  **v0.27.0 does NOT owe prod a backfill** — verified: the partial-final-block
  fix changes 0 of the 10,215 histograms in the prod store (the 4 recovered
  files are archive-only and were never ingested).
 - **The "offset" hardware fault has its own journal** --
  `docs/offset_investigation.md`. **5 of 45 units (11%)**, and the fault is
  **persistent** — it stays until the geophone is serviced. Detect it with
  `scratch/offset_scan3.py`: the resting floor in the **pre-trigger** window,
  required to hold across pre/middle/end. Never score only the dominant-peak
  axis and never use the mean — both produce false recoveries (see the
  retraction banner in the journal). Instantel's autozero procedure and its
  2027-2069 acceptance window are recorded there too. Best open lead is
  `SUB 0x0E` (unimplemented), which may carry those very numbers.
 When new information about the protocol is discovered, please update the instantel_protocol_reference.md with the findings in addition to this document
 ---
 ## Changelog & release convention
 **Feature branches do NOT touch `CHANGELOG.md`.  Write the entry on `dev`, as
 part of finishing the merge, under `## Unreleased`.  Cut the version on `dev` in a
 dedicated release commit when you are ready to ship to `main`.**
 - **The changelog is written on `dev`, never on a feature branch.**  With
  several branches in flight they all edit the same few lines at the top of
  the file and conflict every time.  Writing it once, after the merge, also
  lets it describe what actually *landed* — including anything that changed
  during conflict resolution.
 - ⚠ **The merge is not finished until `## Unreleased` is updated.**  Same sitting,
  not "later" — that is the one failure mode of writing it after the fact.
  Reconstruct from the branch's own commit messages:
  `git log --oneline dev..<branch>` before you merge, or
  `git log --oneline <merge-base>..<branch>` after.
 - **No preamble under `## Unreleased`** — just the `### Added` / `### Changed` /
  `### Fixed` lists.  The themed opening paragraph gets written at release
  time, when the whole release is visible and can be named honestly.  A theme
  written when the first item landed is stale by the third.
 - ⚠ **State the operational consequence** on any entry touching the codec, the
  waveform store, or the DB — **including when it is "none."**  "requires
  `backfill_sidecars.py` + `backfill_event_shape.py`, ~2 h on the NAS",
  "`TOOL_VERSION` bumped", "no schema change, no migration".  Silence is
  ambiguous; "none" is information.  This repo's changelog is how future-you
  learns whether a deploy costs two hours.
 - **Releases are cut on judgement, not on a schedule or a merge.**  `Unreleased`
  is the staging area for whatever is going into the next release; when enough
  has accumulated to be worth shipping, it gets a number and a date.  Nothing
  about a merge to `dev` triggers a release.
 - **Cutting a release** is its own `chore(release): vX.Y.Z — <theme>` commit on
  `dev`, renaming `## Unreleased` → `## vX.Y.Z — YYYY-MM-DD` and touching:
  `CHANGELOG.md`, `pyproject.toml`, the version line in `CLAUDE.md` and
  `README.md`, and `minimateplus/event_file_io.py` (`TOOL_VERSION`) **when the
  codec changed** — that constant gates `.h5` regeneration.
 - **`main` carries only released versions.**  No `## Unreleased` section there;
  it lands via the `dev` → `main` PR.  `main` lagging `dev` by a version is
  normal.
 ---
 ## Architecture: three-tier conceptual model
 seismo-relay is a **suite of cooperating components**, not a single app.
@@ -100,20 +195,34 @@ should not import from `sfm/`, must not touch a DB, and have no I/O
 beyond reading files passed as arguments.  Keep them pure — both
 tiers can then depend on them without circularity.
-#### Thor IDF binary codec (2026-05-28)
+#### Thor IDF binary codec (updated 2026-09-10)
 `micromate/idf_file.read_idf_file()` decodes both Thor IDFW
-(waveform) and IDFH (histogram) binaries.
+(waveform) and IDFH (histogram) binaries.  **Verified per-sample
 against Thor's own CSV exports** — see
 `scratch/verify_thor_against_csv.py`.
- **IDFW** reuses `decode_waveform_v2()` on the body at fixed file
+- **IDFW** uses the series-3 record-chain `decode_waveform_v2()`.  The
-  offset `0x0f1f`.  Sample fidelity is 87–99% byte-exact on quiet
+  body offset is **not** fixed: it is `<chain-head record> + 7`, found
-  events; loud events hit the BW codec's known walker-stops-early
+  by `_find_waveform_body_offset()` anchoring on record headers.  All
-  limitation.
+  **153/153** genuine Thor waveform files decode per-sample exact
- **IDFH** has its own segment-based decoder: `[len_be][0a 00 00 00]
+  (1,057,536/1,057,536 samples).
-  [00 NN][05 3f]` + N × 72-byte interval records (4 × 16-byte
+- **IDFH** segment header is `[len_be][0a 00 00 00][counter_be][05 3f]`,
-  per-channel min/max/halfp).  All 859 Thor IDFH corpus files
+  where `counter` is a **uint16 cumulative interval index** — it must
-  decode (181,071 intervals); peak matches sidecar within ~1.8%
+  not be constrained to a zero high byte (that capped histograms at 250
-  (ADC quantization).
+  intervals).  Intervals whose `min > max` on all channels are unwritten
  slots carrying a ±full-scale seed and are skipped.  858/858 files land
  within 2% of Thor's PPV (median -0.004%).
 - **Geo LSB is `0.000310308` in/s per count** (full scale 10.0 in/s =
  32226.05 counts).  Series-3's 32000-count scale does NOT apply.
 - **Record modes** are `02 00` deltas (14 B header), `01 00` absolute,
  `00 03` raw 12-bit, and `00 00` **raw int16** (all 10 B headers).
  `01 00` and `00 00` are also valid as the implicit segment-0 preamble.
 ⚠ **Thor's histogram PPV has a 0.0050 in/s display floor.**  41.4% of
 prod IDFH sidecars report a component PPV exceeding their own vector
 sum — impossible.  On quiet files our decode is *more* accurate than
 the reference; do not "fix" the decoder to match it.
 The two outlier `BE9439_*` files in the Thor example corpus are
 actually Series III Blastware binaries that share the `.IDFW`/`.IDFH`
@@ -379,15 +488,19 @@ with zero mismatches.  Before: 1 of 1196.
  `BE12599/N599LPWJ.980W` @849, `BE9558/K558LOF2.820W` @1485.
 (The series-3 histogram codec was fixed 2026-08-25 — see below.)
- **Micromate (UM-series) IDF decode is ~1000× low** — e.g.
+- ~~**Micromate (UM-series) IDF decode is ~1000× low**~~ — FIXED 2026-09-10.
-  `UM11402_20260406130113.IDFW` gives a Tran peak of 0.0009 in/s against
+  `UM11402_20260406130113.IDFW` now decodes Tran 1.1168 / Vert 4.3220 /
-  a device-reported 1.1168.  The Thor IDF path decodes sanely, so this
+  Long 0.9135, matching the device report exactly.  Root cause was the
-  is UM-specific.
+  body-offset search landing inside a record header plus the unhandled
- **Thor IDF per-count LSB** — after the 32000 geo full-scale
+  `00 00` record mode, not anything UM-specific.
-  correction, series-4 Thor peaks sit at a median 0.983 of the
+- ~~**Thor IDF per-count LSB**~~ — RESOLVED 2026-09-10.  The 0.983 ratio was
-  device-reported peak (was 0.960 under 32768).  Closer but not exact;
+  exactly `0.0003 / 0.000310308`.  Thor's geo LSB is **0.000310308 in/s per
-  Thor likely uses its own per-count LSB rather than the BW
+  count** (full scale 10.0 in/s = 32226.05 counts), pinned to ±6e-11 by
-  16-count/0.005 in/s convention.
+  intersecting 991,415 rounding constraints from Thor's own exports and
  corroborated by the ±full-scale seed (`±32226`) left in unwritten IDFH
  interval slots.  Series-3's 32000-count scale does **not** carry over.
  Note `10.0/32226` is very slightly wrong — see
  `docs/idf_protocol_reference.md`.
 ### Decoded sample counts (across the fixture bundle)
@@ -1803,4 +1916,4 @@ body) because writing a dial string may require DLE escaping for embedded contro
 To parse BW TX captures: use `bridges/captures/` scripts or adapt the `find_write_frames()` pattern
 in `/tmp/analyze_write_payload.py` — it correctly handles `0x10 0x03` DLE-escaped ETX bytes
-inside write frame data (the naive parser terminates early at the escaped `0x03`).
+inside write frame data (the naive parser terminates early at the escaped `0x03`).
@@ -1,4 +1,4 @@
-# seismo-relay  `v0.26.0`
+# seismo-relay  `v0.30.0`
 A ground-up replacement for **Blastware** — Instantel's aging Windows-only
 software for managing seismographs.  Supports both the **MiniMate Plus
@@ -177,6 +177,8 @@ class AchSession:
        store: "WaveformStore",
        clear_after_download: bool = False,
        restart_monitoring: bool = False,
        rescue_stop_monitoring: bool = False,
        rescue_disable_ach: bool = False,
        force_redownload: bool = False,
    ) -> None:
        self.sock                 = sock
@@ -190,6 +192,9 @@ class AchSession:
        self.store                = store
        self.clear_after_download = clear_after_download
        self.restart_monitoring   = restart_monitoring
        # Rescue actions for a runaway unit — fired before the event walk.
        self.rescue_stop_monitoring = rescue_stop_monitoring
        self.rescue_disable_ach     = rescue_disable_ach
        # `force_redownload` tells this session to ignore ach_state and
        # re-download every event currently on the device, regardless of any
        # (key, timestamp) match.  Useful as a manual override when state has
@@ -290,6 +295,41 @@ class AchSession:
        root_logger.addHandler(fh)
        try:
            # ── Step 1.5: rescue actions ──────────────────────────────────────
            # Fired BEFORE the event walk so a runaway unit is quieted as early
            # in the session as possible.  A unit whose geophone sits above the
            # trigger threshold records back-to-back and, with ACH set to "after
            # event recorded", re-dials every time — saturating its own firmware
            # so it never services inbound requests.  See
            # docs/runbooks/wedged_unit_recovery.md.
            #
            # Each action is independently guarded: a failure here must not
            # abort the download that follows.
            if self.rescue_stop_monitoring or self.rescue_disable_ach:
                rescue: dict = {"peer": self.peer, "ts": ts}
                if self.rescue_stop_monitoring:
                    log.info("Step 1.5: RESCUE — stop monitoring (SUB 0x97)")
                    try:
                        client.stop_monitoring()
                        rescue["stop_monitoring"] = "ok"
                        log.info("  stop monitoring OK — device should stop recording")
                    except Exception as exc:
                        rescue["stop_monitoring"] = f"failed: {exc}"
                        log.error("  stop monitoring FAILED: %s", exc)
                if self.rescue_disable_ach:
                    log.info("Step 1.5: RESCUE — disable auto call home (SUB 0x2C/0x7E/0x7F)")
                    try:
                        client.set_call_home_config(auto_call_home_enabled=False)
                        rescue["disable_ach"] = "ok"
                        log.info("  disable ACH OK — unit should stop calling home")
                    except Exception as exc:
                        rescue["disable_ach"] = f"failed: {exc}"
                        log.error("  disable ACH FAILED: %s", exc)
                _save_json(session_dir / "rescue.json", rescue)
            # ── Step 2: device info ───────────────────────────────────────────
            device_info = None
            if not self.events_only:
@@ -747,6 +787,13 @@ def serve(args: argparse.Namespace) -> None:
    print(f"  Max events per session: {max_ev if max_ev else 'unlimited'}")
    print(f"  Clear device after download: {'YES' if args.clear_after_download else 'no'}")
    print(f"  Restart monitoring after download: {'YES' if args.restart_monitoring else 'no'}")
    _stop_mon = args.stop_monitoring or args.rescue
    _dis_ach  = args.disable_ach or args.rescue
    print(f"  RESCUE stop monitoring on connect:  {'YES' if _stop_mon else 'no'}")
    print(f"  RESCUE disable auto call home:      {'YES' if _dis_ach else 'no'}")
    if _stop_mon and args.restart_monitoring:
        print("  !! --restart-monitoring will re-start the unit after download,")
        print("     undoing --stop-monitoring.  Drop one of them.")
    print(f"  Force re-download all (ignore state): {'YES' if args.force_redownload_all else 'no'}")
    print(f"{'='*60}")
    print(f"\n  Point your test unit's ACEmanager call-home settings to:")
@@ -788,6 +835,8 @@ def serve(args: argparse.Namespace) -> None:
                    store=store,
                    clear_after_download=args.clear_after_download,
                    restart_monitoring=args.restart_monitoring,
                    rescue_stop_monitoring=args.stop_monitoring or args.rescue,
                    rescue_disable_ach=args.disable_ach or args.rescue,
                    force_redownload=args.force_redownload_all,
                )
                t = threading.Thread(target=session.run, daemon=True, name=f"ach-{peer}")
@@ -862,6 +911,32 @@ def parse_args() -> argparse.Namespace:
            "DCD on disconnect — without this the unit stays idle after a call-home."
        ),
    )
    p.add_argument(
        "--stop-monitoring",
        action="store_true",
        default=False,
        help=(
            "RESCUE: send SUB 0x97 (stop monitoring) immediately after the "
            "handshake, before any event download. Use on a unit that is "
            "recording back-to-back because of a stuck-triggered geophone."
        ),
    )
    p.add_argument(
        "--disable-ach",
        action="store_true",
        default=False,
        help=(
            "RESCUE: disable Auto Call Home on the device (SUB 0x2C read → "
            "0x7E write → 0x7F confirm) immediately after the handshake. The "
            "unit stops dialing out until ACH is explicitly re-enabled."
        ),
    )
    p.add_argument(
        "--rescue",
        action="store_true",
        default=False,
        help="Shorthand for --stop-monitoring --disable-ach.",
    )
    p.add_argument(
        "--clear-after-download",
        action="store_true",
@@ -6,7 +6,15 @@ Series IV event-file format.  Sibling to
 Series III "Rosetta Stone") — this doc holds what we know so far and
 the open questions still to crack.
-**Status (2026-05-28):** ASCII text sidecar fully decoded (1,014
+> ⚠ **The "Status (2026-05-28)" block below is SUPERSEDED.**  Its geo LSB
 > (0.0003), its IDFH scale (`/32768 × 10`), its fixed body offset (`0x0f1f`)
 > and its "87–99% byte-exact / loud events truncate" caveat were all wrong or
 > incomplete.  See **[Verified against Thor's own exports
 > (2026-09-10)](#verified-against-thors-own-exports-2026-09-10)** — the
 > decoder is now per-sample exact on 1,057,536/1,057,536 samples.  The block
 > is kept only for the reverse-engineering trail.
 **Status (2026-05-28, SUPERSEDED):** ASCII text sidecar fully decoded (1,014
 sample files round-trip).  **Thor IDFW** binary now decodes via
 `micromate.idf_file.read_idf_file()` — reuses the BW segment-rotated
 block codec verbatim at fixed body offset `0x0f1f`; metadata (serial,
@@ -44,6 +52,220 @@ signature and raises `NotImplementedError` pointing callers at
 time-of-peak); the two uint16 fields (probably PVS contributions);
 8-byte interval tail (PVS data); mic dB(L) exact conversion constant.
 ## Verified against Thor's own exports (2026-09-10)
 **The series-4 decoder is now per-sample exact.**  1,057,536 / 1,057,536
 geophone samples across all 153 genuine Thor waveform files reproduce Thor's
 own CSV export exactly; histogram peaks land within 2% on 858/858 files
 (median error −0.004%).
 ### Ground truth — it was there all along
 Thor writes `TXT/`, `CSV/`, `XML/` and `PDF/` exports beside every binary:
 ```
 <serial dir>/UM13981_20220207084555.IDFW
 <serial dir>/CSV/UM13981_20220207084555.IDFW.csv
 ```
 The **CSV carries a per-sample block** — four columns (Tran, Vert, Long, Mic)
 in in/s and psi, after the 2-column report header.  That is the series-4
 equivalent of Blastware's `_ASCII.TXT` exports, and it gives 1,012 paired
 files (152 IDFW + 860 IDFH).  Earlier notes in this file and in
 `micromate/idf_file.py` asserted "Thor has no ASCII ground truth in the
 corpus"; that was wrong, and it is why the decoder sat pinned to a
 superseded walker with a scaling constant nobody could check.
 Harness: `scratch/verify_thor_against_csv.py`.
 ### Geo LSB = 0.000310308 in/s per count (NOT 0.0003)
 The old 0.0003 was read off the smallest non-zero sample in the exports —
 but that is Thor's **4-decimal display rounding of the LSB, not the LSB**.
 It read every series-4 geophone sample **3.3% low**.  The quantisation
 ladder gives it away: counts 1..6 export as 0.0003, 0.0006, 0.0009, 0.0012,
 0.0016, 0.0019 — an LSB of exactly 0.0003 would end 0.0015, 0.0018.
 Each exported sample constrains the LSB to the window that rounds to its
 printed value.  Intersecting 991,415 such constraints gives
 ```
 LSB ∈ [0.000310307933, 0.000310308057]      width 1.2e-10
 ```
 so `_GEO_LSB_IPS = 0.000310308`, i.e. full scale 10.0 in/s = **32226.05
 counts**.  Corroboration: an IDFH interval that never recorded keeps its
 min/max accumulator at its ±full-scale seed, and that seed is
 `(min=+32226, max=-32226)`.  ⚠ The tempting closed form `10.0/32226` is
 very slightly wrong — it lands 4.5e-10 above the feasible window and loses
 78 boundary samples while never winning one.  **Series III uses 32000 counts
 for the same 10.0 in/s, so the two generations do not share a scale.**
 Independently confirmed on 8 production units (UM6047, UM11402, UM11719,
 UM12947, UM13981, UM14133, UM20146, UM20147): every unit's median PPV error
 against its device-reported peak moved from −3.3% to within ±0.03%.  It is a
 global constant, not a per-unit calibration.
 ### IDFH segment header: the counter is a uint16, and it is cumulative
 ```
 [length_be 2B][0a 00 00 00][counter_be 2B][05 3f]
 ```
 `counter` is the **0-based cumulative index of the last interval in the
 segment** — 9, 19, 29, ... for the usual 10-intervals-per-segment layout
 (`length` = 730).
 The validator used to require `counter`'s high byte to be `0x00`.  That
 silently **capped every histogram at 250 intervals**: once the cumulative
 counter passed 255 the high byte went non-zero and every later segment was
 rejected.  Any run longer than ~4 hours lost its tail — frequently the part
 holding the event peak, so the file's PPV read low.  **540 of 858 corpus
 files were affected**; fixing it moved histogram peaks from 48.3% to 93.8%
 within 0.5% of Thor's reported PPV.
 ### Unwritten interval slots carry a ±full-scale seed
 An interval the device reserved but never wrote keeps `min = +32226`,
 `max = -32226` on all four channels — `min > max`, impossible for real data.
 Decoded naively it yields a 10.0 in/s peak on every channel and, being a
 max-over-intervals, poisons the whole file's PPV.  Rare but real: exactly 1
 of 497,611 corpus intervals, and it inflated that file's Long PPV from
 0.0081 to 10.0 in/s.  The inversion is all-or-nothing across channels (0
 partial cases), so requiring every channel to be inverted is a safe test.
 ### Record mode `00 00` — raw int16 absolute (MODE_RAW16)
 The record chain's mode field at `off+8` takes a fourth value:
 | mode | meaning | header |
 |---|---|---|
 | `02 00` | deltas + two int16 anchors | 14 B |
 | `01 00` | absolute, tagged blocks | 10 B |
 | `00 03` | raw 12-bit absolute, untagged | 10 B |
 | **`00 00`** | **raw int16 BE absolute, untagged** | **10 B** |
 A `MODE_RAW16` record with `length = 1032` carries exactly
 `(1032 - 8) / 2 = 512` samples and reproduced Thor's export **512/512
 exactly** on first test.  Thor uses it for segment 0 (the pre-trigger
 window) on some events.  Before this mode existed the record fell through
 the dispatch unhandled, so the channel silently lost its first 512 samples —
 which is what produced the "loud events truncate" symptom.
 `MODE_ABSOLUTE` is also valid as a **preamble** (the implicit segment-0 Tran
 record); its tagged blocks start at `body[3]`, not `body[7]`, because its
 header is 10 bytes rather than 14.
 ### Body offset is not fixed at 0x0f1f — and 0x0f1f is really a record + 7
 A "body offset" is `<record start> + 7`, so that `body[0]` is the segment
 index and `body[1:3]` is the mode.  The canonical `0x0f1f` is simply the
 record at `0x0f18`.
 Searching for the literal preamble `00 02 00` finds only MODE_DELTA bodies,
 and worse, it **matches the `[seg][mode]` bytes inside any record header**,
 so the scan could pick a candidate part-way down the chain.  That decodes a
 plausible-looking but rotation-shifted body which drops each channel's
 segment 0 — the real cause of the remaining truncations.
 `_find_waveform_body_offset()` now anchors on record headers (the
 `<channel_id> 00 00` signature at `+4`, validated with `is_record()`),
 takes the **chain head** — a record no other record's length field points at
 — and trial-decodes `head + 7`, preferring the candidate where all four
 channels come out the same length.
 ⚠ Do **not** scan for candidate preambles instead: `MODE_RAW16` is
 `00 00`, so every run of three zero bytes looks like a body start and each
 costs a full trial decode (~0.5 s/file measured, vs 6 ms/file now).
 ### `40 NN` is not capped at NN=8 (2026-09-11)
 `data_block_len()` rejected any `40 NN` int16 block with `NN > 0x08`.  The cap
 had no evidence behind it — every corpus available when it was written used
 only NN ∈ {1, 2, 3, 4, 8}, so it was never exercised.  Loud events use much
 wider blocks:
 | corpus | `40 NN` values | walker stops |
 |---|---|---|
 | first + 3-channel corpora | 1, 2, 3, 4, 8 | none |
 | UM12947 2025-07..09 | 2, 4, 8, **12, 16, 20 … 196** | every value > 8 |
 Because `walk_body`/`run` stop at the first unrecognised tag rather than
 raising, this surfaced as **silently short channels** — e.g. Tran 1812 /
 Vert 2132 / Long 2324 on a file whose export has 2324 for all three.  The
 real bound is the buffer (and the caller's record end), not a magic constant.
 Verified against Thor's exports for UM12947 (2025-07-14 … 2025-09-25, 167
 waveforms): length mismatches **22 → 0**, and **1,476,242 / 1,476,249**
 samples exact.
 ⚠ These events are **not** truncated recordings, which was the competing
 hypothesis — the exports carry the full sample count.
 **The 7 residual samples are Thor's rounding, not ours.**  Each differs by
 exactly one 4th-decimal tick (e.g. decoded 3.3551 vs export 3.3550).
 Intersecting the per-sample rounding constraints over this corpus is
 **infeasible** — the binding pair (count 2013 → 0.6247, count 4351 → 1.3501)
 contradict by 2.3e-11, i.e. 7e-5 relative.  No single linear LSB can
 reproduce every printed value, so Thor is not doing plain round-half-up on
 `count × LSB`.  Do not retune `_GEO_LSB_IPS` to chase these; it is already
 pinned to ~1e-11.
 ### Mic-disabled units are a distinct shape (2026-09-10, second corpus)
 Some units run with the microphone disabled — **3 channels, not 4** — and that
 changes two structural things.  Confirmed on the `9-10-26-csv-req` corpus
 (UM11402, UM12947, UM20147): 139/139 waveforms and 877/877 histograms.
 **Waveform: the body starts earlier.**  A 3-channel unit has a shorter fixed
 header and puts its record chain head at **`0x0dba`**, below the old
 `_BODY_SCAN_FLOOR` of `0x0E00`.  The head was therefore invisible to the scan,
 which fell through to the *Vert* segment-0 record and decoded a body shifted
 one position around the channel rotation.  The signature is unmistakable:
 ```
 Tran 3072 / Vert 2560 / Long 3072 / MicL 0      <- Vert exactly 512 short
 ```
 46 of 139 files in that corpus were affected; all 46 became per-sample exact
 once the floor dropped to `0x0C00`.  Note the body-offset scoring also had to
 stop requiring four channels — `len(lengths) >= 3`, not `== 4`, or `equal` is
 permanently False for these events and the pick falls back to raw sample count.
 **Histogram: the interval record is 56 bytes, not 72.**
 ```
 interval_size = 16 × n_channels + 8        (72 for 4 channels, 56 for 3)
 ```
 It is **not a constant**, and it cannot be inferred from `length` alone.
 Derive the interval count from the segment counter — it is cumulative, so
 `n = counter - previous_counter` — and then `stride = (length - 10) / n`.
 `n_channels` follows from `(stride - 8) / 16`.
 Assuming 72 read 7 intervals out of each 10-interval segment and then walked
 off alignment into garbage that decoded as ~10 in/s peaks — inflating those
 files' PPV by up to 191,000%.  Fixing it moved the second corpus from 56.6% to
 **100.0%** of histograms within 2% of Thor's reported PPV, and recovered 4
 files that previously decoded no intervals at all.
 ### What is still open
 - ~~23 of 575 production IDFW files~~ — **RESOLVED 2026-09-11.** Production
  IDFW is now **575/575** with zero truncations and zero decode failures
  (median PPV error −0.0007%).  See "`40 NN` is not capped at NN=8" above.
 - Mic → psi scale is still the rough `2.14e-6` regression, not derived.
 - Per-channel `int16 field4` in the IDFH interval record (possibly
  time-of-peak) and the 8-byte tail (PVS data) remain undecoded.
 ⚠ **Thor's histogram PPV has a display floor of 0.0050 in/s.**  In the
 production store 6,080 sidecar PPV values are exactly 0.0050 (next most
 common value: 275 occurrences), and **41.4% of IDFH sidecars report a
 component PPV larger than their own vector sum** — geometrically impossible.
 On those quiet files the decoder's ~0.0025 in/s is *more* accurate than the
 reference; do not "fix" the decoder to match it.
 ### Codec breakthroughs (2026-05-28)
 - **Body offset is a fixed `0x0f1f`** across 151/154 corpus IDFW
@@ -3524,6 +3524,11 @@ of them was, for a while.
 ### E.1 The "offset" fault
 > **The full investigation now lives in `docs/offset_investigation.md`** --
 > base rate, detector definition, per-unit case files, ruled-out hypotheses,
 > and Instantel's own autozero procedure with its 2027-2069 acceptance
 > window.  This appendix is kept as the protocol-side summary.
 **Symptom.**  One geophone channel's baseline steps away from zero and
 stays there.  The trace still carries the real AC signal, but it rides
 on a DC pedestal of a few tenths of an in/s.  Operators call this an
@@ -0,0 +1,987 @@
 # The "offset" fault — investigation journal
 > ## ⚠ CORRECTED 2026-08-28 (same day) — the v1 detector was wrong
 >
 > Brian pushed back on the finding that offsets "come and go": in the field,
 > once a unit develops one it stays broken until the geophone is replaced.
 > He was right, and the challenge exposed **two real flaws** in the v1 detector:
 >
 > 1. **It scored only the axis with the largest peak.** A real event on one axis
 >    hid a persistent pedestal on another. BE12599 on 2026-08-21 read "clean"
 >    solely because Long had a 1.065 in/s event — Tran was sitting at
 >    **+0.4732 in/s** at that moment and was never examined.
 > 2. **It used the MEAN**, which a real transient perturbs. The **median** is the
 >    resting baseline — most samples sit at it, so a blast does not move it.
 >    Same event, Long channel: mean **+0.0783** vs median **-0.0050**.
 >
 > Both flaws manufactured false recoveries. The corrected detector
 > (`scratch/offset_scan2.py`, per-channel median) shows the pedestal is
 > **persistent**, exactly as the field experience says. See §2b and §3b.
 >
 > **Then Brian proposed a better detector still** — measure the floor during
 > the *pre-trigger* window, and require it to hold across pre/middle/end.
 > That is now the detector of record (§2c). Final answer: **5 of 45 units
 > (11%)**, stable across a 2x threshold range.
 >
 > Sections below that were written against v1 are marked; v1 numbers are kept
 > for the reasoning trail, not as current fact.
 A running record of the **offset** hardware fault on Instantel Series III
 seismographs: a geophone channel whose trace sits displaced from zero rather
 than centred on it.
 This is a *journal*, not a spec. Findings are dated, dead ends are kept with
 the reason they died, and every number says where it came from. When something
 here is superseded, strike it and say why rather than deleting it — the point
 is that a future session can tell what was actually established from what was
 merely believed at the time.
 Companion material:
 - `scratch/offset_scan.py` — the detector
 - `scratch/verify_against_ascii.py` — decoder verification harness
 - `docs/instantel_protocol_reference.md` — wire protocol, incl. the
  unimplemented `SUB 0x0E` this investigation now wants
 ---
 ## TL;DR (current state, 2026-08-28)
 - **It is real device data, not a decode bug.** Settled early and confirmed
  against Blastware's own ASCII exports.
 - **Base rate: 5–6 of 45 units (11–13%)** across the full DL2 archive,
  2018–2026. This *confirms* the earlier 2-of-21 (9.5%) estimate from the much
  smaller Terra-View DB — survivorship bias from deleted events had **not**
  concealed a wave of cases.
 - **The fault is bimodal, not a drift continuum.** A unit is either clean or
  grossly off. Loosening the amplitude threshold 11× adds no new units.
 - **The unit's own sensor check cannot see it.** 102 offset events, zero
  sensor-check failures. Do not try to use it as a screen.
 - **Cause is still unsettled.** Instantel's autozero fixes the minority of
  cases; the rest are hardware. We cannot yet tell which is which remotely.
 - **The histogram corpus (63,535 files, 9.7x the waveforms) is now scanned too** —
  see §8b. It independently confirms BE18438 and BE9558 with a clean 2.5x
  separation, but detects only **2 of the 5** confirmed units, cannot attribute a
  channel, and resolves time to ~a month. **A negative histogram result is not
  evidence of health** — DC leakage into the interval peak varies 45x between units.
 - **`offset_scan3.py` has a label defect** (§8b): its spread gate discards 18.8% of
  high-|pre| rows onto units currently counted as clean. Re-cut before quoting any
  precision number again.
 - **Best open lead:** `SUB 0x0E` (channel sensor data, 8 channels × 10 bytes,
  unimplemented) may carry the very numbers Instantel says to check against
  **2027–2069**. Untested.
 ---
 ## 1. What the fault looks like
 A healthy geophone trace is centred on zero. An offset channel is parked away
 from zero, so the channel **mean approaches its own peak**. In Blastware the
 signature is "parallel lines above or below the zero line" (Instantel's own
 wording).
 Consequences observed in the field:
 - The unit can **self-trigger on its own offset** when the displacement exceeds
  the geo trigger level, producing streams of junk events with no ground
  motion. Instantel has a separate FAQ for this symptom (13-0-22, *"Unit
  triggers continuously without activity"*).
 - Recorded PPV for that channel is meaningless while the fault persists.
 ---
 ## 2. The detector
 Implemented in `scratch/offset_scan.py`. Operates on raw BW binaries only — no
 DB, no sidecars.
 ```
 for each series-3 waveform binary:
    decode -> per-channel ADC counts
    dominant axis = channel with the largest |peak|
    flag when   |mean| / peak  >  0.70
          and   |mean|         >= 0.90 x the unit's geo trigger level
 episodes = per-serial runs of flagged events, split on a >12 h gap
 ```
 Why each term:
 | term | purpose |
 |---|---|
 | `\|mean\|/peak > 0.7` | the discriminator. A DC-parked trace has mean ≈ peak. |
 | `\|mean\| >= 0.9 × trigger` | amplitude floor — suppresses quiet traces where mean and peak are both tiny and the ratio is meaningless. |
 | dominant axis only | the fault is per-channel; scoring all three dilutes it. |
 | 12 h episode gap | separates deployments/visits rather than counting events. |
 Trigger level comes from a paired `_ASCII.TXT` when one exists, else the
 per-serial median learned from that unit's ASCII files, else 0.2 in/s.
 **Known limitation.** Event traces contain real ground motion, so this can only
 see offsets large enough to *dominate* the trace. A mild offset on a real blast
 is invisible. Instantel's A/D-mode check (§5) is the only thing that sees the
 mild end. Our base rate is therefore a **gross-offset** rate.
 ### 2b. Detector v2 — per-channel median (CURRENT)
 `scratch/offset_scan2.py`. Supersedes the above.
 ```
 for each series-3 waveform binary:
    for each geo channel independently:
        pedestal = median(samples)          # resting baseline, robust to blasts
        flag the CHANNEL when |pedestal| >= 0.025 in/s   (5 A/D counts)
 a unit has a real fault when a channel is flagged on >=3 CONSECUTIVE events
 ```
 Why median: a DC pedestal shifts every sample, so it moves the median. A real
 event moves only a minority of samples, so it does not. This removes the need
 for the `m/p` ratio guard entirely — that guard existed only to compensate for
 using the mean.
 Why per-channel: the fault is on one geophone axis. Scoring only the dominant
 axis means any event with motion elsewhere hides it.
 Why "3 consecutive": the 0.025 in/s floor is only ~2x a healthy channel's
 resting median (observed 0.010-0.015), so isolated flags are noise. Persistence
 is the discriminator — and it is what the field experience predicts.
 ---
 ## 3b. Archive results, corrected (v2)
 | | v1 (dominant axis, mean) | **v2 (per-channel median)** |
 |---|---|---|
 | units with any flagged event | 6 of 45 | 19 of 45 |
 | **units with a sustained pedestal (>=3 consecutive)** | — | **8 of 45 (18%)** |
 | runs of >=3 consecutive | — | 29 |
 | runs of 1-2 events (noise) | — | 69 |
 Units with a sustained pedestal: **BE9558, BA10895, BE11007, BE11529, BE12599,
 BE13117, BE18003, BE18438**. BA10895 and BE18003 were invisible to v1.
 **The affected channel is most often Vert**, which v1 got wrong — it named
 whichever axis had the largest peak. BE13117 and BE18438 are both Vert faults.
 Longest / clearest runs:
 | unit | ch | span | events | median in/s |
 |---|---|---|---|---|
 | BE13117 | Vert | 2023-05-03 → 05-04 | 194 | 0.035 → **1.915** |
 | BE18438 | Vert | 2026-02-25 → 02-26 | 75 | 0.180 → 0.370 |
 | BE9558 | Vert | 2020-02-11 (6 h) | 33 | 0.065 → 0.090 |
 | BE12599 | Tran | 2026-08-14 → 08-23 | 8 | 0.030 → **0.565** |
 | BE18003 | Vert | 2021-03-17 → 06-11 | 3 | 0.040 → 0.060 |
 BE12599 began **2026-08-14**, not 08-17 as v1 reported, and was still faulting
 at the last event in the archive.
 ### 2c. Detector v3 — PRE-TRIGGER floor + constant-floor test (CURRENT)
 `scratch/offset_scan3.py`. Brian's method, and better than v2 for a reason
 worth naming: **the pre-trigger window is definitionally quiet** — it is the
 buffer captured before the trigger fired — whereas a whole-record median is
 merely *robust* to the event. `pretrig_samples` comes from the STRT record.
 ```
 per channel:
    pre    = median of the first pretrig_samples samples
    mid    = median of the middle third
    end    = median of the final third
    spread = max(pre,mid,end) - min(pre,mid,end)
    offset when  |pre| >= floor  AND  spread <= 0.02 in/s
 real fault when a channel is flagged on >=3 CONSECUTIVE events
 ```
 A DC offset is a **constant floor** — present before the trigger, during, and
 after. The spread test rejects transients (settling, handling, a long event
 tail) that move one segment relative to the others, which is what v2's
 whole-record median could not do.
 **The empirical noise floor justifies the threshold.** Across 19,244
 non-flagged channel-events the pre-trigger floor distributes as:
 | floor | share |
 |---|---|
 | −1 unit (−0.005) | 18.4% |
 | **0.000** | **62.7%** |
 | +1 unit (+0.005) | 13.4% |
 **94.5% within ±1 quantisation unit; median exactly +0.0000, mean −0.0008.**
 So there is **no systematic zero-point bias in the decoder** — an independent
 confirmation of the 32000-count scale. A healthy channel really does read
 0.000, and "any constant floor that is not 0.000" is the right signal, with
 ±1 unit of slack for quantisation.
 **The result is threshold-insensitive**, which is what distinguishes a real
 signal from a tuned one:
 | floor | units flagged | sustained units |
 |---|---|---|
 | 2 units (0.010) | 34 | 15 ← into the noise |
 | 3 units (0.015) | 26 | 8 |
 | **4 units (0.020)** | 17 | **5** |
 | **5 units (0.025)** — Instantel's | 12 | **5** |
 | **8 units (0.040)** | 8 | **5** |
 ### FINAL RESULT: 5 of 45 units (11%)
 **BE9558, BE11529, BE12599, BE13117, BE18438.**
 Unchanged across a 2x threshold range. BE11007 and BA10895 drop out — the
 spread test identifies them as transients, not pedestals.
 The 11% headline happens to match v1's, but the reasoning and the unit list
 differ: v1 included BE11007 and named the wrong *channel* on most units.
 ---
 ## 3. Archive results (2026-08-28)
 Source: DL2 event export, 6,577 **unique** series-3 waveforms, 45 units.
 See [`dl2-archive`](#8-data-and-tooling) for the `Sent/` mirror trap.
 **283 suspect events, 15 episodes, 6 of 45 units (13.3%).**
 Excluding BE11007 (§4, likely not an offset at all): **5 of 45 = 11.1%**.
 ### Threshold sensitivity — the bimodality result
 Re-scoring the same corpus at a range of amplitude floors, with two
 ratio cut-offs (1 A/D count = 0.005 in/s, see §5):
 | \|offset\| floor | m/p > 0.7 | m/p > 0.9 |
 |---|---|---|
 | 5 cts (0.025 in/s) — *Instantel's own* | 333 ev / 6 units | 279 ev / **5 units** |
 | 10 cts (0.050) | 294 / 6 | 274 / 5 |
 | 20 cts (0.100) | 250 / 5 | 244 / 4 |
 | 40 cts (0.200) | 209 / 5 | 203 / 4 |
 | 80 cts (0.400) | 152 / 4 | 148 / 2 |
 | 160 cts (0.800) | 144 / 2 | 141 / 1 |
 Relaxing the floor by 11× (0.27 → 0.025 in/s) adds ~14% more events and **no
 new units**. There is no population of mild offsets hiding below our threshold
 *in event data*. Either a unit is clean or it is grossly off.
 ---
 ## 4. Per-unit case files
 Ordered by severity. `m/p` medians are on the offending channel.
 ### BE13117 — one violent day, never again
 `145 / 454 events (32%)`, **1 episode**, 2023-05-04, 6.8 h.
 Offset climbed **0.393 → 1.875 in/s within the episode**. `m/p` median
 **0.996** — the trace is almost pure DC. No recurrence in the rest of its 454
 events. No ASCII files in the archive, so no calibration history.
 ### BE18438 — recurring, months apart
 `87 / 293 (30%)`, **2 episodes**: 2025-11-15 (1.2 h, n=12, 0.279 → 0.369) and
 2026-02-25 (**28.8 h**, n=75, 0.183 → 0.366). `m/p` median 0.967.
 Clean across all 196 events preceding its 2025-08-12 calibration.
 ### BE9558 — six years apart
 `38 / 196 (19%)`, **4 episodes**: 2020-02-11 (6.3 h, n=33, but only
 0.068 → 0.086 — very mild), then 2026-04-14, 2026-04-29, 2026-05-04
 (0.28–0.45). `m/p` median 0.919. Calibrated 2026-06-26; 0/7 events flagged
 after, but n=7 is far too small to call it fixed.
 ### BE12599 — the live case ⚠
 `6 / 77 (8%)`, **6 single-event episodes, one per day at exactly 05:00**,
 2026-08-17 → 2026-08-23. Offset rose 0.383 → 0.565 then fell back to 0.345.
 `m/p` ≈ 0.965, geo trigger 0.3 in/s — **the offset exceeds the trigger level,
 so the unit is triggering on its own fault**. Last calibrated 2025-08-12.
 This is the most recent and the most useful: a currently-faulting unit is the
 natural experiment for the re-zero-vs-repair question (§7).
 ### BE11529 — marginal
 `4 / 99 (4%)`, 1 episode 2025-07-08, 0.4 h, offsets only 0.051 → 0.058 in/s.
 `m/p` median 0.959, so DC-dominated, but the magnitude is near the noise of
 this method. Treat as unconfirmed.
 ### BE11007 — probably NOT an offset
 `3 / 70 (4%)`, 1 episode 2022-01-17, offsets 5.500 → 6.904 in/s — by far the
 largest. But `m/p` is only **0.719–0.738** against ≥0.9 for every other unit,
 and the peaks are 7.6–9.4 in/s on a 10 in/s range. That reads as a **large
 real blast with asymmetric ground motion**, not a parked trace. Excluded from
 the headline base rate.
 ---
 ## 5. Instantel's own procedure and thresholds
 From two Instantel technical-support FAQs supplied 2026-08-28
 (answers **13-0-21** *"How to determine offsets"* and **12-0-10** *"Removing
 offsets on an Instantel Series III monitor"*; created 2008/2007, last updated
 2009-03-06).
 ### Identifying (13-0-21)
 1. Create or use an event with the **manual minimum trigger** set for the
   connected geophone and microphone — i.e. an event that recorded no real data.
 2. Save it and open in Blastware.
 3. An offset shows as **parallel lines above or below the zero line**.
 4. Put the unit in **A/D mode** — on Series III, press and hold `OPTION`, then
   press `START MONITOR`.
 5. **Display counts higher than 5**, with no vibration or overpressure present,
   indicate an offset.
 ### Removing — the autozero (12-0-10)
 1. Be in a **quiet area with low vibration**.
 2. Power on the Blastmate III / Minimate Plus.
 3. Connect the geophone and microphone — **LINEAR mic only**.
   ⚠ *Do not connect an "A" weight microphone, regardless of what the monitor
   displays.*
 4. Press `Test`.
 5. Wait for the **Sensor Check** results to appear.
 6. Press `OPTION` and `START MONITOR` **simultaneously**.
 7. `Performing Autozero` appears; press `Enter`.
 8. Confirm the sensors are properly connected; press `Enter`.
 9. Wait for the autozero to complete.
 10. Press `Enter` twice → Main Menu, *Ready To Monitor*, offset corrected.
 ### The go/no-go number — 2027 to 2069
 > When you perform an Autozero on any Series III unit, the lists of numbers in
 > the **X1 and X8 gains should all be between 2027 and 2069**. If not, repeat
 > the Autozero. **If the numbers are extremely out of the specified range, then
 > the unit should be sent in for repair.**
 >
 > If this process does not remove the offset problem, return the unit **and
 > sensors** to Instantel for repair.
 This is the documented explanation for the field experience (Brian's dad,
 2026-08-28) that **a re-zero works maybe 10% of the time** — the autozero only
 recovers units whose zero reference is still near-correct.
 ### Scale derivation (inference, well-supported — not proven)
 2048 is 12-bit midscale. Our codec's geo full scale is 32000 internal counts =
 10 in/s, with 1 decoder unit = 16 counts = exactly 0.005 in/s
 (`geo-full-scale-is-32000-counts`). ±2000 A/D counts about 2048 therefore maps
 to ±10 in/s at **0.005 in/s per A/D count**. That makes:
 - Instantel's ">5 counts" threshold ≈ **0.025 in/s**
 - the 2027–2069 window = **±21 counts = ±0.105 in/s** of tolerated zero error
 Consistent and mutually corroborating, but we have not confirmed the A/D-count
 scale directly from a device reading.
 ---
 ## 6. Ruled out — keep these dead
 ### Condensation / humidity — DEAD (2026-08-25)
 Proposed, then killed by its own controls: BE18438 stayed flat across a 10-hour
 overnight gap, and only 2 of 21 units showed the fault while 19 sat in the same
 weather. The apparent "diurnal cycle" was an artifact of binning by hour-of-day
 across two days. See `waveform-dc-offset-is-real-device-data`.
 ### Clipping as a false-positive source — RULED OUT (2026-08-28)
 A rail-hitting trace would fake an offset (mean → peak). It isn't happening:
 median suspect peak is only **10% of full scale**, p90 is 18.6%. Only BE11007's
 3 events exceed 50% FS, and none reach 98%.
 ### The sensor check as a predictor — DOES NOT WORK (2026-08-28)
 Tested on 102 offset events across 4 units:
 | unit | state | n | failed | median ratio | median freq |
 |---|---|---|---|---|---|
 | BE11529 | offset | 4 | **0** | 3.90 | 7.6 |
 | BE11529 | clean | 14 | 0 | 3.80 | 7.5 |
 | BE12599 | offset | 6 | **0** | 4.00 | 7.4 |
 | BE12599 | clean | 13 | 0 | 4.00 | 7.6 |
 | BE18438 | offset | 87 | **0** | 3.70 | 7.6 |
 | BE18438 | clean | 25 | 0 | 3.80 | 7.5 |
 | BE9558 | offset | 5 | **0** | 3.90 | 7.8 |
 | BE9558 | clean | 44 | 0 | 3.80 | 7.5 |
 Zero failures on either side and indistinguishable ratios/frequencies. The
 swing test measures geophone frequency response and damping — it never examines
 DC zero. **A grossly offset unit passes its own self-check.** This is why the
 fault goes unnoticed until somebody looks at waveforms.
 ### "Offsets are transient / come and go on their own" — RETRACTED 2026-08-28
 v1 reported episodes lasting hours that ended spontaneously. **This was an
 artifact of the v1 detector** (see the banner at the top). With the per-channel
 median, the pedestal persists. Every clear case reads clean again only after a
 multi-day-to-multi-month gap consistent with service: BE13117 6 days, BE18438
 24 days, BE9558 63 days **with a confirmed Instantel calibration inside the
 gap**. BE12599 never reads clean — it is still faulting at the end of the
 archive. This matches the operational experience: once a unit develops an
 offset it stays broken until the geophone is replaced.
 ### "Offsets develop N months after calibration" — CONFOUNDED, NOT A FINDING
 Tempting, and it looked strong:
 | unit | suspect before latest cal | after |
 |---|---|---|
 | BE18438 | 0 / 196 | 87 / 97 |
 | BE12599 | 0 / 62 | 6 / 15 |
 | BE11529 | 0 / 82 | 4 / 17 |
 | BE9558 | 38 / 189 | 0 / 7 |
 But bucketing suspects by months-since-calibration gives **one unit per bucket**:
 `0–3mo={BE11529}`, `3–6 & 6–9mo={BE18438}`, `9–12mo={BE9558}`,
 `12–15mo={BE12599}`. The apparent "51% failure rate at 6–9 months" is entirely
 BE18438's single February 2026 episode. Five units with roughly one episode
 each cannot support a population trend. **Do not re-derive this.**
 Also note: all affected units are calibrated on a **~12–13 month cadence**, so
 "sent to Instantel" is the routine annual schedule, not evidence of a
 fault-driven return.
 ---
 ## 7. Open questions
 ### Q1 — Is it a latched bad zero or analog degradation?
 The question that decides everything. A latched zero is correctable (possibly
 over the wire); degradation means a repair. Instantel's 2027–2069 rule implies
 *both* populations exist, with the split roughly 10/90 in the field.
 **BE12599 is the natural experiment** — faulting as of 2026-08-23. Read its
 values, run the autozero, read them again.
 ### Q2 — Can we read the autozero numbers over the wire? (best lead)
 Instantel says to check *"the lists of numbers in the **X1 and X8 gains**"* —
 4 sensors × 2 gains = **8 channels**. The protocol reference already documents
 an unimplemented command with exactly that shape:
 ```
 SUB 0x0E  ->  RSP 0xF1     "channel sensor data"
  2-step read; channel selector in params[6:8] = 0x0000..0x0007
  data length 0x0A (10 bytes) per channel
 ```
 Blastware's *Unit Channel Test* sequence:
 `POLL×N → 0x15 → 0x01 → 0x08 → 0x01 → 0x0E×8 → 0x98×2 → 0x0E×8`
 — note the **second `0x0E` pass carries live ADC readings**.
 **Hypothesis (untested):** `0x0E` returns the numbers Instantel wants compared
 against 2027–2069. If true, SFM could diagnose an offset remotely *and* predict
 whether a re-zero will succeed — converting a 10%/90% shipping gamble into a
 decision made before packing a box.
 **How to test.** `bridges/ach_mitm.py` is a generic TCP proxy:
 ```bash
 python bridges/ach_mitm.py --bw-host <MODEM_IP> --bw-port 9034 --listen-port 9999
 ```
 Point Blastware at the proxy and run **Unit Channel Test**.
 ⚠ In this topology the output filenames are reversed — the tool labels the
 *connecting* side "unit", so `raw_s3_*.bin` holds Blastware's bytes and
 `raw_bw_*.bin` the unit's.
 Capture priority: (1) BE12599 while faulting, (2) a known-good unit as control,
 (3) before/after an autozero on the same unit. Eight 10-byte payloads with an
 expected value near 2048 is a very constrained puzzle.
 ### Q3 — What is the mild-offset rate?
 Unmeasurable from event files (§2). Only the A/D-mode check sees it. Would
 need a fleet sweep in A/D mode, or Q2 to succeed.
 ### Q4 — Does an offset recur on the same unit after service?
 BE9558 shows episodes in 2020 and 2026; BE18438 twice in four months. Suggestive
 of recurrence, but service records aren't in the data — only calibration dates.
 ---
 ## 8. Data and tooling
 | what | where |
 |---|---|
 | detector | `scratch/offset_scan.py` |
 | current results | `/home/serversdown/dl2-archive/offset_archive.csv` |
 | earlier candidate list (Terra-View DB, 274 events) | `scratch/offset_candidates.csv` |
 | archive working copy | `/home/serversdown/dl2-archive/files/` |
 | archive source | NAS `DeathStar` 10.0.0.2, `/volume1/Uploads/TMI/DL2-Event-backup-8-25-26/Event/autocall home/` |
 ⚠ **The DL2 export keeps a byte-identical `Sent/` mirror of its root.** 13,077
 waveform paths are 6,577 distinct files. Always dedupe by basename — this
 doubled two reported figures before it was caught.
 ---
 ## 8b. The histogram corpus — the other 90% of the archive (2026-09-04)
 Every result above §8 comes from **waveform** files. `offset_scan3.py` filters on
 `\.[A-Za-z0-9]{2}0[Ww]$`, so the corpus it scanned is 6,577 unique binaries. The
 archive also holds **63,535 unique histograms** — 9.7x more files — which the
 pre-trigger method cannot touch, because a histogram carries no samples: only a
 per-interval, per-channel peak and half-period.
 `scratch/offset_hist_scan.py` scans them. **63,505 of 63,535 decoded (99.95%),
 43 units, 77.9M intervals.** Two of the 45 units have no histograms at all.
 Output: `/home/serversdown/dl2-archive/offset_hist.csv` (190,515 channel-rows).
 ### The premise, and how far it actually holds
 A histogram file is hours of continuous monitoring, so most of its intervals are
 definitionally quiet, and a channel parked off zero cannot report a peak below
 its own displacement. The signal is real — two within-unit contrasts, siblings
 unmoved in both:
 | unit | channel | in-episode floor | outside | waveform \|pre\| same window |
 |---|---|---|---|---|
 | BE18438 | Vert | 0.0350 | 0.0050 | +0.18 .. +0.37 |
 | BE12599 | Tran | 0.0250 | 0.0050 | +0.03 .. +0.49 |
 But the **leakage from a waveform pedestal into the histogram floor is bimodal,
 not merely partial**: measured ratio ~0.9 on BE18438 Vert, ~0.7 on BE9558,
 **~0.02 on BE12599** — two orders of magnitude on one instrument. The device
 evidently measures each interval peak against a running baseline, and how much
 DC survives that varies per unit. **Consequence: a negative histogram result
 carries almost no information.** Do not read "clean in the histograms" as clean.
 ### The detector that survived
    dmin(file, ch) = min[ch] - min over the other two geo channels, SAME file
    gates (both hard):  n_intervals >= 60   AND   mic_p5 <= 5 raw counts
    day statistic:      median of dmin over that day's qualifying files
    flag day at        dmin >= 0.020 in/s (4 A/D counts)
    episode at         >= 3 CONSECUTIVE observed days
 **Result: BE18438|Vert, BE9558|Tran, BE9558|Long.** Threshold-insensitive —
 the journal's own test for a real signal against a tuned one — and this is the
 first operating point in the investigation that passes it cleanly. The identical
 answer holds across: statistic `min` or `p5`; length gate 10/30/60/120/300; mic
 gate 3/5/8/10; threshold 0.015–0.035 (a 2.3x span); persistence K = 2,3,4,5,7.
 Separation, ranked by highest floor sustained over 3 consecutive gated days
 across all 135 unit-channels:
 | unit-channel | best3 |
 |---|---|
 | BE18438 Vert | 0.1650 |
 | BE9558 Long | 0.0350 |
 | BE9558 Tran | 0.0250 |
 | *(2.5x gap)* | |
 | BE7145 Tran | 0.0100 |
 | entire rest of fleet | <= 0.0050 (one quantisation count) |
 Day-level false alarm: **37 of 99,432 gated unit-channel-days = 0.037%.**
 ### What it does NOT do — read this before trusting it
 - **It finds 2 of the 5 confirmed units, not 5.** The site-quiet gate is what
  makes it work and it is also what costs BE11529 and BE12599. BE11529's
  four-day single-axis ramp (Tran 0.025 -> 0.055, both siblings pinned at 0.005)
  is the most offset-shaped thing in the corpus outside the two detections, and
  the gate discards it.
 - **The positive class is two units.** Every threshold here is fitted to
  BE18438 and BE9558, which contribute 22 of the 37 flagged days in the entire
  corpus. No cross-validation is possible at n=2.
 - **Per-channel attribution is NOT established.** Rotating the three geo channel
  labels within each file — preserving every value, file and day, destroying
  only channel identity — reproduces the episode *count* with p = 0.769 and the
  label agreement at p = 0.038–0.077. Report a **unit and a window**; do not
  name a geophone axis on the strength of this detector alone.
 - **Timing resolution is ~1 month, not ~1 day.** A 30-day label shift still
  scores 2 of 9 episode hits; the signal dies only past ~60 days. The day-level
  series look far crisper than they are.
 - **Ground truth here is a sibling detector, not a service record.** Agreement
  between the two corpora is corroboration of a shared method. Nothing in this
  section has been checked against an actual repair, calibration or RMA.
 ### Dead ends — keep these dead
 - **Absolute floor (min / p1 / p5 / p10 / p25, thresholded alone) — RETIRED.**
  Not fleet-comparable and mostly not about the channel. Scoring each cell using
  *only the other two channels* — a statistic containing zero information about
  the suspect channel — reaches AUC 0.746 against the same labels, versus 0.872
  for the absolute floor itself. **66% of its apparent discrimination is "that
  day was noisy at that site."** Interval size alone moves its p99 7x (0.0350 at
  1 min vs 0.0050 at 2 s). And of all files with any channel above 0.025, 56.5%
  have **all three** channels above it — common-mode, i.e. the wrong physics.
 - **Zero-fraction — STRUCTURALLY IMPOSSIBLE, not merely weak.** The device never
  reports a zero histogram interval peak. The value is a max over hundreds of
  samples of a channel that always carries at least 1 count of noise, so it is
  clamped at 1 A/D count (0.005 in/s). There is no zero to count.
 - **Interval size, sample rate, geo range, firmware — refuted as confounds for
  the differential.** All four are *file-level scalars*: they move all three geo
  channels together, so they cannot produce a single-channel lift and the
  within-file differential is immune to them by construction. Geo range is
  identical across the three geo channels in **63,535 of 63,535** binaries.
  (Interval size remains fatal to the *absolute*-floor version, above.)
 ### Two findings that are independent of the histogram detector
 **1. `offset_scan3.py`'s `spread <= 0.02` gate is discarding real signal.**
 It rejects **113 of the 600 channel-rows with |pre| >= 0.025 (18.8%)**, and the
 rejections are not random — 92 of them fall across 41 unit-channels currently
 labelled NEGATIVE. Four would become sustained positives under an
 amplitude-only >=3-consecutive rule: **BE12599|Long (run of 8), BE18003|Vert
 (4), BA10895|Vert (3), BE12844|Tran (3).** Until this is re-cut, the fleet label
 is **three-state — POSITIVE / NEGATIVE / SPREAD-REJECTED(unknown)** — and the
 third state should be excluded from both TP and FP counts rather than silently
 scored as healthy. Every precision figure computed against the two-state label,
 in this section and in §3, is affected.
 **2. The waveform corpus sees ~7% of the days a unit was deployed.** 2,627
 (unit, day) observations against the histogram corpus's 35,105 — 13.4x — with a
 per-unit median ratio of 0.070. BE12599, a confirmed unit, is waveform-observed
 on 39 of its 1,666 histogram-observed days (**2.3%**). Any statement of the form
 "the fault was absent before date X" that rests on waveform coverage alone is
 much weaker than its event count suggests.
 ### BA10895 — reclassified (see also §4)
 Previously dismissed as a transient. The histogram record shows its **Vert**
 quiet-minute floor at 0.005 on 62/62 qualifying files from 2023-07-07, then
 0.010–0.015 on 48/58 files from 2023-08-03 to 08-27, while Tran moves on 2/58
 and Long on 9/58 and the site mic floor never leaves 1–3 counts. Independently,
 **42 of its 85 waveform events (49.4%) are single-axis-dominant** — one geo peak
 >= 10x both siblings and >= 0.05 in/s — the **highest rate in the 45-unit
 fleet** (BE13117 36.1%, BE18438 29.4%), and **100% of it on Vert**. Vert
 excursions of 0.1–1.5 in/s with Tran/Long at 0.005–0.035 are not ground motion.
 This is a genuine Vert-channel hardware fault, but **not the classic pedestal** —
 the differential is only one A/D count. Caveat: its entire histogram record is a
 single 52-day deployment ending 2023-08-27, so nothing says whether it
 persisted, was serviced, or resolved.
 The other six marginal units — BE11007, BE17354, BE18004, BE18104, BE9557,
 BE18003 — are **clean**. All seven cap at +0.005 to +0.007 (one A/D count)
 lifetime under the quiet-site gate, against +0.175 for BE18438 Vert and +0.062
 for BE9558 Long. Three individual waveform flags fall in windows with **zero**
 histogram coverage and are NO-DATA, not clean: BE18004|Tran 2024-10-16,
 BE9557|Tran 2021-06-28, BE9557|Vert 2025-06-12.
 ### Still open in this section
 - **The 11 thin-coverage units were not screened** (BE10202, BE11462, BE13779,
  BE15760, BA15957, BE16754, BE16758, BE8081, BE8626, BA9229, BE9887 — each
  under 20 waveform events, several with hundreds of histograms). This is the
  population most likely to hold a previously unknown offset, and it is the one
  slice of the plan that did not run. BE11462 was incidentally scored clean by
  the full-archive pass; BE10202 has no histogram files at all.
 - **No completeness audit was run** over the above.
 - Re-cutting the ground truth three-state (finding 1) and re-scoring everything
  against it.
 ---
 ## 8c. Mechanism — five hypotheses tested, all dead (2026-09-06)
 **The mechanism is still unknown.** Five campaigns, ~105 effectively independent
 tests, seven nominally significant results against **5.2 expected by chance**
 under a global null. Every one died to its own confound analysis. What the
 campaign bought is a set of *shape constraints* and a long list of dead ends.
 ### ⚠ Two things retracted from this journal
 **1. "Polarity is perfectly consistent — 11 of 11, zero mixed cases."** That is
 a **tautology of the spread gate**, not a property of the fault. `spread <= 0.02`
 requires pre/mid/end to agree, which forces one sign. Amplitude-only at the same
 0.025 threshold: **12 of 53 unit-channels are mixed**, including BE18438|Vert
 (88+/1−) and BE9558|Vert (1+/35−). Withdrawn.
 **2. "5 of 45 units, unchanged across a 2x threshold range."** The
 threshold-insensitivity is also a property of the gate. Amplitude-only gives
 **9 units at 0.020, 8 at 0.025** (adding BA10895, BE12844, BE18003), 5 at 0.040.
 The fleet is **8–9 units, not 5**.
 **3. "Persistent — it stays until the geophone is serviced."** Weakened, not
 withdrawn. There are **23 recoveries after runs of >=3 flagged events, median
 gap 6.03 days**, three inside ten minutes. BE18438|Vert reads `pre=mid=end=
 +0.0000` on 2026-02-10, +0.185→+0.370 across 02-25/26, and `+0.0000` again on
 2026-03-22 — identical Project, Seis Loc, calibration date, geo range and
 trigger throughout. The one thing that cannot be excluded is a **field
 autozero**: it is a button sequence at the unit and writes nothing into the
 event header. So "persistent" may be "persistent unless somebody pressed the
 buttons," and the archive cannot tell those apart.
 ### The one positive finding: onset is a RAMP, minutes to hours
 Both onsets resolvable at minute cadence are ramps. **BE18438|Vert,
 2026-02-20** — the histogram corpus collapses a 14 d 21 h waveform bracket to
 **one minute**:
 ```
 ~14,200 consecutive quiet minutes at 0.000–0.005   (ten full daily files)
 09:32 +0.005   09:39 +0.045   10:20 +0.125   16:00 +0.165
 09:33 +0.010   09:42 +0.070   13:13 +0.150   20:17 +0.185  plateau
 ```
 **50% of the excursion in 7 minutes**, the rest asymptotic over ~10 h, **>=25
 distinct one-minute intermediates**. Validated **75/75** against Blastware's own
 ASCII export. Its 2025-11-15 onset is the same shape over 2.7 h. BE13117 stage B
 is a 91-minute monotone rise, +0.035 → +1.745 in/s over ~40 samples.
 **This kills both poles of the original dichotomy** (journal Q1): not an
 instantaneous latched step (a bad autozero, a stuck trim-DAC), and not slow
 component degradation over days or weeks.
 ⚠ It rests on **2 of 45 instruments**. Clopper-Pearson on 4/4 resolved onsets
 gives 95% CI [0.40, 1.00] — a mixed population with up to 60% true steps is not
 excluded. BE13117 has zero paired ASCII, so its ramp rests on our decoder alone.
 ### The methodological corollary — more important than the finding
 **A waveform-only bracket manufactures the appearance of a step, and the spread
 gate is blind to onsets by construction.**
 The offset is what fires the trigger, so no waveform event can exist until the
 ramp has nearly reached the trigger level. BE18438's first event of each episode
 sits at 0.280 against a 0.300 trigger, and 0.185 against 0.200. At daily cadence
 against a 3 h ramp, P(catching an intermediate) = **0.125**.
 And `spread <= 0.02` rejects any record in which the floor is *moving* — which
 is exactly what an onset is. **The gate rejected the very BE18438 record where
 the ramp is visible.** If the operational goal is catching a fault early, before
 the unit floods the store with junk events, the current detector is the wrong
 shape for the job.
 ### The surviving shape
 An **electrical, reversible, two-time-constant settling process** (~10 min and
 ~hours), saturating at a ceiling, with occasional sub-3-minute discrete jumps
 superposed (BE18438 2026-02-26: 13:24 pre +0.180 / mid +0.240 / end +0.255 →
 13:27 +0.325, identical metadata). That is the signature of a **bias or leakage
 path charging a high-impedance node** — the class of fault Instantel's autozero
 recovers ~10% of the time, and what the X1/X8 gains measure.
 **It is a shape constraint, not a mechanism. Do not write it up as one.**
 ### Dead — with the evidence, so none of this is re-derived
 | Killed | Evidence |
 |---|---|
 | **Latched step at onset** | >=25 one-minute intermediates over ~10 h, ASCII-validated. Direct observation, not a test. |
 | **Slow degradation over days/weeks** | Same observation — bulk of the excursion in 7 min to 2.7 h. |
 | **Thermal driving of pedestal magnitude** | BE13117, 365-count pedestal, n=128: full-day modulation **−0.42% ± 0.42%**, 95% CI [−1.25%, +0.40%]. Healthy-fleet seasonal zero drift totals **~0.3 A/D counts** — 15x to 1200x too small. Best-powered result in the campaign. |
 | **Ground-motion shock** | 30-day window-max percentile ranks 0.03/0.98/0.15/0.01/0.68/0.24, median **0.194** against a null of 0.5. **0 of 7 events >=9 in/s** was followed by an onset within 30 d. BE12599 hit 10.220 in/s (2023-11) and 10.005 (2025-04) and did not onset until 2026-08-14. |
 | **Handling / redeployment** | **0 of 9** onsets had a Project/Client/Seis Loc change. Widened to 30 d: 2 observed vs 4.90 expected, P(X>=2)=0.995 — *depleted*, the wrong direction. The apparent gap effect (p=0.035) died on histogram coverage: BE18438's "59.7-day gap" contains 122 histogram files; true silence 0.52 d. |
 | **Mechanical resonance / damping change** | BE18438|Vert at a 64-count pedestal (3x outside Instantel's ±21): ΔTest-Freq **CI [−0.090, +0.021]** against 0.127 Hz for a real calibration. Block permutation p=0.658. |
 | **Accumulated-duty threshold** | ~4 clean units logged more monitoring than the largest positive onset dose; BE18193 logged **13.45M intervals, 6.2x**. A counterexample — no power argument weakens it. |
 | **Firmware** | **14,338 of 14,340** exports read `V 10.72-8.17`. A constant cannot explain a variable. |
 | **Unit age** | Serial rank-sum 118.0 vs null 115.0, p=0.549; unchanged on the 8-unit re-cut (p=0.586). Serial is a poor age proxy anyway (Spearman +0.113 against archive entry). |
 | **Strong seasonal clustering** | 25 onsets, exposure-weighted permutation **p=0.59**. Excludes >=75%-in-one-season only; a 2x seasonal hazard is *not* excluded. |
 Also retire two overstated bounds. H6's dose-response exclusion "|r| > 0.03" is
 a **10x overstatement** once clustering is corrected — the honest bound is
 |r| > 0.1–0.3, so a real r=0.2 is not excluded. And **any statistic quoted
 per-event**: 512 flagged channel-events collapse to **4.9 effective independent
 observations** (unequal-cluster design effect 104.6 at ICC=1), and **55% of the
 flagged corpus is one instrument on two calendar days** (BE13117, 2023-05-03/04).
 ### Power — read every negative in this section as bounded
 Fisher exact, 5 positives of 45, one-sided α=0.05, exposure a third of the fleet:
 | relative risk | power |
 |---|---|
 | 1.5 | 0.059 |
 | 2 | 0.112 |
 | 3 | 0.231 |
 | 6 | 0.497 |
 | 15 | 0.753 |
 80% power needs **RR ≈ 13–20**. Even a *perfect* split reaches p<0.05 only if
 the exposed group is <=25 of 45 units. **This archive can detect only
 near-deterministic unit-level causes.** Every negative above excludes a strong
 effect, not a real one.
 ### What this archive can NEVER answer
 - **The A/D zero and the X1/X8 gains.** The 2027–2069 numbers appear in no file,
  header or decoded record. They exist only on a live device behind `SUB 0x0E`.
  Q1 is structurally unanswerable from data.
 - **Unit-level vs component-level cause.** **Zero of 14,340** exports carry a
  geophone or sensor serial. Q4 is dead — there is no way to know whether the
  same physical geophone came back after service.
 - **Service history.** The only service-adjacent field is `Calibration: <date>`
  — 30 distinct dates fleet-wide, none before 2023, ASCII corpus entirely
  2025–26. BE9558's 2020 and BE13117's 2023 episodes have no calibration record.
 - **Temperature.** Zero exports carry it. Battery Level is a verified coarse
  thermometer (+0.204 V winter over summer, 20/20 unit-years, p=9.5e−7, matching
  lead-acid tempco) but quantised at 0.1 V ≈ 10 °C — useless within a day. The
  archive can *bound* thermal; it can never *test* it.
 - **BE13117 specifically** — 55% of the flagged corpus, the largest pedestal at
  1.92 in/s, **zero** ASCII exports, histogram record ending eight months before
  its episode. The most informative case in the archive is permanently outside
  every metadata test.
 - **The mild-offset rate**, and therefore the base rate's denominator. Event
  files only see offsets large enough to dominate the trace.
 ### The experiment to run — `SUB 0x0E`, one afternoon
 Point Blastware at `bridges/ach_mitm.py` and run **Unit Channel Test** against
 (1) a faulting unit, (2) a known-good control, (3) the same unit before and
 after an autozero. BW's sequence is `0x0E x8 → 0x98 x2 → 0x0E x8`, the second
 pass carrying live ADC. Eight 10-byte payloads with expected values near 2048 is
 a very constrained puzzle.
 - **Proves:** whether the X1/X8 gains are readable over the wire, and whether
  the fault sits at or upstream of the ADC zero reference. Gains walk out of
  2027–2069 with the pedestal → the fault *is* the zero reference, Q1 answered.
  Gains hold while the trace moves → the fault is downstream, look at the front
  end.
 - **§8c hands it a falsifiable time course:** poll at ~1-minute cadence and the
  numbers should **ramp over minutes-to-hours, not step**. If they step while
  the trace ramps, the two are decoupled.
 - **Payoff:** converts the 10%/90% ship-it-or-not gamble into a decision made
  before packing a box, remotely, for the whole fleet.
 - ⚠ In the MITM topology filenames are reversed — `raw_s3_*.bin` holds
  Blastware's bytes.
 **Second: swap the geophone** between a faulted base and a healthy one. Fault
 follows the sensor → element or cable. Fault stays with the base → front-end
 board. One afternoon, zero code, and it settles the one question the archive is
 permanently blind to.
 **Third: log a faulting unit for 72 h untouched.** Every recovery we have is
 confounded by a possible field autozero. A shelf and a logger settles whether
 the fault genuinely self-reverses.
 **Fourth, free: re-cut the fleet label** — drop the spread gate, re-score
 amplitude-only, screen the 11 unscreened thin-coverage units. Might reach 9–10
 positives. Be honest about the gain: power against "older half carries 3x the
 hazard" rises only 0.23 → 0.30.
 **Highest-value item overall, and not an experiment: the RMA/repair records.**
 Which unit went back, when, what was done (autozero vs geophone replaced vs
 board), and the geophone serial fitted. "Same channel after a documented
 geophone *replacement*" is component-level-negative in one observation.
 ---
 ### 8d. The non-motion test — Brian's "it doesn't cross zero" (2026-09-07)
 Looking at BE12599's 2026-08-09 event, Brian noted it reports no ZC frequency
 **because the trace never crosses zero**. That observation is the best detector
 in this investigation, and it comes from physics rather than a threshold.
 A geophone is a velocity sensor with no DC response, so its output over a record
 must integrate to ~zero — the ground does not relocate. Real motion therefore
 sits roughly half below zero. Anything electrical is one-sided.
    mp = |mean| / peak        ~0 for motion, ~1 for a fault
    frac_neg = share of samples < 0
 `scratch/nonmotion_scan.py`, all 6,577 waveforms, 19,731 channel-rows.
 Restricted to peak >= 0.05 in/s (n = 12,068), the distribution is **bimodal
 with an empty middle**:
 | mp band | channel-events |
 |---|---|
 | 0.0–0.1 | 11,384 |
 | 0.1–0.2 | 293 |
 | **0.15–0.85 (dead zone)** | **131 = 1.09%** |
 | 0.9–1.0 | 278 |
 At `mp >= 0.8` with >=3 events it returns **exactly the five confirmed units** —
 BE9558, BE11529, BE12599, BE13117, BE18438 — stable from 0.5 to 0.9. Two
 detectors on entirely different principles agreeing on the unit list is the
 strongest corroboration that list has.
 **BE11007 is settled: NOT an offset.** It reaches mp 0.75–0.89, but with
 `frac_neg = 0.99` at peaks of **7.4–9.4 in/s** — parked *negative* during a
 near-full-scale blast. §4's guess was right. `mp` alone cannot separate a
 pedestal from a large one-sided blast; pair it with a peak ceiling or with
 sign-consistency across events.
 ⚠ **Not a rediscovery of the retracted v1 detector.** v1 scored only the
 largest-peak axis and used the mean as a *baseline estimator* where the median
 was required. Here the mean is the signal itself, per channel — that is what the
 physics licenses.
 **Correction to §8c.** That section says the spread gate is "blind to onsets by
 construction." Too strong: of 87 BE18438|Vert events at mp >= 0.5 the gate
 rejected **one** — the transitional record. It does not lose onsets
 systematically; it loses the transition specifically.
 ### 8e. BE12599 — a connector, not a geophone (2026-09-07)
 Waveform shapes across its August episode, measured rather than eyeballed:
 | date | channel | shape |
 |---|---|---|
 | Aug 09 05:29 | Long | **unipolar +**, 0/2304 samples below zero, decay tau **26 ms** |
 | Aug 09 05:35 | Long | unipolar +, 3 spikes at irregular gaps (744, 1032 ms), tau **38 ms** |
 | Aug 14 05:00 | Long | single lobe, bipolar, tau **118 ms** |
 | Aug 17–23 | Tran | **flat DC pedestal**, sd/level 0.015–0.020, 0 zero crossings |
 **Unipolar impulses with an RC tail are not mechanical.** Fast rise, exponential
 decay, one polarity, irregular timing — that is charge dumped into a
 capacitively-coupled input and draining through the input resistance. The
 progression 26 ms -> 118 ms -> never recovers, over 14 days, is a leakage path
 worsening.
 **And the fault moved channels** — Long on Aug 9/14, Tran on Aug 17–23, Long
 again on Aug 21 (1.065 in/s) while Tran held its pedestal. Vert stayed clean
 throughout. **A failing geophone element cannot hop channels. A connector can.**
 That single fact explains what had been puzzling:
 - **The sensor self-check keeps passing** (7.4/7.5/7.6 Hz, ratios 3.6–4.2, all
  four channels Passed, on the very events where Long throws 0.5 in/s spikes).
  The swing test drives the element; the element is fine. The fault is in the
  wiring to it.
 - **Why Instantel's autozero fixes only ~10%** — it cannot fix a connector.
 - **Why onset "ramps" over minutes to hours** — contact resistance drifting.
 All seven Aug 17–23 events are stamped **05:00:14**, the same second, and their
 filename extensions run `8E → WE → KE → 8E → WE → KE → 8E` — the documented
 3-day cycle for a fixed daily time. Clock-scheduled, not physically triggered:
 the modem powers up, draws a surge, and a marginal connection responds.
 **Field action: inspect and photograph the geophone connector BEFORE reseating
 anything** — an intermittent contact clears the moment it is disturbed.
 ⚠ Scoped to BE12599. BE18438's onset was a smooth 7-minute ramp with no spikes,
 which looks like a different failure mode wearing the same signature.
 ---
 ### ⚠ Serial prefixes — four of these units are BlastMates, not MiniMates
 Corrected 2026-09-06, after Brian queried "BA10895?" against a report that
 said BE10895. He was right. The BW filename encodes the serial **number
 only** — `L895` -> 10895 — and every offset scanner synthesised the family
 prefix as `"BE"`. Four of the 43 archive units are **BA** (BlastMate, the
 MiniMate Plus's bigger sibling; same Series III, byte-identical data):
 **BA9229, BA10060, BA10895, BA15957.**
 Read off the file bodies, which carry the serial verbatim. No analysis
 changed — grouping was always on the numeric part, and no unit number maps
 to two serials — but every earlier reference to "BE10895" and the other
 three is a label error and has been corrected throughout this document.
 The same assumption was live in two production sites and is fixed
 (`sfm/waveform_store.py`, `minimateplus/client.py`): the store would have
 filed a BlastMate under a unit that does not exist, and the monitor-log
 decoder lost the geo threshold along with the serial. See commit `9ceff65`.
 ---
 ## 9. Chronology
 | date | event |
 |---|---|
 | 2026-08-25 | Reported as a *waveform decode bug* — traces with a DC offset. Investigation shows the offset is **real device data**; the decoder is correct. |
 | 2026-08-25 | Brian relays his dad's description: a known hardware fault called an "offset"; usually sent to Instantel. |
 | 2026-08-25 | First detection pass over the Terra-View DB: **2 of 21 units**, 274 events, 11 months. Flagged as vulnerable to survivorship bias — flooded events were routinely deleted. |
 | 2026-08-25 | Condensation hypothesis proposed, then **killed by its own controls**. |
 | 2026-08-25 | Parked pending the multi-year archive. |
 | 2026-08-28 | DL2 archive pulled (33 GB, 546k files; 6.6 GB working set). |
 | 2026-08-28 | Archive scan: **6 of 45 units**, 283 events, 15 episodes. Prior base rate **confirmed**, not overturned. |
 | 2026-08-28 | Clipping ruled out; `m/p` established as the discriminator; BE11007 reclassified as probably a real blast. |
 | 2026-08-28 | Calibration-timing correlation attempted and **rejected as confounded**. |
 | 2026-08-28 | Instantel FAQs supplied: autozero procedure, the **2027–2069** window, the **>5 counts** threshold. Explains the ~10% re-zero success rate. |
 | 2026-08-28 | Bimodality established; sensor check proven **blind** to offsets; `SUB 0x0E` identified as the best open lead. |
 | 2026-08-28 | **v1 detector retracted.** Brian challenged the "come and go" finding against field experience. Two flaws found: dominant-axis-only scoring and mean-instead-of-median. Corrected detector shows persistent pedestals on **8 of 45 units**, and the gaps are service windows. |
 | 2026-08-28 | **Detector v3 (Brian's method):** pre-trigger floor + pre/mid/end consistency. Healthy channels proven to sit at 0.000 +/-1 unit (94.5%), confirming no decoder zero-point bias. Final: **5 of 45 units (11%)**, threshold-insensitive. |
 | 2026-09-04 | **Histogram corpus scanned** — 63,505 of 63,535 files, 43 units, 77.9M intervals (9.7x the waveform corpus). `scratch/offset_hist_scan.py`. |
 | 2026-09-04 | Absolute-floor statistic **retired**: 66% of its discrimination is a day/site confound (other-channels-only AUC 0.746 vs 0.872). Zero-fraction shown **structurally impossible** — the device clamps every interval peak at >= 1 count. |
 | 2026-09-04 | Site-quiet-gated cross-channel differential established: **BE18438 Vert, BE9558 Tran+Long**, threshold-insensitive over a 2.3x span. Finds only **2 of the 5** confirmed units — leakage into the histogram floor is bimodal (0.9 to 0.02), so a negative result carries almost no information. Per-channel attribution **not** established (channel-scramble p = 0.769). |
 | 2026-09-04 | **BA10895 reclassified** from transient to a genuine Vert fault of a different subtype — 49.4% single-axis-dominant events, the highest in the fleet, 100% on Vert. The other six marginal units are clean. |
 | 2026-09-04 | **Defect found in `offset_scan3.py`**: its `spread <= 0.02` gate discards 18.8% of rows with \|pre\| >= 0.025, concentrated on 41 negative unit-channels; 4 would be sustained positives without it. The fleet label is three-state, not two. |
 | 2026-09-06 | **Four units relabelled BA, not BE** — BA9229, BA10060, BA10895, BA15957 are BlastMates. The BW filename carries only the serial number; the family prefix must be read from the file body. Fixed in the scanners and in two production sites. |
 | 2026-09-06 | **Mechanism campaign — five hypotheses, all dead.** Thermal, ground-motion shock, handling/redeployment, accumulated duty, unit age, firmware and a mechanical element fault are each refuted or bounded. 7 nominally significant results against 5.2 expected by chance. |
 | 2026-09-06 | **Onset is a RAMP of minutes-to-hours, not a step** — BE18438 Vert resolved to one-minute cadence, 50% of the excursion in 7 min, >=25 intermediates, ASCII-validated 75/75. Kills both a latched digital step AND slow component degradation. Surviving shape: a reversible two-time-constant settling process — a bias/leakage path charging a high-impedance node. |
 | 2026-09-06 | **Polarity consistency RETRACTED** (a tautology of the spread gate; amplitude-only gives 12 of 53 unit-channels mixed) and the fleet **re-cut to 8–9 units, not 5**. "Persistent until serviced" weakened: 23 recoveries, median gap 6 days — though a field autozero cannot be excluded. |
 | 2026-09-06 | The spread gate is **blind to onsets by construction** — it rejects a moving floor, which is what an onset is. It rejected the very record in which the ramp is visible. |
 | 2026-09-07 | **The non-motion test** (Brian: "it doesn't cross zero"). `\|mean\|/peak` is bimodal with a 1.09% dead zone and returns exactly the 5 confirmed units from physics, not a threshold. Independent corroboration of the unit list. **BE11007 settled as NOT an offset** — a one-sided 9 in/s blast. |
 | 2026-09-07 | **BE12599 is a connector fault, not a geophone fault.** Unipolar spikes with a 26→118 ms RC tail progressing to a flat pedestal, and the fault MOVES between Long and Tran while the sensor self-check passes on every event. An element cannot hop channels; a connector can. Inspect before reseating. |
@@ -0,0 +1,135 @@
 # USBM RI8507 / OSMRE Blasting Compliance Curve — Reference
 Reference for the **velocity-vs-frequency blasting compliance chart** Blastware
 draws on its Event Report ("USBM RI8507 And OSMRE"), and how seismo-relay
 reproduces it. Implemented in [`sfm/compliance.py`](../sfm/compliance.py); the
 spectral (FFT) side lives in [`waveform_fft.py`](../waveform_fft.py).
 Reverse-engineered 2026-09-14 against 7 BE12844 (MiniMate Plus) events, each
 with a Blastware Event Report + FFT Report as ground truth. Curve values from
 USBM RI8507 Appendix B and 30 CFR 816.67.
 ---
 ## What it is
 Two closely-related sources for the same limit curve:
 - **USBM RI8507** — Bureau of Mines *Report of Investigations 8507* (Siskind
  et al., 1980), *"Structure Response and Damage Produced by Ground Vibration
  From Surface Mine Blasting."* The curve is **Figure B-1**, Appendix B
  ("Alternative Blasting Level Criteria"), p.73–74.
 - **OSMRE / OSM** — the Office of Surface Mining Reclamation and Enforcement
  codified it as **30 CFR 816.67, Figure 1**. "CFR" = the U.S. Code of Federal
  Regulations. Same curve, regulatory force.
 The chart plots each geophone channel's significant vibration cycles as
 `(frequency, peak velocity)` points against this limit. A point **below** the
 line passes; **above** fails.
 ---
 ## The limit curve
 A structure has a resonance band (~4–12 Hz for whole structures) where it is
 most vulnerable, so the safe velocity is **lower** at those frequencies and
 **higher** away from them. The curve captures this by alternating two kinds of
 bound:
 - **Constant-velocity** segments — a flat horizontal line at a fixed PPV.
 - **Constant-displacement** segments — a fixed peak *displacement* `d`. For
  simple harmonic motion, peak velocity `v = 2πf·d`, so on a velocity-vs-
  frequency **log-log** plot this is a straight line of slope +1 (velocity rises
  with frequency). This is why the low- and high-frequency bounds are sloped.
 ### Two lines — structure type
 RI8507 gives two lines for two interior-wall constructions (Table 13, p.67):
 | line | construction | plateau PPV |
 |---|---|---|
 | **Drywall** (solid) | modern gypsum wallboard | **0.75 in/s** |
 | **Plaster** (dashed) | older plaster on wood lath | **0.50 in/s** |
 Plaster-on-lath is more damage-prone, hence the lower limit. You apply **one**
 line depending on the monitored structure.
 ### The four segments (Figure B-1, p.74)
 Going low → high frequency, each line is:
 1. **Ultimate low-frequency bound** — constant displacement **0.030 in**
   (`v = 2πf·0.030`). Only relevant below ~4 Hz.
 2. **Plateau** — constant velocity **0.75** (Drywall) / **0.50** (plaster) in/s.
 3. **Rising diagonal** — constant displacement **0.008 in** (`v = 2πf·0.008`),
   climbing from the plateau up to the high-frequency cap.
 4. **High-frequency cap** — constant velocity **2.0 in/s** above ~40 Hz.
 The segments are drawn **continuous**: each bound is used over the frequency
 range where it is the binding (lowest) limit, and consecutive bounds meet where
 they are equal — so there are no vertical steps. Transition frequencies come
 straight from the values (`f = V / (2π·d)`):
 | transition | formula | Drywall | Plaster |
 |---|---|---|---|
 | 0.030 in → plateau | `V_mid / (2π·0.030)` | 3.98 Hz | 2.65 Hz |
 | plateau → 0.008 in | `V_mid / (2π·0.008)` | 14.92 Hz | 9.95 Hz |
 | 0.008 in → 2.0 in/s | `2.0 / (2π·0.008)` | 39.79 Hz | 39.79 Hz |
 Because both lines share the same **0.008 in** rising diagonal, above ~15 Hz
 they lie on the *same* line (both reach 2.0 in/s at ~40 Hz) — RI8507's literal
 construction merges them there. Blastware renders the dashed line as a separate
 parallel diagonal, but that is cosmetic: above ~15 Hz both structure types carry
 the identical limit, so compliance is unaffected.
 > ⚠ RI8507's *Table 13* is a simpler two-range criterion with a **sharp
 > discontinuity at 40 Hz** (flat plateau, then a jump to 2.0). Figure B-1 is the
 > **smoothed** version that adds the 0.008 in transition — that is the one drawn
 > on reports and implemented here.
 ---
 ## The compliance scatter (the points)
 The cloud is **not** the FFT spectrum. It is a per-cycle, time-domain measure by
 the **zero-crossing method** (`channel_compliance_points`):
 - Split the channel's waveform at its zero crossings.
 - Each half-cycle contributes one point: **frequency** `= 1 / (2 · half-period)`
  (from the samples between the two crossings), **velocity** `= peak |amplitude|`
  in that half-cycle.
 This yields ~90–110 points per channel, and — by construction — each channel's
 **highest** point equals that channel's PPV. Verified against Blastware: the
 cloud shape, density, and ceiling all match.
 ### Why not the FFT?
 A broadband blast spreads its energy across many FFT bins, so no single bin
 reaches the time-domain peak — the FFT amplitudes come out ~10× below the
 compliance-chart velocities. The compliance chart is a *per-cycle peak* view;
 the **FFT** is a separate analysis (Blastware's *FFT Report*), reproduced by
 [`waveform_fft.py`](../waveform_fft.py) and used for the dominant-frequency
 readout and the #10 FFT view — not for this scatter.
 ---
 ## Implementation
 - `sfm/compliance.py`
  - `limit_at(freq, curve)` — the limit PPV at a frequency (`curve` = `"Drywall"`
    or `"Plaster"`); curves are data in `_CURVES`, so more standards can be added.
  - `channel_compliance_points(samples, sps)` — the zero-crossing scatter.
  - `draw_compliance_chart(ax, channels, sps)` — matplotlib rendering (both
    limit lines + per-channel scatter, Blastware's tick scales and channel
    markers: Tran `+` red, Vert `×` green, Long `o` blue).
 - Tests: `tests/test_compliance.py`.
 ---
 ## Sources
 - USBM **RI8507** (Siskind, Stagg, Kopp, Dowding, 1980), Appendix B / Figure B-1,
  p.73–74; Table 13, p.67. (`ref-stuff/usbm-ri8507-ground_vibration.pdf`.)
 - **30 CFR 816.67**, "Use of explosives: Control of adverse effects," Figure 1 —
  <https://www.ecfr.gov/current/title-30/chapter-VII/subchapter-K/part-816/section-816.67>
@@ -1,6 +1,7 @@
 # Runbook — Recovering a wedged unit stuck in a call-home loop
-**Original incident:** BE9558H at `166.246.130.1:9034`, recovered 2026-05-17.
+**Incidents:** BE9558H at `166.246.130.1:9034`, 2026-05-17 (Method B) ·
 BE12599 at `166.246.64.226:9034`, 2026-09-16 (Method A).
 A field unit with a stuck-triggered geophone (or any hardware fault causing
 constant event triggering) will record events back-to-back, and if Auto Call
@@ -14,6 +15,33 @@ This runbook describes how to break the loop and recover control.
 ---
 ## ⚠ Two cures for one disease — intercept first
 Both incidents below are the **same failure**: a geophone offset crosses the
 trigger level, the unit records back-to-back, ACH set to "after event
 recorded" dials continuously, and the unit becomes unreachable because its
 modem is in client mode almost all of the time.
 There are two ways to get a Stop Monitoring command into it.
 | | **A — intercept the call** (preferred) | **B — catch it between calls** (original) |
 |---|---|---|
 | Idea | Be the server it dials.  Point the modem's Destination at our own ACH server and answer it. | Clear the Destination so it stops dialing, then race a Stop into the gap. |
 | Needs inbound? | **No — the unit calls us** | Yes: working inbound TCP to the modem |
 | Determinism | Deterministic — it dials every ~75 s, we only have to be listening | A race.  BE9558H took ~7 h of attempts before one landed. |
 | Tool | `bridges/ach_server.py --stop-monitoring` | `scripts/slow_drip.sh` |
 | Proven on | BE12599, 2026-09-16 | BE9558H, 2026-05-17 |
 **Method A is the standard procedure now.**  The unit won't answer us because
 it is on the phone — so stop dialing it and be the one it calls.  It rings,
 we pick up, take its data, and tell it to stop calling here.
 Method B is kept because it is proven, and because A needs a listener the
 modem can actually reach (public IP + forwarded port).  When you have that,
 don't race it — intercept it.
 ---
 ## Symptoms
 - Terra-View / SFM `/device/info` either hangs or fails on `count_events()`.
@@ -31,9 +59,85 @@ If you see *all* of these, the unit is in this exact failure mode.
 ---
-## Quick reference — how to recover
+## Method A (preferred) — intercept the call
-You need **ACEmanager access** to the unit's modem.
+You need **ACEmanager access** and a host the modem can dial: public IP with
 the listener's port forwarded to it.
 ### A1 — start the listener BEFORE touching the modem
 ```bash
 cd /home/serversdown/seismo-relay
 tmux new -s rescue
 .venv/bin/python -u bridges/ach_server.py --port 12345 \
  -o bridges/captures/<unit>-diag --stop-monitoring -v
 ```
 ⚠ **Listener first, always.**  A Destination pointed at a dead port is the
 worst state available — the device still dials, the modem still flips to
 client mode, inbound stays blocked, and nothing gets delivered.
 Do **not** add `--events-only` (it silently breaks dedup — see gotchas), and
 do **not** add `--disable-ach` yet (see A4).
 ### A2 — point the modem at it
 ACEmanager → **Serial → Port Configuration**:
 | Field | Set to |
 |---|---|
 | **Destination Address** | the listener's public IP |
 | **Destination Port** | the listener's port (e.g. `12345`) |
 Apply.  The modem auto-dials its Destination whenever serial data arrives
 while the serial port is closed — so the unit's own retry cycle now lands on
 you instead of nowhere.
 ### A3 — answer, and stop the bleeding
 Within ~75 s you should see a call-in.  `--stop-monitoring` fires SUB 0x97 at
 step 1.5 — after the handshake, **before** the event walk — so the recording
 halts at the earliest possible moment in the session.  Confirm via
 `rescue.json` in the session directory:
 ```json
 {"peer": "166.246.64.226:60921", "stop_monitoring": "ok"}
 ```
 That is the bleeding stopped.  Everything after this is cleanup.
 ### A4 — drain the backlog, THEN disable ACH
 ⚠ **Order matters, and it is counter-intuitive.**  Stopping monitoring also
 removes your call-in trigger: ACH fires on "after event recorded", so with
 recording stopped the unit has no reason to dial again.  The backlog sitting
 in its memory does **not** re-arm it.
 So if the stored events are worth keeping — and on a fault unit they usually
 are, they're the evidence — drain them across however many call-ins it takes
 *before* you silence it.  Only then add `--disable-ach` (or use
 `scripts/rescue_device.sh <host> <port> --no-erase`).
 If the unit has gone quiet and you still need it, cycling the modem produces
 a call-in, and a unit with a scheduled daily call will dial at its configured
 time regardless.
 ### A5 — restore the Destination, and confirm you did
 Put `Destination Address` back to `0.0.0.0` (or the office Instantel ACH
 server) once you are finished, and only stop the listener after that is done.
 ### A6 — do NOT re-enable ACH until the hardware fault is repaired
 Otherwise the loop restarts the moment monitoring resumes and you run this
 runbook again.
 ---
 ## Method B (fallback) — catch it between calls
 The original 2026-05 procedure.  Use when you cannot stand up a listener the
 modem can reach.  You need **ACEmanager access** to the unit's modem.
 ### Step 1: stop the modem's mode-flipping
@@ -253,3 +357,223 @@ service).
 Total time from "i was wondering if its possible to" first attempt to
 recovery: ~7 hours of intermittent debugging across one evening.
 ---
 # Second incident — BE12599, 2026-09-16/17
 **Unit:** BE12599 at `166.246.64.226:9034`, RV50, job *I-80 North Fork Bridge
 — Abut 1 West* (Fay Company).  Same job as BE9558H, which is a coincidence.
 **Fault:** the connector fault documented in `docs/offset_investigation.md`
 §8e progressed until the Tran pedestal reached **0.400 in/s** — its trigger
 level.  Constant triggering → constant recording → ACH "after event recorded"
 → continuous dialing.  Same disease as BE9558H.
 **Same disease, inverted cure.**  Method B's Step 1 *did* work — clearing the
 Destination stopped the dial-outs, confirmed in the ALEOS log.  It was Step 2
 that didn't land, and rather than keep racing we turned the rescue around:
 gave the unit a different server to call, and answered it.
 Total time ≈ 5 h, of which ~90 min went to two red herrings documented below.
 Much of the rest was rediscovering the May procedure, which is why the
 "two cures" table now sits at the top of this file.
 ---
 ## Turn on ALEOS_SERIAL debug FIRST
 This is the single highest-value diagnostic and it should be step zero on any
 future incident.  ACEmanager → **Admin → Log → ALEOS_SERIAL log level →
 DEBUG**, then view the serial log.
 It is the only thing that tells you what the *device* is actually saying.
 Everything before we did this was guesswork.
 ## What the log showed — the unit is on the phone
 Every ~75 seconds, verbatim:
 ```
 ALEOS_SERIAL_HIF: 29 byte(s) in buffer: 'ATQ1^MATE0^MATS0=2^M^MRADIO RING^M'
 ALEOS_SERIAL_HMC: TCP recvhost fd 65535 len 29  state TCPMode::kClosed
 ALEOS_SERIAL_HMC: tcpmode trying to send to invalid socket
 ALEOS_SERIAL_HMC: Connect to IP: 0.0.0.0 Port 0
 ALEOS_SERIAL_HMC: Initialize Auto answer on port 9034
 ALEOS_SERIAL_HMC: Cannot connect to 0.0.0.0
 ```
 Read that carefully:
 - `ATQ1` (quiet) / `ATE0` (echo off) / `ATS0=2` (auto-answer after 2 rings).
  **There is no `ATD`.**  The device is not dialing — it is trying to
  *configure* its modem.
 - The modem's serial port is in TCP data mode, so it never interprets these
  as AT commands.  It treats them as payload and tries to ship them to a TCP
  socket that does not exist.
 - The device therefore never receives `OK`, never progresses, and **retries
  the identical 29 bytes forever**.
 **While it is in this state it is busy placing a call, not listening for
 us.**  This is almost certainly what BE9558H was doing too — we simply never
 turned on ALEOS_SERIAL debug in May to look.  It is not a different disease;
 it is the same one, seen properly for the first time.
 It is also the argument for Method A in one picture: the unit is mid-dial
 every ~75 s, and our inbound Stop has to thread the gaps between those
 attempts.  Give it somewhere to dial and the problem inverts into a
 deterministic one.
 ### Why `slow_drip` lied
 `slow_drip` returned the *success* signature except for the one field that
 mattered:
 ```json
 {"duration_s":120.0,"drips_sent":38,"bytes_sent":920,
 "bytes_received":0,"send_error":null}
 ```
 Full duration, no broken pipe — but zero bytes back.  Cause is in the log
 above: each 75 s cycle re-runs `Initialize Auto answer on port 9034`, which
 orphans the held session (`data in for unknown reason 3 removing from
 select`, `OnMsg recv error: 107 - Transport endpoint is not connected`).  Our
 local TCP stayed open so `sendall` never raised — but the modem stopped
 bridging after the first re-init, so every drip after that went into a socket
 nobody was reading.
 ⚠ **`send_error: null` + full duration is NOT success.  Only
 `bytes_received > 0` is success.**
 ⚠ **In fairness to slow_drip: it got exactly one attempt here**, run ~90 s
 after a modem reboot, with a dead session visible in the log at 20:19:17 in
 that same window.  BE9558H took hours of attempts before one landed.  Method B
 was not ruled out on BE12599 so much as abandoned in favour of something that
 doesn't need luck.
 ---
 ## ⚠ Two red herrings that cost ~90 minutes
 ### 1. The trusted-IP whitelist (this was the real reason inbound never worked)
 The RV50s run with **Security → Trusted IPs (Friends List) enabled**.  A
 source IP that is not on the list is dropped **silently** — inbound presents
 as `Connection error: timed out`, never a refusal.
 Brian's dev-box public IP is **dynamic** and had changed, so `tmi-dev` was no
 longer whitelisted.  Every inbound attempt failed identically across four
 different modem and device states, which looked exactly like the BE9558H
 mode-flipping symptom and sent us chasing modem configuration for over an
 hour.
 **Check this before diagnosing anything else.**  Note that SFM in Docker
 egresses via the *host's public IP*, not its LAN IP.
 ### 2. A 502 from SFM does not mean TCP connected
 `sfm/server.py` raises **502 for both** failure classes:
 ```python
 raise HTTPException(status_code=502, detail=f"Protocol error: {exc}")
 raise HTTPException(status_code=502, detail=f"Connection error: {exc}")
 ```
 We read an early 502 as "TCP connected, modem bridged, device mute" and built
 a whole theory on it.  It was almost certainly a connect timeout.
 **Always read the `detail` string** — "connect failed" and "device didn't
 answer" are completely different problems and the status code will not
 separate them.
 ---
 ## What actually worked — invert the direction
 The key observation is in the log above:
 > `TCP recvhost ... state TCPMode::kClosed` → `Connect to IP: 0.0.0.0 Port 0`
 **The modem auto-dials its Destination whenever serial data arrives while
 closed.**  So instead of fighting for inbound, give it somewhere to dial:
 point `Destination Address` at our own `ach_server` and the device's own
 75-second attempts become **device-initiated sessions the modem bridges
 correctly**.  No race, no contention, worst case a 75-second wait.
 ### Procedure
 1. **Run the rescue server** on a host the modem can reach (public IP +
   forwarded port):
   ```bash
   cd /home/serversdown/seismo-relay
   .venv/bin/python -u bridges/ach_server.py --port 12345 \
     -o bridges/captures/<unit>-diag --stop-monitoring -v
   ```
 2. **Point the modem at it** — ACEmanager → Serial → Port Configuration →
   `Destination Address` = your public IP, `Destination Port` = 12345.
 3. **Wait for the call-in.**  `--stop-monitoring` fires SUB 0x97 at step 1.5,
   after the handshake and *before* the event walk.  Confirm via
   `rescue.json` in the session directory:
   ```json
   {"peer": "166.246.64.226:60921", "stop_monitoring": "ok"}
   ```
 4. **Restore the modem's Destination** once you are done, then finish the
   device side (disable ACH, erase) through whichever channel works.
 On BE12599 the first call-in landed at 20:58:11 and reported
 `stop_monitoring: ok`; a second at 20:58:20 confirmed it.  `is_monitoring:
 false` was still true **6½ hours later** — the fix is durable.
 ---
 ## Hard-won gotchas (do not re-derive)
 - **Never leave the Destination pointed at a host with nothing listening.**
  That is the worst state available: the device still dials, the modem still
  flips, inbound stays blocked, and nothing is delivered.  An 8-minute gap
  with the listener down produced a spurious inbound timeout that cost
  another round of misdiagnosis.
 - **Stopping monitoring removes your call-in channel.**  ACH is "after event
  recorded"; no new events means no new dials.  The backlog sitting in memory
  does *not* re-arm it.  After a successful stop the unit goes quiet and you
  need the modem cycled (works — produced a call-in), the scheduled daily call
  (BE12599 calls at **05:00:14 device-local**, per §8e), or working inbound.
  **Plan the order before you fire the stop.**
 - **`--events-only` silently breaks dedup.**  It skips the device-info step,
  so the serial is never read; `ach_state.json` then keys on
  `peer:ephemeral_port`, which is unique per connection.  Every session looks
  like a new unit, starts from key 0, and re-downloads the same event.  Four
  sessions on BE12599 downloaded the identical event four times and made zero
  progress on the backlog.  Events also file as `serial=UNKNOWN` with a
  `M000…` BW filename (serial_numeric 0) instead of `N599…`.
  **Do not use `--events-only` when you intend to download anything.**
 - **`/device/events/index` reported `lifetime_count: 0`** on a unit with years
  of history.  Suspected decode bug in the SUB 0x08 field offset — do not
  trust that number.  The 88-byte payload is preserved in the `raw_hex` field
  if someone wants to chase it.
 - **Memory used cross-checks the event keys exactly:**
  `last_key − buffer_start = memory_total − memory_free`.  On BE12599:
  `0x011230ec − 0x01110000 = 78,060` and `983,028 − 904,968 = 78,060`.
  Useful sanity check that you are reading the keys right.
 ---
 ## Final state (2026-09-17 ~01:30 local)
 - `is_monitoring: false`, held 6½ hours
 - Battery 6.76 V
 - Memory 78,060 / 983,028 bytes used (8%)
 - `first_key 01121728`, `last_key 011230ec` — ~6.6 KB of addressable event
  chain, roughly 3 events
 - ACH still **enabled** — to be disabled after the backlog is preserved
 - Modem Destination still pointed at tmi-dev — to be restored
 - ⚠ **Do not re-enable ACH until the connector is serviced.**  Tran is still
  sitting at 0.400 and the loop restarts the moment monitoring resumes.
@@ -0,0 +1,134 @@
 # Plan — "Rescue Listener": a first-class tool for the inverted rescue
 **Status:** proposal, not started.  Written 2026-09-17 ~01:40 local, straight
 off the BE12599 incident.  Open questions at the bottom need Brian's answer
 before anything is built.
 **Background:** `docs/runbooks/wedged_unit_recovery.md`, "Second incident —
 BE12599".  The manual version of this worked; this plan is about making it a
 tool instead of a sequence of remembered steps at 1 AM.
 ---
 ## The problem, stated plainly
 When a unit is wedged in the BE12599 mode — geophone offset above trigger,
 recording back-to-back, ACH dialing constantly, device stuck repeating an AT
 modem-init string and therefore **deaf to S3 over inbound** — the only channel
 that works is the one the *device* opens.
 Recovering it currently means:
 1. Remember that `bridges/ach_server.py` exists and takes the right flags
 2. Start it by hand on a box the modem can reach, with a public port forwarded
 3. Go into ACEmanager and repoint the modem's Destination
 4. Watch a terminal for a call-in
 5. Read `rescue.json` to find out whether it worked
 6. Go back into ACEmanager and repoint the modem to where it belongs
 7. **Not forget step 6**, because leaving the Destination pointed at a dead
   listener is worse than never having started
 That is six manual steps and one landmine, executed under pressure while a
 unit floods the office server.
 ## What the tool should be
 **A "rescue listener" an operator can start for one unit, which handles
 whatever that unit says when it calls in, and refuses to go away until the
 operator confirms the modem has been pointed back.**
 Lifecycle:
 1. **Start** — operator names the target unit and starts a rescue listener.
   The tool reports the exact address/port to enter in ACEmanager, plus the
   actions it will take.
 2. **Operator repoints the modem** to that address.
 3. **Wait** — listener sits there.  Live status: "waiting for call-in",
   elapsed, last-seen.
 4. **Act** — on call-in, run the configured rescue actions automatically,
   in a safe order, each independently guarded.  Report per-action outcome.
 5. **Hold** — the listener **stays up** and keeps reporting, because the
   modem is still pointed at it.
 6. **Confirm & stop** — the operator explicitly confirms the Destination has
   been restored (to `0.0.0.0`, or to the office Instantel ACH server).
   Only then does the listener shut down.
 Step 6 is the whole point of making this a tool.  It is the step that is
 easiest to skip and most expensive to skip.
 ## Default action set
 Ordered deliberately — see "order matters" below.
 | # | Action | Default | Why |
 |---|---|---|---|
 | 1 | **Stop monitoring** (SUB 0x97) | ✅ on | Halts recording; ends the trigger→record→dial loop at its source.  Already implemented as `--stop-monitoring`. |
 | 2 | **Drain events** to a diagnostics store | ⚙ configurable | The backlog is usually evidence, not garbage — see the BE12599 offset investigation.  Must NOT land in the prod SFM DB. |
 | 3 | **Disable ACH** (SUB 0x2C/0x7E/0x7F) | ❌ off by default | Stops the dialing — **and stops your only channel**.  Opt-in, and ideally gated on step 1 having succeeded. |
 | 4 | **Erase events** | ❌ off by default | Destructive.  Only after a verified drain. |
 ### Order matters — the lesson from BE12599
 Stopping monitoring *removes the call-in trigger*.  ACH fires on "after event
 recorded"; with recording stopped, the unit has no reason to dial again, even
 though the backlog is still sitting in its memory.  So a naive
 "stop + disable + erase, all at once" rescue can silence the unit before
 you've collected anything, leaving you with no channel and a device full of
 evidence.
 The tool should either sequence around this or warn loudly about it.  My
 instinct is: **stop monitoring immediately** (it's the bleeding), then drain
 across however many call-ins it takes, and treat disable-ACH/erase as a
 separate, explicit "finish" action once the operator is satisfied.
 ## Where it should live — open question, with a proposal
 The natural tier is **SFM** (device-side, per the three-tier model in
 CLAUDE.md).  But the rescue listener must be reachable *from the cellular
 network*, which is a deployment constraint SFM's usual profile doesn't have.
 **Proposal worth considering:** run it at the office, beside the real Instantel
 ACH server, on a **different port** (e.g. 12346 while Instantel holds 12345).
 Then the ACEmanager change is a **port change, not an IP change** — smaller,
 faster, less to get wrong, and trivially reversible.  It also means the office
 public IP (already stable and known) is the destination, rather than whatever
 Brian's dynamic home IP happens to be that week.
 The tmi-dev approach used on BE12599 worked, but required a router forward and
 ran into the dynamic-IP problem in the same session.
 ## Open questions
 1. **Where does it run?**  Office beside Instantel ACH (port swap), SFM on the
   NAS, or ad-hoc on tmi-dev?  Affects everything else.
 2. **What drives it?**  Terra-View admin page (fits "operator UI"), an SFM
   endpoint pair (`POST /device/rescue_listener/start` + `/stop` + `/status`),
   or a CLI wrapper?  A long-lived listener doesn't fit the request/response
   endpoint shape well — probably needs a background task with a status poll.
 3. **How does it identify the unit?**  It can't know the serial until the
   device calls in and the handshake reads it.  Allowlist by modem IP?  Accept
   anything and report what showed up?
 4. **Where do drained events go?**  A per-incident diagnostics store
   (`bridges/captures/<unit>-diag`) seems right — explicitly *not* the prod
   SFM DB.  Does that store need to be a first-class thing with its own
   retention, or is a directory fine?
 5. **How is "confirm the modem is repointed" verified?**  Operator attestation
   (a button), or can we actually probe it?  If the listener stops seeing
   call-ins that's weak evidence; if inbound to the unit starts working that's
   stronger.
 6. **Multi-unit?**  One listener per incident, or one listener that handles any
   unit that dials in?  Probably the former for safety.
 7. **Timeout / abandonment policy.**  If nobody ever confirms, does it run
   forever?  Alert after N hours?
 ## What already exists
 - `bridges/ach_server.py` — the listener itself, with `--stop-monitoring`,
  `--disable-ach`, `--rescue` (added on `feat/ach-rescue-on-connect`, commit
  `9f1050b`), `--clear-after-download`, `--max-events`, `--allow-ip`.
 - Per-session `rescue.json` recording per-action outcomes.
 - Isolated per-output-dir SQLite + waveform store, so a diagnostics capture is
  already separate from prod by construction.
 So the gap is not protocol work — it's lifecycle, operator surface, and the
 confirmation gate.  Most of the risk is in questions 1 and 2.
@@ -47,19 +47,24 @@ from dataclasses import dataclass
 from pathlib import Path
 from typing import Optional, Union
-# Thor IDFW bodies are pinned to the SUPERSEDED tag-dispatch decoder.
+# Thor IDFW bodies use the series-3 record-chain decoder.
 #
-# _find_waveform_body_offset() trial-decodes every candidate offset and keeps
+# This was previously pinned to the SUPERSEDED tag-dispatch walker
-# whichever yields the most samples.  The series-3 record-chain decoder
+# (`decode_waveform_legacy`) on the stated grounds that "Thor has no ASCII
-# correctly returns None where the legacy walker returned garbage, which
+# ground truth in the corpus and its geo scaling is separately suspect".
-# changes that heuristic's winner on 33 of 577 files.  The net effect measured
+# Both premises were false: Thor writes a per-sample CSV export next to every
-# 2026-08-25 was positive (all-channels-equal 8/577 -> 506/577, mean abs PPV
+# binary (see scratch/verify_thor_against_csv.py), and the scaling is now
-# error 0.228 -> 0.173 in/s) but Thor has no ASCII ground truth in the corpus
+# resolved (see _GEO_LSB_IPS).  Measured against that ground truth on
-# and its geo scaling is separately suspect, so the switch is deferred until
+# 2026-09-10, the record chain beats the legacy walker outright:
-# the body-offset search is reworked to use the record chain directly.
+#
-from minimateplus.waveform_codec import (
+#     channel truncation   55/153 files -> 3/153
-    decode_waveform_legacy as decode_waveform_v2,
+#     files exact           98/153      -> 150/153
-)
+#     per-sample exact      99.781%     -> 99.854%
 #
 # The legacy walker stops at the first unrecognised tag and returns whatever
 # channels it had, so its failure mode is silent short channels rather than an
 # error.  Do not re-pin it.
 from minimateplus.waveform_codec import _MODES, decode_waveform_v2, is_record
 from .models import IdfEvent, IdfPeaks, IdfReport
@@ -89,23 +94,70 @@ _BODY_MAGIC = b"\x00\x02\x00"
 # fixed-header region where the same magic legitimately appears inside
 # channel-test records and the compliance block (offsets 0x015d, 0x091c,
 # 0x0ae2, 0x0d30 in observed events).
-_BODY_SCAN_FLOOR = 0x0E00
+# Lowered from 0x0E00 to 0x0C00 (2026-09-10).  Three-channel events -- mic
 # disabled -- have a shorter fixed header and put their record chain head at
 # 0x0dba, below the old floor.  The head was therefore invisible to the scan,
 # which fell through to the *Vert* segment-0 record and decoded a body shifted
 # one position around the channel rotation.  46 of 139 files in the
 # 9-10-26-csv-req corpus were affected; all 46 became per-sample exact once
 # the head was reachable.  The floor still skips the fixed-header region,
 # where `is_record()` can match channel-test records (0x015d, 0x091c, 0x0ae2).
 _BODY_SCAN_FLOOR = 0x0C00
-# Geophone count → in/s, derived from sidecar ground truth: the smallest
+# Cap on trial decodes per file.  Chain-head detection normally yields one
-# non-zero sample in 1,014-file corpus is 0.0003 in/s.
+# or two candidates; the cap only bounds the worst case on a corrupt file.
-_GEO_LSB_IPS = 0.0003
+_MAX_BODY_CANDIDATES = 16
 # Geophone count → in/s.
 #
 # The old value 0.0003 was read off the smallest non-zero sample in the
 # sidecar corpus, but that sample is Thor's *4-decimal display rounding* of
 # the true LSB, not the LSB itself.  It read every series-4 geophone sample
 # 3.3% low.  The quantisation ladder gives it away: counts 1..6 export as
 # 0.0003, 0.0006, 0.0009, 0.0012, 0.0016, 0.0019 — an LSB of exactly 0.0003
 # would end 0.0015, 0.0018.
 #
 # The value below maximises exact 4-dp agreement over 1,046,016 paired
 # samples (454 channel-events, 2 units) at 99.854%, versus 50.7% for 0.0003.
 # It is a global constant, not a per-unit calibration: all 8 UM units in the
 # production store independently agree to within ±0.07% on their
 # device-reported PPV.  1/LSB = 3222.6 counts per in/s.
 #
 # The value is pinned, not guessed.  Each exported sample constrains the LSB
 # to the window that rounds to the printed 4-dp figure; intersecting 991,415
 # such constraints (clean channel-events only) gives
 #
 #     LSB in [0.000310307933, 0.000310308057]     width 1.2e-10
 #
 # 0.000310308 sits at the centre of that window.  Equivalent full scale is
 # 10.0 in/s / 0.000310308 = 32226.05 counts.
 #
 # Corroboration from the device: an IDFH interval that never recorded keeps
 # its min/max accumulator at its ±full-scale seed, and that seed is
 # (min=+32226, max=-32226) — the same magnitude, independently.  Note the
 # tempting closed form 10.0/32226 is very slightly WRONG: it lands 4.5e-10
 # above the feasible window and loses 78 boundary samples to the literal
 # value while never winning one.  Series-3 uses 32000 counts for the same
 # 10.0 in/s, so the two generations do NOT share a scale.
 #
 # Ground truth + harness: scratch/verify_thor_against_csv.py
 _GEO_LSB_IPS = 0.000310308
 # Microphone count → psi, derived from sidecar regression on 50 sample
 # pairs from UM11719_20231219162723.IDFW (mic-heavy event).
 _MIC_LSB_PSI = 2.14e-6
 # IDFH histogram constants.
-_IDFH_INTERVAL_SIZE = 72        # bytes per per-interval record
+# Bytes per interval record = 16 per channel + an 8-byte tail, so a
 # 4-channel unit uses 72 and a mic-disabled 3-channel unit uses 56.  It is
 # NOT a constant: derive it per segment from the interval counter (see
 # decode_idfh_body).  This value survives only as the 4-channel default.
 _IDFH_INTERVAL_SIZE = 72        # bytes per per-interval record (4 channels)
 _IDFH_CHANNEL_BLOCK = 16        # bytes per channel inside an interval record
 _IDFH_INTERVAL_TAIL = 8         # bytes after the per-channel blocks
 _IDFH_SEGMENT_HEADER = 10       # bytes: [len_be 2B][0a 00 00 00 4B][00 NN 2B][05 3f 2B]
 _IDFH_SEGMENT_TAIL   = 2        # bytes after the interval data block, before next marker
 _IDFH_HALFP_FREQ_NUM = 512.0    # freq_hz = NUM / halfp; halfp ≤ 5 means ">100 Hz" sentinel
 _IDFH_GEO_FULL_SCALE = 10.0     # in/s — Normal range
 _IDFH_INT16_FS = 32768.0
 _IDFH_CHANNELS = ("Tran", "Vert", "Long", "MicL")
@@ -223,26 +275,67 @@ def _find_waveform_body_offset(buf: bytes) -> Optional[int]:
    """
    if len(buf) < _BODY_SCAN_FLOOR + 8:
        return None
-    best: Optional[tuple[int, int]] = None   # (total_samples, offset)
+
-    i = _BODY_SCAN_FLOOR
+    # 1. Locate every plausible per-channel record header.  A header carries
    #    [len 2B][channel_id][00][00] at +2..+6, so anchor the search on the
    #    three-byte ``<cid> 00 00`` signature and validate with is_record().
    #    Scanning candidate *preambles* instead is not viable: MODE_RAW16 is
    #    ``00 00``, so every run of three zero bytes would look like a body
    #    start and each would cost a full trial decode (~0.5 s/file measured).
    floor = max(0, _BODY_SCAN_FLOOR - 7)
    starts: list = []
    for cid in (0x46, 0x47, 0x48, 0x49):
        sig = bytes((cid, 0x00, 0x00))
        i = floor
        while True:
-        j = buf.find(_BODY_MAGIC, i)
+            j = buf.find(sig, i)
            if j < 0:
                break
            i = j + 1
            q = j - 4
            if q >= floor and is_record(buf, q):
                starts.append(q)
    if not starts:
        return None
    starts.sort()
    # 2. A body begins at the head of a record chain -- a record that no other
    #    record's length field points at.  The head's own payload is the
    #    implicit segment-0 Tran record, and the body offset is head + 7 (past
    #    [len 2B][cid][00][00][seg]) so that body[1:3] lands on the mode.
    ends = {q + 2 + int.from_bytes(buf[q + 2 : q + 4], "big") for q in starts}
    heads = [q for q in starts if q not in ends] or starts[:1]
    # 3. Trial-decode each head and keep the best.  Prefer a candidate where
    #    all four channels come out the same length: scoring on raw sample
    #    count alone picks false positives sitting *inside* a record header,
    #    which decode a plausible-looking but rotation-shifted body that
    #    silently drops each channel's segment 0.
    best = None
    best_off = None
    for head in heads[:_MAX_BODY_CANDIDATES]:
        j = head + 7
        if j + 3 > len(buf) or (buf[j + 1], buf[j + 2]) not in _MODES:
            continue
        try:
            decoded = decode_waveform_v2(buf[j:])
        except Exception:
            continue
        if not decoded:
            continue
        lengths = [len(v) for v in decoded.values() if v]
        total = sum(len(v) for v in decoded.values())
        # A "real" body has more than just the 2-sample preamble.
        if total <= 2:
            continue
-        if best is None or total > best[0]:
+        # >= 3 rather than == 4: a mic-disabled event has only the three geo
-            best = (total, j)
+        # channels, and demanding four made `equal` permanently False for
-    return best[1] if best else None
+        # them, leaving the pick to raw sample count alone.
        equal = len(lengths) >= 3 and len(set(lengths)) == 1
        score = (equal, total)
        if best is None or score > best:
            best, best_off = score, j
    return best_off
 def _decode_waveform_samples(buf: bytes) -> Optional[dict]:
@@ -299,6 +392,12 @@ class IdfhInterval:
    micl_min:    int
    micl_max:    int
    micl_halfp:  int
    # 4 on a normal unit; 3 when the microphone is disabled, in which case the
    # micl_* fields are absent from the record and read as zero.
    n_channels:  int = 4
    def has_channel(self, channel: str) -> bool:
        return channel != "MicL" or self.n_channels >= 4
    def peak_count(self, channel: str) -> int:
        mn = getattr(self, f"{channel.lower()}_min")
@@ -307,7 +406,11 @@ class IdfhInterval:
    def peak_ips(self, channel: str) -> float:
        """Convert peak count to in/s (geo channels only)."""
-        return self.peak_count(channel) / _IDFH_INT16_FS * _IDFH_GEO_FULL_SCALE
+        # Same geo LSB as the waveform path — verified independently against
        # the IDFH exports: as peak magnitude rises (and 4-dp quantisation
        # noise falls) the implied LSB converges on 0.0003103, matching
        # _GEO_LSB_IPS.  The old 10.0/32768 read histogram peaks 1.7% low.
        return self.peak_count(channel) * _GEO_LSB_IPS
    def freq_hz(self, channel: str) -> Optional[float]:
        halfp = getattr(self, f"{channel.lower()}_halfp")
@@ -316,11 +419,46 @@ class IdfhInterval:
        return _IDFH_HALFP_FREQ_NUM / halfp
-def _decode_idfh_interval(buf72: bytes, offset: int) -> IdfhInterval:
+def _is_unwritten_interval(interval: "IdfhInterval") -> bool:
-    """Decode one 72-byte interval record into per-channel min/max/halfp."""
+    """True for an interval slot the device reserved but never wrote.
    Thor seeds each interval's per-channel accumulators at ``min = +full
    scale`` and ``max = -full scale`` and then narrows them as samples
    arrive.  A slot that never recorded keeps that seed, so ``min > max`` —
    impossible for real data.  Such a record decodes to a full-scale
    10.0 in/s peak on every channel and, being a max-over-intervals, poisons
    the whole file's PPV.
    Rare but real: exactly 1 of 497,611 corpus intervals, and it inflated
    that file's Long PPV from 0.0081 to 10.0 in/s.  The inversion is always
    all-or-nothing across channels (0 partial cases in the corpus), so
    requiring every channel to be inverted keeps this from ever firing on
    genuine data.
    """
    pairs = [
        (interval.tran_min, interval.tran_max),
        (interval.vert_min, interval.vert_max),
        (interval.long_min, interval.long_max),
    ]
    if interval.has_channel("MicL"):
        pairs.append((interval.micl_min, interval.micl_max))
    return all(mn > mx for mn, mx in pairs)
 def _decode_idfh_interval(buf72: bytes, offset: int,
                          n_channels: int = 4) -> IdfhInterval:
    """Decode one interval record into per-channel min/max/halfp.
    The record is ``n_channels`` × 16-byte blocks plus an 8-byte tail, so it
    is 72 bytes on a normal unit and 56 when the microphone is disabled.
    Missing channels read as zero.
    """
    import struct
    fields = []
    for i in range(4):
        if i >= n_channels:
            fields.extend([0, 0, 0])
            continue
        block = buf72[i * 16 : (i + 1) * 16]
        mn = struct.unpack_from(">h", block, 0)[0]
        mx = struct.unpack_from(">h", block, 2)[0]
@@ -336,6 +474,7 @@ def _decode_idfh_interval(buf72: bytes, offset: int) -> IdfhInterval:
        vert_min=fields[3], vert_max=fields[4], vert_halfp=fields[5],
        long_min=fields[6], long_max=fields[7], long_halfp=fields[8],
        micl_min=fields[9], micl_max=fields[10], micl_halfp=fields[11],
        n_channels=n_channels,
    )
@@ -343,36 +482,73 @@ def decode_idfh_body(buf: bytes) -> list:
    """Walk an IDFH file and decode every interval record.
    The body has one or more segments; each segment header is 12 bytes:
-    ``[length_be 2B][0a 00 00 00][00 NN_counter][05 3f]`` where ``length``
+    ``[length_be 2B][0a 00 00 00][counter_be 2B][05 3f]`` where ``length``
    is bytes from the magic through the end of the interval block
    (= 10 + 72 × n_intervals).  Segments are separated by a 2-byte tail
    + next-segment 2-byte prefix (the bytes before the next length field).
-    Confirmed against the 859-file corpus (181,071 intervals decoded; 1
+
-    failure is the sig-B BE9439 file).
+    ``counter`` is a **uint16 BE cumulative interval index** — the 0-based
    index of the LAST interval in this segment.  Segments carry 10
    intervals each, so it runs 9, 19, 29, ... across the file.
    ⚠ This validator used to require ``buf[j + 4] == 0x00``, i.e. that the
    counter's high byte was zero.  That silently capped every histogram at
    **250 intervals**: the moment the cumulative counter passed 255 the high
    byte went non-zero and every later segment was rejected, so any
    monitoring run longer than ~4 hours lost its tail — frequently the part
    holding the event peak, which is why those files' PPV read low.  540 of
    858 corpus files were affected.  Do not reinstate that check.
    """
    intervals: list = []
    i = 0
    prev_counter = -1          # so the first segment's n = counter + 1
    while True:
        j = buf.find(b"\x0a\x00\x00\x00", i)
        if j < 0 or j < 2:
            break
-        # Validate: [length_be][0a 00 00 00][00 NN][05 3f]
+        # Validate: [length_be][0a 00 00 00][counter_be][05 3f].  The counter
-        if buf[j + 4] != 0x00 or buf[j + 6 : j + 8] != b"\x05\x3f":
+        # is deliberately NOT constrained — see the note above.
        if buf[j + 6 : j + 8] != b"\x05\x3f":
            i = j + 1
            continue
        length = int.from_bytes(buf[j - 2 : j], "big")
-        n = (length - _IDFH_SEGMENT_HEADER) // _IDFH_INTERVAL_SIZE
+        counter = int.from_bytes(buf[j + 4 : j + 6], "big")
        header_start = j - 2
        if length < _IDFH_SEGMENT_HEADER or header_start + length > len(buf):
            # Truncated / bogus length — not a real segment header.
            i = j + 1
            continue
        # The counter is the cumulative index of this segment's LAST interval,
        # so the interval count is its delta from the previous segment.  That
        # gives the record stride, which is NOT fixed: 16 bytes per channel
        # plus an 8-byte tail, so 72 for a 4-channel unit and 56 for a
        # mic-disabled 3-channel one.  Assuming 72 unconditionally made every
        # 3-channel histogram read 7 intervals per 10-interval segment,
        # walking off alignment into garbage that decoded as ~10 in/s peaks.
        n = counter - prev_counter
        if n <= 0:
            i = j + 1
            continue
-        header_start = j - 2
+        stride = (length - _IDFH_SEGMENT_HEADER) // n
        n_channels, remainder = divmod(stride - _IDFH_INTERVAL_TAIL,
                                       _IDFH_CHANNEL_BLOCK)
        if remainder or not (1 <= n_channels <= 4):
            i = j + 1
            continue
        interval_start = header_start + _IDFH_SEGMENT_HEADER
        for k in range(n):
-            off = interval_start + k * _IDFH_INTERVAL_SIZE
+            off = interval_start + k * stride
-            if off + _IDFH_INTERVAL_SIZE > len(buf):
+            if off + stride > len(buf):
                break
-            chunk = buf[off : off + _IDFH_INTERVAL_SIZE]
+            chunk = buf[off : off + stride]
-            intervals.append(_decode_idfh_interval(chunk, off))
+            interval = _decode_idfh_interval(chunk, off, n_channels)
            if _is_unwritten_interval(interval):
                # Reserved-but-never-recorded slot: the min/max accumulators
                # still hold their ±full-scale seed.  Counting it would
                # fabricate a 10.0 in/s peak on every channel.
                continue
            intervals.append(interval)
        prev_counter = counter
        # Advance past this segment + the 2-byte tail.
        i = header_start + length + _IDFH_SEGMENT_TAIL
    return intervals
@@ -452,7 +628,12 @@ def read_idf_file(
        peak_long = max((iv.peak_ips("Long") for iv in intervals), default=0.0)
        # Mic peak in psi — Thor stores per-interval mic ADC counts in the
        # binary; convert the max count to psi via the per-count factor.
-        mic_peak_count = max((iv.peak_count("MicL") for iv in intervals), default=0)
+        # Skip on a mic-disabled (3-channel) unit: those records carry no mic
        # block at all, so peak_count("MicL") would report a synthetic zero.
        mic_peak_count = max(
            (iv.peak_count("MicL") for iv in intervals if iv.has_channel("MicL")),
            default=0,
        )
        mic_peak_psi = mic_count_to_psi(mic_peak_count) if mic_peak_count else None
        rep = IdfReport(
            serial_number=md.serial,
@@ -0,0 +1,75 @@
 """Structural annotation of a Series-3 Blastware waveform binary.
 Pure, no I/O: takes the raw file bytes and returns a flat, gap-free tiling of
 labelled :class:`Span` regions for a hex viewer to paint.  Every byte is
 covered — anything the decoder can't account for becomes an ``unknown`` span,
 so undecoded regions (e.g. a stored spectral/FFT block, if one exists) stand
 out instead of hiding.
 File layout (see ``blastware_file.py``):  ``[header][21B STRT][body][26B footer]``.
 The body is the record chain walked by :func:`waveform_codec.walk_records`.
 """
 from __future__ import annotations
 from dataclasses import dataclass
 from typing import List
 from .waveform_codec import walk_records
 _STRT_LEN = 21
 _FOOTER_LEN = 26
@dataclass
 class Span:
    start: int          # inclusive byte offset
    end: int            # exclusive byte offset
    label: str          # human-readable description
    kind: str           # 'header' | 'strt' | 'sample' | 'footer' | 'unknown'
 def _tile(known: List[Span], total: int) -> List[Span]:
    """Sort *known* spans and fill every gap with an ``unknown`` span, so the
    result is a contiguous, non-overlapping tiling of ``[0, total)``.  Overlaps
    are resolved by clamping to the running position (first writer wins)."""
    out: List[Span] = []
    pos = 0
    for s in sorted(known, key=lambda x: (x.start, x.end)):
        if s.end <= pos:
            continue                       # fully behind — dropped overlap
        start = max(s.start, pos)
        if start > pos:
            out.append(Span(pos, start, "unknown", "unknown"))
        out.append(s if start == s.start else Span(start, s.end, s.label, s.kind))
        pos = s.end
    if pos < total:
        out.append(Span(pos, total, "unknown", "unknown"))
    return out
 def annotate_blastware_binary(raw: bytes) -> List[Span]:
    """Annotate a Series-3 waveform binary into a gap-free list of spans."""
    total = len(raw)
    strt_pos = raw.find(b"STRT")
    if strt_pos < 0:
        return [Span(0, total, "unrecognized — no STRT record", "unknown")]
    known: List[Span] = []
    if strt_pos > 0:
        known.append(Span(0, strt_pos, "File header", "header"))
    known.append(Span(strt_pos, strt_pos + _STRT_LEN, "STRT record", "strt"))
    body_start = strt_pos + _STRT_LEN
    footer_start = total - _FOOTER_LEN
    if footer_start >= body_start:
        known.append(Span(footer_start, total, "File footer", "footer"))
    else:
        footer_start = total          # file too short for a footer
    body = raw[body_start:footer_start]
    for rec in walk_records(body):
        hi, lo = rec["mode"]
        label = f"{rec['channel']} record (seg {rec['segment_index']}, mode {hi:02x} {lo:02x})"
        known.append(Span(body_start + rec["offset"], body_start + rec["end"], label, "sample"))
    return _tile(known, total)
@@ -30,6 +30,7 @@ from __future__ import annotations
 import datetime
 import logging
 import re
 import struct
 from typing import Optional
@@ -2532,10 +2533,17 @@ def _decode_0a_partial_header(raw_data: bytes, index: int, key4: bytes) -> Optio
        ts2 = try_ts(raw_data[ts1_end + 1:ts1_end + 1 + ts_size])
    # Extract serial and geo threshold from "BE11529\0" and "Geo: X.XXX in/s\0".
    #
    # Match any two-letter family prefix, not a literal "BE" — a BlastMate
    # reports "BA10895", and the old `find(b"BE")` returned -1 on one.  That
    # skipped this whole block, so the geo threshold went missing along with
    # the serial.  Requiring the NUL terminator in the pattern also makes the
    # match stricter than the bare two-byte search it replaces.
    serial: Optional[str] = None
    geo_ips: Optional[float] = None
-    serial_pos = raw_data.find(b"BE")
+    serial_match = re.search(rb"[A-Z]{2}\d{3,6}(?=\x00)", raw_data)
    serial_pos = serial_match.start() if serial_match else -1
    if serial_pos >= 0:
        # Read null-terminated serial starting at serial_pos.
        null_pos = raw_data.find(b"\x00", serial_pos)
@@ -50,7 +50,7 @@ SIDECAR_KIND   = "sfm.event"
 # bumped without a `pip install` re-run — leading to confusing stale
 # version stamps in sidecars.  Bump this constant and CHANGELOG.md
 # together at release time.
-TOOL_VERSION = "0.26.0"
+TOOL_VERSION = "0.30.0"
 try:
    # Best-effort: prefer the installed metadata when it's NEWER than the
@@ -413,10 +413,18 @@ def detect_multi_interval_stride(body: bytes) -> Optional[int]:
        if (_ctr(stride) - _ctr(0)) & 0xFFFF != 1:
            continue
-        # confirm on a third block when the body is long enough
+        # Confirm on a third block WHEN ONE IS ACTUALLY PRESENT.  A body can
-        if 2 * stride + _MULTI_HEADER_LEN <= len(body):
+        # be longer than two strides and still hold only two real blocks: a
-            if not _is_multi_header(body, 2 * stride):
+        # final *partial* block leaves trailing padding.  E.g. 51 intervals at
-                continue
+        # 2 s = one full 30-interval block + a 21-interval remainder, in a
        # 2787-byte body — long enough to demand a third header at 1224 that
        # does not exist.  Requiring it unconditionally threw away the correct
        # stride and the file decoded to nothing (BE18193 T193L0XM.CI0H).
        # The block-counter check above is the decisive anti-false-positive
        # test; this one is corroboration, so a missing third header means
        # end-of-stream, not disqualification.
        if (2 * stride + _MULTI_HEADER_LEN <= len(body)
                and _is_multi_header(body, 2 * stride)):
            if (_ctr(2 * stride) - _ctr(stride)) & 0xFFFF != 1:
                continue
        return stride
@@ -0,0 +1,146 @@
 r"""Decode the Blastware sensor self-check waveforms from a series-3 event binary.
 Reverse-engineered 2026-09-15 against 7 BE12844 (MiniMate Plus) oracle events.
 After the main waveform record-chain and the trailing metadata / per-channel
 calibration records, the binary carries four length-prefixed records tagged
 0x3c-0x3f: the sensor self-check traces the unit records when it pulses each
 sensor before monitoring.  Blastware draws these as the little waveforms in the
 "Sensor Check" strip on the right of the Event Report.
  * 0x3c / 0x3d / 0x3e = Tran / Vert / Long geophone ring-downs (a damped
    oscillation at the geophone's resonance, ~7-8 Hz at 1024 sps).
  * 0x3f            = MicL, a pulse train at the mic self-test frequency
    (~20 Hz), whose zero-crossing frequency is BW's mic "Channel Test" freq.
 Record framing (per record, all four chained by their length prefix)::
    [len:2 BE][id:1][00 00][Nchan:1][12-byte header][delta stream][40 02][6B]
    \_________________ payload (len bytes) _______________________________/
 The delta stream is ``payload[20 : len-8]`` (the ``40 02`` terminator sits at
 ``len-8``, followed by 6 trailing bytes).  It uses the exact same 10/20/30/00
 delta-block tags as the main waveform codec
 (:mod:`minimateplus.waveform_codec`), decoded here from an implicit anchor of 0
 — so the traces come out in the same 16-count raw units as the main waveform
 (LSB = 0.005 in/s at Normal range for the geophones).
 """
 from __future__ import annotations
 from typing import Dict, List
 from minimateplus.waveform_codec import walk_body
 # Record id → channel.  Order mirrors the trailing per-channel calibration
 # records (Tran / Vert / Long / MicL), confirmed against BW's sensor-check
 # frequencies on all 7 oracle events.
 _ID_TO_CHANNEL = {0x3C: "Tran", 0x3D: "Vert", 0x3E: "Long", 0x3F: "MicL"}
 _CHAIN_IDS = (0x3C, 0x3D, 0x3E, 0x3F)
 _HEADER_LEN = 20    # payload bytes before the delta stream
 _TRAILER_LEN = 8    # 40 02 terminator + 6 trailing bytes after the stream
 def _s4(nib: int) -> int:
    """Sign-extend a 4-bit nibble delta."""
    return nib - 16 if nib >= 8 else nib
 def _i8(byte: int) -> int:
    """Sign-extend an 8-bit int delta."""
    return byte - 256 if byte >= 128 else byte
 def _decode_delta_stream(buf: bytes) -> List[int]:
    """Accumulate a 10/20/30/00 delta-block stream from an anchor of 0,
    stopping at the 0x40 terminator.
    Mirrors the block semantics in
    :func:`minimateplus.waveform_codec.decode_waveform_v2` (fully decoded &
    byte-exact as of 2026-05-11); see that module for the format details.
    """
    out: List[int] = []
    cur = 0
    for blk in walk_body(buf, 0):
        fam = blk.tag_hi & 0xF0
        if fam == 0x10:
            # nibble deltas, high nibble first
            for byte in blk.data:
                for nib in ((byte >> 4) & 0xF, byte & 0xF):
                    cur += _s4(nib)
                    out.append(cur)
        elif fam == 0x20:
            # int8 deltas
            for byte in blk.data:
                cur += _i8(byte)
                out.append(cur)
        elif fam == 0x30:
            # 12-bit signed deltas, packed as tag_lo/4 groups of 6 bytes
            for g in range(blk.tag_lo // 4):
                grp = blk.data[g * 6:(g + 1) * 6]
                if len(grp) < 6:
                    break
                high_word = (grp[0] << 8) | grp[1]
                for k in range(4):
                    nib = (high_word >> (12 - 4 * k)) & 0xF
                    v = (nib << 8) | grp[2 + k]
                    if v >= 0x800:
                        v -= 0x1000
                    cur += v
                    out.append(cur)
        elif fam == 0x00:
            # RLE zero-delta run (wide form carries the high nibble in the tag)
            run = ((blk.tag_hi & 0x0F) << 8) | blk.tag_lo
            out.extend([cur] * run)
        elif fam == 0x40:
            # segment / record terminator
            break
    return out
 def _find_chain(body: bytes):
    """Locate the four length-prefixed sensor-check records.
    Returns a list of ``(offset, id, length)`` or ``None``.  The chain is
    validated by walking the ids 0x3c → 0x3d → 0x3e → 0x3f via their own length
    prefixes, so a stray 0x3c byte in the waveform data cannot match.
    """
    for p in range(len(body) - 6):
        if body[p + 2] == 0x3C and body[p + 3] == 0 and body[p + 4] == 0:
            q = p
            recs = []
            ok = True
            for expect in _CHAIN_IDS:
                if q + 3 > len(body) or body[q + 2] != expect:
                    ok = False
                    break
                length = int.from_bytes(body[q:q + 2], "big")
                recs.append((q, expect, length))
                q = q + 2 + length
            if ok and len(recs) == 4:
                return recs
    return None
 def decode_sensor_check(raw: bytes) -> Dict[str, List[int]]:
    """Decode the four sensor self-check traces from a series-3 event binary.
    Returns ``{"Tran": [...], "Vert": [...], "Long": [...], "MicL": [...]}`` in
    raw decode units (same 16-count LSB as the main waveform), or ``{}`` if the
    binary carries no sensor-check block (a histogram event, a non-series-3
    file, or a unit/firmware that doesn't store it).
    """
    strt = raw.find(b"STRT")
    if strt < 0 or len(raw) < strt + 21 + 26:
        return {}
    body = raw[strt + 21: len(raw) - 26]
    chain = _find_chain(body)
    if not chain:
        return {}
    out: Dict[str, List[int]] = {}
    for off, rid, length in chain:
        payload = body[off + 2: off + 2 + length]
        if len(payload) < _HEADER_LEN + _TRAILER_LEN:
            continue
        stream = payload[_HEADER_LEN: length - _TRAILER_LEN]
        out[_ID_TO_CHANNEL[rid]] = _decode_delta_stream(stream)
    return out
@@ -722,7 +722,18 @@ STREAM_END_ID = 0x06
 MODE_DELTA    = (0x02, 0x00)
 MODE_ABSOLUTE = (0x01, 0x00)
 MODE_RAW12    = (0x00, 0x03)
-_MODES = (MODE_DELTA, MODE_ABSOLUTE, MODE_RAW12)
+# Raw int16 BE absolute samples, 10-byte header, no tags — the same shape as
 # MODE_RAW12 but two bytes per sample instead of 1.5.  Found on Thor/Micromate
 # segment-0 records (2026-09-10): a `len=1032` record carries exactly
 # (1032 - 8) / 2 = 512 samples and reproduces Thor's own export 512/512
 # exactly.  Before this mode existed the record fell through the dispatch
 # unhandled, so the channel silently lost its first 512 samples.
 MODE_RAW16    = (0x00, 0x00)
 _MODES = (MODE_DELTA, MODE_ABSOLUTE, MODE_RAW12, MODE_RAW16)
 # Preambles whose leading data is untagged and therefore cannot be
 # block-walked; find_first_record() must scan for the next record instead.
 _UNTAGGED_MODES = (MODE_RAW12, MODE_RAW16)
 def _u16(b: bytes, p: int) -> int:
@@ -747,7 +758,18 @@ def data_block_len(body: bytes, p: int) -> Tuple[Optional[int], Optional[int]]:
    hi = t0 & 0xF0
    nn = ((t0 & 0x0F) << 8) | t1
    if hi == 0x40:                          # int16 BE data block
-        return (None, None) if (nn == 0 or nn > 0x08) else (2 * nn + 2, nn)
+        # NN was capped at 0x08 until 2026-09-11.  That cap had no basis: the
        # two corpora available at the time only ever used NN in {1,2,3,4,8},
        # so it was never exercised.  Loud UM12947 events use NN of 12, 16,
        # 20 ... up to 196, and every value above 8 halted the walk, which
        # surfaced as silently short channels (walk_body/run stop at the first
        # unrecognised tag rather than raising).  Verified against Thor's own
        # exports: 22 length-mismatched files -> 0, and the affected corpus
        # went to 1,476,242/1,476,249 samples exact.  The real bound is the
        # buffer; the caller additionally clamps to the record end.
        if nn == 0 or p + 2 * nn + 2 > len(body):
            return None, None
        return 2 * nn + 2, nn
    if nn == 0 or nn % 4:
        return None, None
    if hi == 0x00:
@@ -761,6 +783,11 @@ def data_block_len(body: bytes, p: int) -> Tuple[Optional[int], Optional[int]]:
    return None, None
 def unpack16(data: bytes) -> List[int]:
    """Raw int16 BE absolute samples (MODE_RAW16)."""
    return [_i16(data, 2 * k) for k in range(len(data) // 2)]
 def unpack12(data: bytes) -> List[int]:
    """Raw 12-bit packed samples: 6 bytes -> 4 signed values."""
    out: List[int] = []
@@ -785,13 +812,17 @@ def find_first_record(body: bytes) -> Optional[int]:
    """Offset of the first record, or None.
    Under the normal ``00 02 00`` preamble the leading bytes are segment-0's
-    Tran blocks, so walk them.  Under the ``00 00 03`` preamble that data is
+    Tran blocks, so walk them.  Under the untagged preambles (``00 00 03``
-    raw 12-bit with no tags at all and cannot be block-walked — scan instead.
+    raw-12 and ``00 00 00`` raw-16) that data has no tags at all and cannot
    be block-walked — scan for the next record header instead.
    """
-    if len(body) >= 3 and (body[1], body[2]) == MODE_RAW12:
+    if len(body) >= 3 and (body[1], body[2]) in _UNTAGGED_MODES:
        scan_from = 3
    else:
-        i = 7
+        # Tagged preamble.  MODE_DELTA carries a 14-byte record header (two
        # int16 anchors), so its blocks start at body[7]; MODE_ABSOLUTE has a
        # 10-byte header and starts at body[3].
        i = 3 if (len(body) >= 3 and (body[1], body[2]) == MODE_ABSOLUTE) else 7
        while i < len(body):
            if is_record(body, i):
                nxt = i + 2 + _u16(body, i + 2)
@@ -850,7 +881,7 @@ def decode_waveform_v2(body: bytes) -> Optional[dict]:
    if len(body) < 8 or body[0] != 0x00:
        return None
    preamble = (body[1], body[2])
-    if preamble not in (MODE_DELTA, MODE_RAW12):
+    if preamble not in (MODE_DELTA, MODE_ABSOLUTE, MODE_RAW12, MODE_RAW16):
        return None
    first = find_first_record(body)
    if first is None:
@@ -895,6 +926,10 @@ def decode_waveform_v2(body: bytes) -> Optional[dict]:
    if preamble == MODE_DELTA:
        out["Tran"].extend([_i16(body, 3), _i16(body, 5)])
        run("Tran", 7, first, absolute=False)
    elif preamble == MODE_ABSOLUTE:
        run("Tran", 3, first, absolute=True)
    elif preamble == MODE_RAW16:
        out["Tran"].extend(unpack16(body[3:first]))
    else:
        out["Tran"].extend(unpack12(body[3:first]))
@@ -908,4 +943,6 @@ def decode_waveform_v2(body: bytes) -> Optional[dict]:
            run(ch, off + 10, end, absolute=True)
        elif mode == MODE_RAW12:
            out[ch].extend(unpack12(body[off + 10:end]))
        elif mode == MODE_RAW16:
            out[ch].extend(unpack16(body[off + 10:end]))
    return out
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 [project]
 name = "seismo-relay"
-version = "0.26.0"
+version = "0.30.0"
 description = "Python client and REST server for MiniMate Plus seismographs"
 requires-python = ">=3.10"
 dependencies = [
@@ -0,0 +1,91 @@
 #!/usr/bin/env python3
 """Detect NON-MOTION on a geophone channel: |mean| / peak.
 A geophone is a velocity sensor with no DC response, so its output over a
 record must integrate to ~zero — the ground does not relocate. Real motion
 therefore sits roughly half above and half below zero. Anything electrical —
 a charge-injection spike, a step, a parked pedestal — is one-sided.
    mp   = |mean| / peak            ~0 for motion, ~1 for a pedestal
    frac_neg = share of samples < 0  ~0.3-0.5 for motion, ~0 for a fault
 Why this beats the pre-trigger floor (`offset_scan3.py`): that detector's
 `spread <= 0.02` gate rejects any record whose floor is MOVING, which is
 exactly what an onset is — it discarded the one BE18438 record in which the
 ramp was visible. This test is indifferent to whether the fault is a spike,
 a ramp or a flat pedestal; none of them cross zero.
 ⚠ Not a rediscovery of the retracted v1 detector. v1 scored only the
 largest-peak axis and used the mean as a BASELINE estimator, where the median
 was required. Here the mean is the signal itself, per channel, and that is
 what the physics licenses.
 """
 from __future__ import annotations
 import argparse, csv, re, statistics, sys
 from concurrent.futures import ProcessPoolExecutor, as_completed
 from pathlib import Path
 sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
 from minimateplus.event_file_io import read_blastware_file
 GEO=("Tran","Vert","Long"); K=10.0/32000.0
 _WAVE=re.compile(r"\.[A-Za-z0-9]{2}0[Ww]$"); _STEM=re.compile(r"^([B-Z])(\d{3})")
 _SER=re.compile(rb"[A-Z]{2}\d{3,6}")
 def serial_of(name, path=None):
    m=_STEM.match(name)
    if not m: return "?"
    num=(ord(m.group(1))-ord("B"))*1000+int(m.group(2))
    if path is not None:
        try:
            for s in _SER.findall(Path(path).read_bytes()):
                s=s.decode()
                if s[2:].lstrip("0")==str(num): return s
        except Exception: pass
    return f"BE{num}"
 def scan(ps):
    import logging; logging.disable(logging.WARNING)
    p=Path(ps)
    try: ev=read_blastware_file(p)
    except Exception: return None
    s=ev.raw_samples or {}
    if not all(s.get(c) for c in GEO): return None
    ts=ev.timestamp
    stamp=(f"{ts.year:04d}-{ts.month:02d}-{ts.day:02d}T"
           f"{ts.hour:02d}:{ts.minute:02d}:{ts.second:02d}") if ts else ""
    ser=serial_of(p.name,p); out=[]
    for ch in GEO:
        a=[x*K for x in s[ch]]
        pk=max(abs(x) for x in a)
        if pk<=0: continue
        out.append({"serial":ser,"timestamp":stamp,"filename":p.name,"channel":ch,
                    "peak":round(pk,4),
                    "mean":round(statistics.fmean(a),4),
                    "mp":round(abs(statistics.fmean(a))/pk,4),
                    "frac_neg":round(sum(1 for x in a if x<0)/len(a),4),
                    "n":len(a)})
    return out
 COLS=["serial","timestamp","filename","channel","peak","mean","mp","frac_neg","n"]
 def main():
    ap=argparse.ArgumentParser()
    ap.add_argument("--dir",required=True); ap.add_argument("--out",required=True)
    ap.add_argument("--jobs",type=int,default=4)
    a=ap.parse_args()
    seen=set(); files=[]
    for q in sorted(Path(a.dir).rglob("*")):
        if q.is_file() and _WAVE.search(q.name) and q.name not in seen:
            seen.add(q.name); files.append(str(q))
    print(f"unique waveform binaries: {len(files)}",flush=True)
    rows=[]
    with ProcessPoolExecutor(max_workers=a.jobs) as ex:
        for i,f in enumerate(as_completed([ex.submit(scan,p) for p in files]),1):
            r=f.result()
            if r: rows.extend(r)
            if i%1000==0: print(f"  {i}/{len(files)}",flush=True)
    with open(a.out,"w",newline="") as fh:
        w=csv.DictWriter(fh,fieldnames=COLS); w.writeheader(); w.writerows(rows)
    print(f"\nwrote {a.out}  ({len(rows)} channel-rows)")
 if __name__=="__main__": main()
@@ -0,0 +1,234 @@
 #!/usr/bin/env python3
 """Offset detector — HISTOGRAM corpus (the other 90% of the archive).
 `offset_scan3.py` measures the pre-trigger floor in *waveform* samples.  That
 covers 6,577 of the archive's 70,112 unique series-3 files; the remaining
 63,535 are **histograms**, which carry no samples — only a per-interval,
 per-channel peak + half-period.  So the pre-trigger method cannot run on them.
 The histogram analogue of "the resting floor" is the **low percentile of the
 per-interval peaks**.  A histogram file is typically hours of continuous
 monitoring, so the great majority of its intervals are definitionally quiet;
 the bottom of that distribution is what the channel reads when nothing is
 happening.  A healthy channel bottoms out at 0.000-0.005 in/s.  A channel
 parked off zero cannot report a peak below its own displacement, so its floor
 is pinned up.
 ⚠ The DC leakage into the histogram peak is PARTIAL.  Measured within-unit
 against episodes already established from the waveform scan:
    BE18438  Vert  in-episode 0.0350 vs 0.0050 outside  (waveform pre = +0.18..+0.37)
    BE12599  Tran  in-episode 0.0250 vs 0.0050 outside  (waveform pre = +0.03..+0.49)
 so the device's per-interval peak is evidently measured against a running /
 AC-coupled baseline that removes most, but not all, of the DC.  The residual
 is real and channel-specific, but the margin is ~5 quantisation counts rather
 than the ~70 the waveform detector enjoys.  Do not carry the waveform
 detector's 0.025 in/s floor across unexamined — calibrate on the CSV.
 Because the absolute floor also moves with site noise (traffic, wind, a
 generator), the statistic that matters most is the **cross-channel
 differential**: a channel's floor minus the quietest of the other two geo
 channels in the same file.  Site noise lifts all three together and cancels;
 a DC offset lifts one.
 This script does not decide anything.  It emits every candidate statistic per
 (file, channel) so thresholds can be calibrated against the waveform-derived
 ground truth in `offset_v3.csv` rather than guessed.
 Usage:
    python scratch/offset_hist_scan.py --dir /home/serversdown/dl2-archive/files \
        --out /home/serversdown/dl2-archive/offset_hist.csv --jobs 4
 """
 from __future__ import annotations
 import argparse
 import csv
 import datetime
 import logging
 import re
 import statistics
 import sys
 from concurrent.futures import ProcessPoolExecutor, as_completed
 from pathlib import Path
 sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
 from minimateplus.event_file_io import read_blastware_file  # noqa: E402
 GEO = ("Tran", "Vert", "Long")
 K = 10.0 / 32000.0          # ADC count -> in/s (see CLAUDE.md: full scale 32000)
 _HIST = re.compile(r"\.[A-Za-z0-9]{2}0[Hh]$")
 _STEM = re.compile(r"^([B-Z])(\d{3})")
 _B36 = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"
 _SERIAL_RE = re.compile(rb"\b([A-Z]{2}\d{3,6})\b")
 def serial_of(name: str, path=None) -> str:
    """Real serial for a BW file.
    The filename encodes only the NUMBER: `<letter><3 digits>` where
    letter = chr(ord('B') + serial // 1000).  The two-letter family prefix
    ("BE", "BA", ...) is **not** in the filename, so it must be read out of
    the file body.  Four units in the DL2 archive are BA, not BE — assuming
    "BE" mislabels BA9229, BA10060, BA10895 and BA15957.
    """
    m = _STEM.match(name)
    if not m:
        return "?"
    num = (ord(m.group(1)) - ord("B")) * 1000 + int(m.group(2))
    if path is not None:
        try:
            for s in _SERIAL_RE.findall(Path(path).read_bytes()):
                s = s.decode()
                if s[2:].lstrip("0") == str(num):
                    return s
        except Exception:
            pass
    return f"BE{num}"          # last-resort fallback; prefix unverified
 def stem_time(name: str):
    """Decode the filename's base-36 timestamp.  Epoch 1985-01-01, 1296 s/tick.
    Preferred over the file's own footer timestamp only because it costs
    nothing; the caller falls back to the decoded event when this fails.
    """
    try:
        base, ext = name.rsplit(".", 1)
        n = 0
        for c in base[4:8].upper():
            n = n * 36 + _B36.index(c)
        ab = _B36.index(ext[0].upper()) * 36 + _B36.index(ext[1].upper())
        return datetime.datetime(1985, 1, 1) + datetime.timedelta(seconds=n * 1296 + ab)
    except Exception:
        return None
 def _pct(sorted_vals, q):
    """Nearest-rank percentile on an already-sorted list."""
    if not sorted_vals:
        return None
    i = min(len(sorted_vals) - 1, max(0, int(len(sorted_vals) * q / 100.0)))
    return sorted_vals[i]
 def scan(path_str: str):
    logging.disable(logging.WARNING)   # per-worker: the codec warns on undecodables
    p = Path(path_str)
    try:
        ev = read_blastware_file(p)
    except Exception:
        return None
    s = ev.raw_samples or {}
    if not any(s.get(c) for c in GEO):
        return None
    ts = stem_time(p.name) or ev.timestamp
    stamp = ""
    if ts is not None:
        stamp = (f"{ts.year:04d}-{ts.month:02d}-{ts.day:02d}T"
                 f"{ts.hour:02d}:{ts.minute:02d}:{ts.second:02d}")
    # Per-channel floor candidates, in in/s.
    stats = {}
    for ch in GEO:
        v = sorted(s.get(ch) or [])
        if not v:
            continue
        stats[ch] = {
            "n": len(v),
            "min": v[0] * K,
            "p1": _pct(v, 1) * K,
            "p5": _pct(v, 5) * K,
            "p10": _pct(v, 10) * K,
            "p25": _pct(v, 25) * K,
            "med": statistics.median(v) * K,
            "peak": v[-1] * K,
            "zeros": sum(1 for x in v if x == 0) / len(v),
        }
    if len(stats) < 2:      # need at least one sibling channel for the differential
        return None
    # Mic floor as a site-noise proxy (raw counts; the dB conversion is not
    # needed — only its relative movement matters here).
    mic = sorted(s.get("MicL") or [])
    mic_p5 = _pct(mic, 5) if mic else ""
    rows = []
    for ch, st in stats.items():
        others = [stats[o]["p5"] for o in stats if o != ch]
        rows.append({
            "serial": serial_of(p.name, p),
            "timestamp": stamp,
            "filename": p.name,
            "channel": ch,
            "n_intervals": st["n"],
            "min": round(st["min"], 4),
            "p1": round(st["p1"], 4),
            "p5": round(st["p5"], 4),
            "p10": round(st["p10"], 4),
            "p25": round(st["p25"], 4),
            "median": round(st["med"], 4),
            "peak": round(st["peak"], 4),
            "frac_zero": round(st["zeros"], 4),
            # the site-noise-cancelling statistic: this channel's floor above
            # the quietest sibling geo channel in the same file
            "diff_p5": round(st["p5"] - min(others), 4),
            "mic_p5": mic_p5,
        })
    return rows
 COLS = ["serial", "timestamp", "filename", "channel", "n_intervals",
        "min", "p1", "p5", "p10", "p25", "median", "peak", "frac_zero",
        "diff_p5", "mic_p5"]
 def main():
    ap = argparse.ArgumentParser()
    ap.add_argument("--dir", required=True)
    ap.add_argument("--out", required=True)
    ap.add_argument("--jobs", type=int, default=4)
    ap.add_argument("--limit", type=int, default=0, help="stop after N files (smoke test)")
    a = ap.parse_args()
    # Dedupe by basename — the DL2 export keeps a byte-identical `Sent/`
    # mirror of its root, which doubled two figures before it was caught.
    seen, files = set(), []
    for q in sorted(Path(a.dir).rglob("*")):
        if q.is_file() and _HIST.search(q.name) and q.name not in seen:
            seen.add(q.name)
            files.append(str(q))
    if a.limit:
        files = files[:a.limit]
    print(f"unique histogram binaries: {len(files)}", flush=True)
    rows, undecodable = [], 0
    with ProcessPoolExecutor(max_workers=a.jobs) as ex:
        futs = [ex.submit(scan, f) for f in files]
        for i, fut in enumerate(as_completed(futs), 1):
            r = fut.result()
            if r:
                rows.extend(r)
            else:
                undecodable += 1
            if i % 5000 == 0:
                print(f"  {i}/{len(files)}", flush=True)
    with open(a.out, "w", newline="") as fh:
        w = csv.DictWriter(fh, fieldnames=COLS)
        w.writeheader()
        w.writerows(rows)
    files_ok = len({r["filename"] for r in rows})
    units = len({r["serial"] for r in rows})
    ivals = sum(r["n_intervals"] for r in rows) // 3
    print(f"\ndecoded {files_ok}/{len(files)} files "
          f"({undecodable} undecodable), {units} units, ~{ivals/1e6:.1f}M intervals")
    print(f"wrote {a.out}  ({len(rows)} channel-rows)")
 if __name__ == "__main__":
    main()
@@ -0,0 +1,171 @@
 #!/usr/bin/env python3
 """Scan series-3 waveform binaries for the 'offset' hardware fault.
 A healthy geophone trace is centred on zero.  An offset unit sits displaced,
 so the channel mean approaches its own peak.  Detector (unchanged from the
 2026-08-25 run, see memory note `offset-archive-analysis-backlog`):
    dominant-axis  |mean| / peak > 0.7
    AND            |mean| >= 0.9 * the unit's geo trigger level
 Trigger level is read from a paired _ASCII.TXT where one exists, otherwise
 from a per-serial median learned across that unit's ASCII files, otherwise
 --default-trigger.
 Serial is decoded from the BW filename: prefix letter encodes thousands
 (chr(ord('B') + n)), next 3 digits the remainder -- T193 -> BE18193.
 Usage:
  python scratch/offset_scan.py --dir <path> [--jobs N] --out offsets.csv
 """
 from __future__ import annotations
 import argparse, csv, json, re, sys
 from collections import defaultdict
 from concurrent.futures import ProcessPoolExecutor, as_completed
 from pathlib import Path
 sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
 from minimateplus.event_file_io import read_blastware_file
 from minimateplus.bw_ascii_report import parse_report
 GEO = ("Tran", "Vert", "Long")
 _GEO_FS_COUNTS = 32000.0
 _WAVE_RE = re.compile(r"\.[A-Za-z0-9]{2}0[Ww]$")
 _STEM_RE = re.compile(r"^([B-Z])(\d{3})")
 MEAN_OVER_PEAK_MIN = 0.7
 TRIGGER_FRACTION   = 0.9
 def serial_from_name(name: str):
    m = _STEM_RE.match(name)
    if not m:
        return None
    letter, digits = m.group(1), m.group(2)
    return f"BE{(ord(letter) - ord('B')) * 1000 + int(digits)}"
 def counts_to_ips(c, gr):
    return c * (gr or 10.0) / _GEO_FS_COUNTS
 def scan_one(path_str: str, default_trigger: float) -> dict | None:
    p = Path(path_str)
    try:
        gr, trig = 10.0, None
        ap = p.with_name(p.name.replace(".", "_", 1) + "_ASCII.TXT") \
            if False else p.parent / (p.stem + "_" + p.suffix.lstrip(".") + "_ASCII.TXT")
        if ap.exists():
            rep = parse_report(ap.read_text(errors="replace"))
            gr = rep.geo_range_ips or 10.0
            trig = rep.geo_trigger_level_ips
        ev = read_blastware_file(p)
        s = ev.raw_samples or {}
        if not all(s.get(c) for c in GEO):
            return None
        best = None
        for ch in GEO:
            arr = s[ch]
            n = len(arr)
            if n == 0:
                continue
            mean = sum(arr) / n
            peak = max(abs(v) for v in arr)
            if peak == 0:
                continue
            ratio = abs(mean) / peak
            if best is None or peak > best["peak_counts"]:
                best = {"channel": ch, "mean_counts": mean,
                        "peak_counts": peak, "ratio": ratio}
        if best is None:
            return None
        ts = ev.timestamp
        return {
            "serial": serial_from_name(p.name) or "?",
            "timestamp": (f"{ts.year:04d}-{ts.month:02d}-{ts.day:02d}T"
                          f"{ts.hour:02d}:{ts.minute:02d}:{ts.second:02d}") if ts else "",
            "filename": p.name,
            "channel": best["channel"],
            "offset_ips": round(counts_to_ips(best["mean_counts"], gr), 4),
            "peak_ips": round(counts_to_ips(best["peak_counts"], gr), 4),
            "mean_over_peak": round(best["ratio"], 3),
            "trigger_level_ips": trig if trig is not None else "",
            "geo_range_ips": gr,
        }
    except Exception:
        return None
 def main():
    ap = argparse.ArgumentParser()
    ap.add_argument("--dir", required=True)
    ap.add_argument("--jobs", type=int, default=4)
    ap.add_argument("--limit", type=int, default=0)
    ap.add_argument("--default-trigger", type=float, default=0.2)
    ap.add_argument("--out", required=True)
    a = ap.parse_args()
    # The DL2 export keeps a byte-identical `Sent/` mirror of the root, so
    # enumerate paths but keep only the first occurrence of each basename —
    # otherwise every event is counted twice.
    seen = set()
    files = []
    for q in sorted(Path(a.dir).rglob("*")):
        if q.is_file() and _WAVE_RE.search(q.name) and q.name not in seen:
            seen.add(q.name)
            files.append(q)
    if a.limit:
        files = files[: a.limit]
    print(f"waveform binaries to scan: {len(files)}", flush=True)
    rows = []
    with ProcessPoolExecutor(max_workers=a.jobs) as ex:
        futs = [ex.submit(scan_one, str(p), a.default_trigger) for p in files]
        for n, f in enumerate(as_completed(futs), 1):
            r = f.result()
            if r:
                rows.append(r)
            if n % 2000 == 0:
                print(f"  {n}/{len(files)}", flush=True)
    # learn per-serial trigger levels from the rows that had an ASCII
    by_serial = defaultdict(list)
    for r in rows:
        if r["trigger_level_ips"] != "":
            by_serial[r["serial"]].append(float(r["trigger_level_ips"]))
    med = {}
    for k, v in by_serial.items():
        v.sort()
        med[k] = v[len(v) // 2]
    for r in rows:
        if r["trigger_level_ips"] == "":
            r["trigger_level_ips"] = med.get(r["serial"], a.default_trigger)
        r["suspect"] = int(
            r["mean_over_peak"] > MEAN_OVER_PEAK_MIN
            and abs(r["offset_ips"]) >= TRIGGER_FRACTION * float(r["trigger_level_ips"])
        )
    cols = ["serial", "timestamp", "filename", "channel", "offset_ips", "peak_ips",
            "mean_over_peak", "trigger_level_ips", "geo_range_ips", "suspect"]
    with open(a.out, "w", newline="") as fh:
        w = csv.DictWriter(fh, fieldnames=cols)
        w.writeheader()
        w.writerows(rows)
    sus = [r for r in rows if r["suspect"]]
    print(f"\nscanned {len(rows)} decodable waveforms")
    print(f"suspect events: {len(sus)}")
    per = defaultdict(int)
    for r in sus:
        per[r["serial"]] += 1
    print(f"units with >=1 suspect event: {len(per)} of {len({r['serial'] for r in rows})}")
    for s, n in sorted(per.items(), key=lambda x: -x[1])[:20]:
        print(f"  {s:10} {n}")
    print(f"\nwrote {a.out}")
 if __name__ == "__main__":
    main()
@@ -0,0 +1,108 @@
 #!/usr/bin/env python3
 """Offset detector v2 — per-channel MEDIAN pedestal.
 Supersedes the dominant-axis / mean detector in offset_scan.py, which had two
 flaws that manufactured false "recoveries":
  1. It scored only the axis with the largest peak, so a real event on one axis
     hid a persistent pedestal on another.  BE12599 2026-08-21 read "clean"
     because Long had a 1.065 in/s event, while Tran sat at +0.47 in/s.
  2. It used the MEAN, which a real transient perturbs.  The median is the
     resting baseline: most samples sit at it, so a blast does not move it.
     Same event, Long: mean +0.0783 vs median -0.0050.
 Flags a CHANNEL when |median| >= --floor in/s (default 0.025 = 5 A/D counts,
 Instantel's own criterion; 1 A/D count = 0.005 in/s).
 Emits one row per (event, channel) so persistence can be tracked per channel.
 """
 from __future__ import annotations
 import argparse, csv, re, statistics, sys
 from concurrent.futures import ProcessPoolExecutor, as_completed
 from pathlib import Path
 sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
 from minimateplus.event_file_io import read_blastware_file
 GEO = ("Tran", "Vert", "Long")
 K = 10.0 / 32000.0                      # ADC counts -> in/s at the 10 in/s range
 _WAVE_RE = re.compile(r"\.[A-Za-z0-9]{2}0[Ww]$")
 _STEM_RE = re.compile(r"^([B-Z])(\d{3})")
 def serial_from_name(n):
    m = _STEM_RE.match(n)
    return f"BE{(ord(m.group(1))-ord('B'))*1000+int(m.group(2))}" if m else "?"
 def scan_one(ps):
    p = Path(ps)
    try:
        ev = read_blastware_file(p)
        s = ev.raw_samples or {}
        if not all(s.get(c) for c in GEO):
            return None
        ts = ev.timestamp
        stamp = (f"{ts.year:04d}-{ts.month:02d}-{ts.day:02d}T"
                 f"{ts.hour:02d}:{ts.minute:02d}:{ts.second:02d}") if ts else ""
        out = []
        for ch in GEO:
            a = s[ch]
            out.append({
                "serial": serial_from_name(p.name), "timestamp": stamp,
                "filename": p.name, "channel": ch,
                "median_ips": round(statistics.median(a) * K, 4),
                "mean_ips":   round(statistics.fmean(a) * K, 4),
                "peak_ips":   round(max(abs(v) for v in a) * K, 4),
            })
        return out
    except Exception:
        return None
 def main():
    ap = argparse.ArgumentParser()
    ap.add_argument("--dir", required=True)
    ap.add_argument("--jobs", type=int, default=4)
    ap.add_argument("--floor", type=float, default=0.025)
    ap.add_argument("--out", required=True)
    a = ap.parse_args()
    seen, files = set(), []
    for q in sorted(Path(a.dir).rglob("*")):
        if q.is_file() and _WAVE_RE.search(q.name) and q.name not in seen:
            seen.add(q.name); files.append(str(q))
    print(f"unique waveform binaries: {len(files)}", flush=True)
    rows = []
    with ProcessPoolExecutor(max_workers=a.jobs) as ex:
        for n, f in enumerate(as_completed([ex.submit(scan_one, p) for p in files]), 1):
            r = f.result()
            if r: rows.extend(r)
            if n % 2000 == 0: print(f"  {n}/{len(files)}", flush=True)
    for r in rows:
        r["offset"] = int(abs(r["median_ips"]) >= a.floor)
    cols = ["serial","timestamp","filename","channel","median_ips","mean_ips","peak_ips","offset"]
    with open(a.out, "w", newline="") as fh:
        w = csv.DictWriter(fh, fieldnames=cols); w.writeheader(); w.writerows(rows)
    from collections import defaultdict
    ev_flagged = {(r["serial"], r["filename"]) for r in rows if r["offset"]}
    ev_all     = {(r["serial"], r["filename"]) for r in rows}
    per = defaultdict(set)
    for r in rows:
        if r["offset"]: per[r["serial"]].add(r["filename"])
    tot = defaultdict(set)
    for r in rows: tot[r["serial"]].add(r["filename"])
    print(f"\nfloor = {a.floor} in/s ({a.floor/0.005:.0f} A/D counts)")
    print(f"events with >=1 offset channel: {len(ev_flagged)} of {len(ev_all)}")
    print(f"units affected: {len(per)} of {len(tot)}")
    for s in sorted(per, key=lambda s: -len(per[s])):
        print(f"  {s:9} {len(per[s]):4} / {len(tot[s]):4} events")
    print(f"\nwrote {a.out}")
 if __name__ == "__main__":
    main()
@@ -0,0 +1,117 @@
 #!/usr/bin/env python3
 """Offset detector v3 — pre-trigger floor, with pre/mid/end consistency.
 Brian's method, and better than v2's whole-record median for one reason: the
 pre-trigger window is *definitionally* quiet (it is the buffer captured before
 the trigger fired), whereas a whole-record median is merely robust to the event.
 Per channel:
    pre  = median of the first `pretrig_samples` samples (STRT record)
    mid  = median of the middle third
    end  = median of the final third
    spread = max(pre,mid,end) - min(pre,mid,end)
 A DC offset is a *constant floor*: |pre| at or above the floor AND a small
 spread. A transient (settling, handling, a long-tailed event) moves one segment
 relative to the others and is rejected by the spread test.
 Floor default 0.025 in/s = 5 A/D counts (Instantel's own criterion; 1 count =
 0.005 in/s).  Quantisation is 0.005 in/s, so `spread` is measured in units of it.
 """
 from __future__ import annotations
 import argparse, csv, re, statistics, sys
 from concurrent.futures import ProcessPoolExecutor, as_completed
 from pathlib import Path
 sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
 from minimateplus.event_file_io import read_blastware_file
 GEO=("Tran","Vert","Long"); K=10.0/32000.0
 _WAVE=re.compile(r"\.[A-Za-z0-9]{2}0[Ww]$"); _STEM=re.compile(r"^([B-Z])(\d{3})")
 _SERIAL_RE = re.compile(rb"\b([A-Z]{2}\d{3,6})\b")
 def serial_of(name: str, path=None) -> str:
    """Real serial for a BW file.
    The filename encodes only the NUMBER: `<letter><3 digits>` where
    letter = chr(ord('B') + serial // 1000).  The two-letter family prefix
    ("BE", "BA", ...) is **not** in the filename, so it must be read out of
    the file body.  Four units in the DL2 archive are BA, not BE — assuming
    "BE" mislabels BA9229, BA10060, BA10895 and BA15957.
    """
    m = _STEM.match(name)
    if not m:
        return "?"
    num = (ord(m.group(1)) - ord("B")) * 1000 + int(m.group(2))
    if path is not None:
        try:
            for s in _SERIAL_RE.findall(Path(path).read_bytes()):
                s = s.decode()
                if s[2:].lstrip("0") == str(num):
                    return s
        except Exception:
            pass
    return f"BE{num}"          # last-resort fallback; prefix unverified
 def scan(ps):
    p=Path(ps)
    try:
        ev=read_blastware_file(p); s=ev.raw_samples or {}
        if not all(s.get(c) for c in GEO): return None
        pre_n=ev.pretrig_samples
        ts=ev.timestamp
        stamp=(f"{ts.year:04d}-{ts.month:02d}-{ts.day:02d}T"
               f"{ts.hour:02d}:{ts.minute:02d}:{ts.second:02d}") if ts else ""
        out=[]
        for ch in GEO:
            a=s[ch]; n=len(a); t=n//3
            pre = a[:pre_n] if (pre_n and 0 < pre_n < n) else a[:t]
            mid, end = a[t:2*t], a[2*t:]
            if not pre or not mid or not end: continue
            v=[statistics.median(x)*K for x in (pre,mid,end)]
            out.append({"serial":serial_of(p.name, p),"timestamp":stamp,
                        "filename":p.name,"channel":ch,
                        "pretrig_n": pre_n or 0,
                        "pre":round(v[0],4),"mid":round(v[1],4),"end":round(v[2],4),
                        "spread":round(max(v)-min(v),4),
                        "peak":round(max(abs(x) for x in a)*K,4)})
        return out
    except Exception:
        return None
 def main():
    ap=argparse.ArgumentParser()
    ap.add_argument("--dir",required=True); ap.add_argument("--jobs",type=int,default=4)
    ap.add_argument("--floor",type=float,default=0.025)
    ap.add_argument("--max-spread",type=float,default=0.02)
    ap.add_argument("--out",required=True)
    a=ap.parse_args()
    seen=set(); files=[]
    for q in sorted(Path(a.dir).rglob("*")):
        if q.is_file() and _WAVE.search(q.name) and q.name not in seen:
            seen.add(q.name); files.append(str(q))
    print(f"unique waveform binaries: {len(files)}",flush=True)
    rows=[]
    with ProcessPoolExecutor(max_workers=a.jobs) as ex:
        for i,f in enumerate(as_completed([ex.submit(scan,p) for p in files]),1):
            r=f.result()
            if r: rows.extend(r)
            if i%2000==0: print(f"  {i}/{len(files)}",flush=True)
    for r in rows:
        r["offset"]=int(abs(r["pre"])>=a.floor and r["spread"]<=a.max_spread)
    cols=["serial","timestamp","filename","channel","pretrig_n","pre","mid","end","spread","peak","offset"]
    with open(a.out,"w",newline="") as fh:
        w=csv.DictWriter(fh,fieldnames=cols); w.writeheader(); w.writerows(rows)
    from collections import defaultdict
    per=defaultdict(set); tot=defaultdict(set)
    for r in rows:
        tot[r["serial"]].add(r["filename"])
        if r["offset"]: per[r["serial"]].add(r["filename"])
    print(f"\nfloor={a.floor} in/s ({a.floor/0.005:.0f} counts)  max spread={a.max_spread}")
    print(f"units affected: {len(per)} of {len(tot)}")
    for s in sorted(per,key=lambda s:-len(per[s])):
        print(f"  {s:9} {len(per[s]):4} / {len(tot[s]):4} events")
    print(f"\nwrote {a.out}")
 if __name__=="__main__": main()
@@ -0,0 +1,221 @@
 #!/usr/bin/env python3
 """Verify the series-3 decoder against preserved Blastware ASCII exports.
 Pairs each `<stem>_<ext>_ASCII.TXT` with its binary `<stem>.<ext>`, decodes the
 binary with the production codec, and compares against BW's own export:
  waveform  — per-channel sample counts, then every sample value
  histogram — interval count, then every per-interval channel peak
 ADC counts convert as  ips = counts * geo_range_ips / 32000  (1 decoder unit =
 16 counts = 0.005 in/s at the 10 in/s range; see CLAUDE.md).
 Usage:
  python scratch/verify_against_ascii.py --dir <path> [--limit N] [--jobs N]
                                         [--out results.json] [--kind w|h|all]
 """
 from __future__ import annotations
 import argparse, json, re, sys, traceback
 from concurrent.futures import ProcessPoolExecutor, as_completed
 from pathlib import Path
 sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
 from minimateplus.event_file_io import read_blastware_file
 from minimateplus.bw_ascii_report import parse_report_file, parse_report
 GEO = ("Tran", "Vert", "Long")
 _GEO_FS_COUNTS = 32000.0
 _ASCII_SUFFIX_RE = re.compile(r"_ASCII\.TXT$", re.IGNORECASE)
 def binary_for(ascii_path: Path) -> Path:
    """H907KXOW_WC0H_ASCII.TXT -> H907KXOW.WC0H"""
    stem = _ASCII_SUFFIX_RE.sub("", ascii_path.name)
    if "_" not in stem:
        return ascii_path.with_name(stem)
    head, _, ext = stem.rpartition("_")
    return ascii_path.with_name(f"{head}.{ext}")
 def counts_to_ips(counts, geo_range_ips):
    r = geo_range_ips if geo_range_ips else 10.0
    return counts * r / _GEO_FS_COUNTS
 def agrees(got, exp, geo_range_ips, tol=0.0006):
    """True when decoded `got` matches BW's exported `exp`.
    Saturation carve-out: when an event clips, BW clamps its export to the
    channel's range maximum (and writes OORANGE for the summary PPV), while
    the decoder faithfully reproduces raw counts that can sit a decoder unit
    or two past nominal full scale (32016 counts observed = 10.005 in/s on
    the 10 in/s range).  Same sign and both at/above the ceiling is agreement,
    not a decode error.
    """
    if abs(got - exp) <= tol:
        return True
    r = geo_range_ips if geo_range_ips else 10.0
    if abs(exp) >= r - tol and abs(got) >= r - tol and (got >= 0) == (exp >= 0):
        return True
    return False
 def parse_interval_table(text: str):
    """Histogram interval rows: time, Tpk, Tfq, Vpk, Vfq, Lpk, Lfq, PVS, ..., micdB, micfq"""
    rows = []
    seen_header = False
    for line in text.splitlines():
        if "\t" not in line:
            continue
        cols = [c.strip().strip('"') for c in line.split("\t")]
        cols = [c for c in cols if c != ""]
        if not seen_header:
            if any(c in ("Tran", "Vert", "Long") for c in cols):
                seen_header = True
            continue
        if len(cols) < 7:
            continue
        if not re.match(r"^\d{1,2}:\d{2}:\d{2}$", cols[0]):
            continue
        def num(s):
            try:
                return float(s)
            except ValueError:
                return None
        rows.append({"time": cols[0], "Tran": num(cols[1]),
                     "Vert": num(cols[3]), "Long": num(cols[5])})
    return rows
 def check_one(ascii_path_str: str) -> dict:
    ap = Path(ascii_path_str)
    bp = binary_for(ap)
    res = {"ascii": ap.name, "binary": bp.name, "status": "?",
           "kind": None, "detail": ""}
    try:
        if not bp.exists():
            res["status"] = "no_binary"
            return res
        text = ap.read_text(errors="replace")
        rep = parse_report(text, parse_samples=True)
        ev = read_blastware_file(bp)
        gr = rep.geo_range_ips
        res["kind"] = kind = ("histogram"
                              if (rep.event_type or "").lower().startswith(("full histogram", "histogram"))
                              else "waveform")
        samples = ev.raw_samples or {}
        dec_n = {c: len(samples.get(c) or []) for c in GEO}
        if kind == "histogram":
            rows = parse_interval_table(text)
            res["n_ascii"] = len(rows)
            res["n_decoded"] = dec_n["Tran"]
            if not rows:
                res["status"] = "no_ascii_table"
                return res
            if dec_n["Tran"] == 0:
                res["status"] = "decode_empty"
                return res
            if dec_n["Tran"] != len(rows):
                res["status"] = "count_mismatch"
                res["detail"] = f"decoded {dec_n['Tran']} vs ascii {len(rows)}"
                return res
            bad = 0
            worst = 0.0
            for i, row in enumerate(rows):
                for ch in GEO:
                    exp = row[ch]
                    if exp is None:
                        continue
                    got = counts_to_ips(samples[ch][i], gr)
                    if not agrees(got, exp, gr):
                        bad += 1
                        worst = max(worst, abs(got - exp))
            res["worst_abs"] = round(worst, 6)
            res["status"] = "exact" if bad == 0 else "value_mismatch"
            if bad:
                res["detail"] = f"{bad} interval-channel values off"
            return res
        # waveform
        asc = rep.samples or []
        res["n_ascii"] = len(asc)
        res["n_decoded"] = dec_n["Tran"]
        if not asc:
            res["status"] = "no_ascii_table"
            return res
        if dec_n["Tran"] == 0:
            res["status"] = "decode_empty"
            return res
        if len({dec_n[c] for c in GEO}) != 1:
            res["status"] = "channel_len_mismatch"
            res["detail"] = str(dec_n)
            return res
        if dec_n["Tran"] != len(asc):
            res["status"] = "count_mismatch"
            res["detail"] = f"decoded {dec_n['Tran']} vs ascii {len(asc)}"
            return res
        bad = 0
        worst = 0.0
        for i, quad in enumerate(asc):
            for j, ch in enumerate(GEO):
                exp = quad[j]
                got = counts_to_ips(samples[ch][i], gr)
                if not agrees(got, exp, gr):
                    bad += 1
                    worst = max(worst, abs(got - exp))
        res["worst_abs"] = round(worst, 6)
        res["status"] = "exact" if bad == 0 else "value_mismatch"
        if bad:
            res["detail"] = f"{bad} sample values off"
        return res
    except Exception as e:
        res["status"] = "error"
        res["detail"] = f"{type(e).__name__}: {e}"
        return res
 def main():
    ap = argparse.ArgumentParser()
    ap.add_argument("--dir", required=True)
    ap.add_argument("--limit", type=int, default=0)
    ap.add_argument("--jobs", type=int, default=8)
    ap.add_argument("--kind", choices=["w", "h", "all"], default="all")
    ap.add_argument("--out", default=None)
    a = ap.parse_args()
    root = Path(a.dir)
    files = sorted(p for p in root.rglob("*")
                   if p.is_file() and p.name.upper().endswith("_ASCII.TXT"))
    if a.kind != "all":
        want = "0W" if a.kind == "w" else "0H"
        files = [p for p in files
                 if _ASCII_SUFFIX_RE.sub("", p.name).upper().endswith(want)]
    if a.limit:
        files = files[: a.limit]
    print(f"pairs to check: {len(files)}", flush=True)
    out = []
    from collections import Counter
    tally = Counter()
    with ProcessPoolExecutor(max_workers=a.jobs) as ex:
        futs = {ex.submit(check_one, str(p)): p for p in files}
        for n, f in enumerate(as_completed(futs), 1):
            r = f.result()
            out.append(r)
            tally[(r["kind"], r["status"])] += 1
            if n % 500 == 0:
                print(f"  {n}/{len(files)}", flush=True)
    print("\n=== results ===")
    for (kind, status), n in sorted(tally.items(), key=lambda x: -x[1]):
        print(f"  {str(kind):10} {status:22} {n}")
    if a.out:
        Path(a.out).write_text(json.dumps(out, indent=1))
        print(f"\nwrote {a.out}")
 if __name__ == "__main__":
    main()
@@ -0,0 +1,228 @@
 #!/usr/bin/env python3
 """Verify the Thor / Micromate (series-4) IDF decoder against Thor's own exports.
 Sister harness to ``scratch/verify_against_ascii.py`` (series-3 / Blastware).
 Ground truth is the ``.IDFW.csv`` / ``.IDFH.csv`` file Thor writes next to each
 binary, under a sibling ``CSV/`` directory:
    <dir>/UM13981_20220207084555.IDFW
    <dir>/CSV/UM13981_20220207084555.IDFW.csv
 For waveforms the CSV carries a per-sample block of four columns
 (Tran, Vert, Long, Mic) in in/s and psi -- i.e. true per-sample ground truth,
 exactly what the BW ASCII exports give us for series-3.  The leading 2-column
 rows are the report header (PPV, sample rate, geo range, ...).
 Usage:
    python scratch/verify_thor_against_csv.py [--root DIR] [--lsb FLOAT]
                                              [--limit N] [--kind idfw|idfh|both]
 """
 from __future__ import annotations
 import argparse
 import csv
 import os
 import statistics
 import sys
 from collections import Counter, defaultdict
 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
 from micromate import idf_file as M
 DEFAULT_ROOT = "/home/serversdown/thor-watcher/example-data"
 GEO = ("Tran", "Vert", "Long")
 def parse_export(path):
    """Return (header_dict, sample_rows) from a Thor CSV export."""
    hdr, rows = {}, []
    with open(path, newline="", encoding="utf-8", errors="replace") as fh:
        for rec in csv.reader(fh):
            if len(rec) == 2:
                hdr[rec[0].strip()] = rec[1].strip()
            elif len(rec) >= 3:
                try:
                    rows.append([float(x) for x in rec])
                except ValueError:
                    pass
    return hdr, rows
 def index_corpus(root):
    """Map BASENAME.IDFW -> (binary_path, csv_path) for every paired file."""
    exports, binaries = {}, {}
    for dirpath, _dirs, files in os.walk(root):
        for name in files:
            up = name.upper()
            full = os.path.join(dirpath, name)
            if up.endswith(".IDFW.CSV") or up.endswith(".IDFH.CSV"):
                exports.setdefault(name[:-4].upper(), full)
            elif up.endswith(".IDFW") or up.endswith(".IDFH"):
                binaries.setdefault(up, full)
    return {k: (binaries[k], exports[k]) for k in binaries.keys() & exports.keys()}
 def hdr_float(hdr, key):
    raw = hdr.get(key)
    if not raw:
        return None
    try:
        return float(raw.split()[0])
    except (ValueError, IndexError):
        return None
 def verify_waveform(binpath, csvpath, lsb):
    """Compare one IDFW against its export.  Returns a result dict."""
    out = {"file": os.path.basename(binpath), "status": "ok"}
    try:
        res = M.read_idf_file(binpath)
    except NotImplementedError:
        out["status"] = "not-thor"
        return out
    except Exception as exc:  # noqa: BLE001 - harness reports, never raises
        out["status"] = "decode-error"
        out["detail"] = f"{type(exc).__name__}: {exc}"
        return out
    hdr, rows = parse_export(csvpath)
    if not rows:
        out["status"] = "no-gt-samples"
        return out
    gt = {ch: [r[i] for r in rows] for i, ch in enumerate(GEO)}
    out["gt_len"] = len(rows)
    out["geo_range"] = hdr.get("GeoRange")
    exact = total = 0
    lens, chan_status = {}, {}
    ppv_err = {}
    for ch in GEO:
        arr = res.samples.get(ch, [])
        ref = gt[ch]
        lens[ch] = len(arr)
        if len(arr) != len(ref):
            chan_status[ch] = "length"
            continue
        if not arr:
            chan_status[ch] = "empty"
            continue
        hits = sum(1 for c, v in zip(arr, ref) if abs(c * lsb - v) < 5e-5)
        exact += hits
        total += len(arr)
        chan_status[ch] = "exact" if hits == len(arr) else "value"
        gp = hdr_float(hdr, f"{ch}PPV")
        if gp:
            ppv_err[ch] = (max(abs(c) for c in arr) * lsb - gp) / gp
    out["lens"] = lens
    out["chan_status"] = chan_status
    out["exact"] = exact
    out["total"] = total
    out["ppv_err"] = ppv_err
    if all(v == "exact" for v in chan_status.values()):
        out["status"] = "exact"
    elif any(v == "length" for v in chan_status.values()):
        out["status"] = "length-mismatch"
    else:
        out["status"] = "value-mismatch"
    return out
 def verify_histogram(binpath, csvpath, lsb):
    out = {"file": os.path.basename(binpath), "status": "ok"}
    try:
        res = M.read_idf_file(binpath)
    except NotImplementedError:
        out["status"] = "not-thor"
        return out
    except Exception as exc:  # noqa: BLE001
        out["status"] = "decode-error"
        out["detail"] = f"{type(exc).__name__}: {exc}"
        return out
    hdr, _rows = parse_export(csvpath)
    out["n_intervals"] = len(res.intervals or [])
    errs = {}
    for ch, attr in (("Tran", "transverse_ips"), ("Vert", "vertical_ips"),
                     ("Long", "longitudinal_ips")):
        gp = hdr_float(hdr, f"{ch}PPV")
        dv = getattr(res.event.peaks, attr, None)
        if gp and dv:
            errs[ch] = (dv - gp) / gp
    out["ppv_err"] = errs
    out["status"] = "peaks" if errs else "no-gt-peaks"
    return out
 def main():
    ap = argparse.ArgumentParser()
    ap.add_argument("--root", default=DEFAULT_ROOT)
    ap.add_argument("--lsb", type=float, default=M._GEO_LSB_IPS)
    ap.add_argument("--limit", type=int, default=0)
    ap.add_argument("--kind", choices=("idfw", "idfh", "both"), default="both")
    ap.add_argument("--show", type=int, default=15, help="worst-N detail rows")
    args = ap.parse_args()
    pairs = index_corpus(args.root)
    keys = sorted(pairs)
    if args.kind != "both":
        keys = [k for k in keys if k.endswith(args.kind.upper())]
    if args.limit:
        keys = keys[: args.limit]
    print(f"root: {args.root}")
    print(f"geo LSB under test: {args.lsb!r} in/s per count")
    print(f"paired files: {len(keys)}\n")
    wf, hg = [], []
    for k in keys:
        binpath, csvpath = pairs[k]
        if k.endswith(".IDFW"):
            wf.append(verify_waveform(binpath, csvpath, args.lsb))
        else:
            hg.append(verify_histogram(binpath, csvpath, args.lsb))
    if wf:
        st = Counter(r["status"] for r in wf)
        ex = sum(r.get("exact", 0) for r in wf)
        tot = sum(r.get("total", 0) for r in wf)
        print("=" * 68)
        print(f"WAVEFORM (IDFW): {len(wf)} files")
        for s, n in st.most_common():
            print(f"    {s:16} {n:5d}  ({100*n/len(wf):5.1f}%)")
        if tot:
            print(f"  per-sample exact: {ex}/{tot} = {100*ex/tot:.3f}%")
        errs = [e for r in wf for e in r.get("ppv_err", {}).values()]
        if errs:
            print(f"  PPV rel-error: median {statistics.median(errs):+.4%}  "
                  f"mean {statistics.mean(errs):+.4%}  "
                  f"max|.| {max(abs(e) for e in errs):.4%}")
        bad = [r for r in wf if r["status"] not in ("exact",)]
        if bad:
            print(f"\n  worst {min(args.show, len(bad))} of {len(bad)} non-exact:")
            for r in bad[: args.show]:
                print(f"    {r['file']:42} {r['status']:16} "
                      f"lens={r.get('lens')} gt={r.get('gt_len')} "
                      f"{r.get('detail','')}")
    if hg:
        st = Counter(r["status"] for r in hg)
        print("=" * 68)
        print(f"HISTOGRAM (IDFH): {len(hg)} files")
        for s, n in st.most_common():
            print(f"    {s:16} {n:5d}  ({100*n/len(hg):5.1f}%)")
        errs = [e for r in hg for e in r.get("ppv_err", {}).values()]
        if errs:
            print(f"  PPV rel-error: median {statistics.median(errs):+.4%}  "
                  f"mean {statistics.mean(errs):+.4%}  "
                  f"max|.| {max(abs(e) for e in errs):.4%}")
            within = lambda t: 100*sum(1 for e in errs if abs(e) <= t)/len(errs)
            print(f"  within 0.5%: {within(0.005):.1f}%   "
                  f"within 2%: {within(0.02):.1f}%   within 5%: {within(0.05):.1f}%")
    return 0
 if __name__ == "__main__":
    raise SystemExit(main())
@@ -1,12 +1,12 @@
 #!/usr/bin/env python3
-"""Backfill events.shape_* from each event's .h5 waveform samples. Idempotent."""
+"""Backfill events.shape_* and shape_offset_* from each event's .h5 samples. Idempotent."""
 from __future__ import annotations
 import argparse, logging, sys
 from pathlib import Path
 sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
 from sfm.database import SeismoDb
 from sfm.waveform_store import WaveformStore
-from sfm.shape_metrics import shape_from_h5
+from sfm.shape_metrics import shape_from_h5, offset_from_h5
 log = logging.getLogger("backfill_event_shape")
@@ -21,6 +21,7 @@ def backfill_shape(db: SeismoDb, store: WaveformStore, *, dry_run: bool = False)
        if not h5_path.exists():
            counts["skipped_no_h5"] += 1; continue
        shape = shape_from_h5(h5_path)
        offset = offset_from_h5(h5_path)
        if shape is None:
            # The .h5 can no longer yield a shape (fewer than 2 samples, or a
            # flat trace).  Clear any previously stored value rather than
@@ -28,22 +29,32 @@ def backfill_shape(db: SeismoDb, store: WaveformStore, *, dry_run: bool = False)
            # from and silently feeds the false-trigger detector.  Seen after
            # a decoder fix shrinks an event: 493 rows in the prod snapshot
            # were carrying metrics from a superseded decode (2026-08-25).
-            if row.get("shape_crest_factor") is not None:
+            if (row.get("shape_crest_factor") is not None
                    or row.get("shape_offset") is not None):
                if not dry_run:
                    with db._connect() as conn:
                        conn.execute(
                            "UPDATE events SET shape_crest_factor=NULL, "
                            "shape_near_peak_count=NULL, shape_sample_count=NULL, "
-                            "shape_axis=NULL WHERE id=?", (row["id"],))
+                            "shape_axis=NULL, shape_offset=NULL, shape_offset_axis=NULL, "
                            "shape_offset_pre=NULL, shape_offset_spread=NULL WHERE id=?",
                            (row["id"],))
                counts["cleared_stale"] += 1
            counts["skipped_no_samples"] += 1; continue
        if not dry_run:
            with db._connect() as conn:
                conn.execute(
                    "UPDATE events SET shape_crest_factor=?, shape_near_peak_count=?, "
-                    "shape_sample_count=?, shape_axis=? WHERE id=?",
+                    "shape_sample_count=?, shape_axis=?, shape_offset=?, "
                    "shape_offset_axis=?, shape_offset_pre=?, shape_offset_spread=? "
                    "WHERE id=?",
                    (shape["crest_factor"], shape["near_peak_count"],
-                     shape["sample_count"], shape["axis"], row["id"]))
+                     shape["sample_count"], shape["axis"],
                     (1 if offset["offset"] else 0) if offset else None,
                     offset["axis"] if offset else None,
                     offset["pre"] if offset else None,
                     offset["spread"] if offset else None,
                     row["id"]))
        counts["updated"] += 1
    log.info("backfill_shape: %s", counts)
    return counts
@@ -54,6 +54,7 @@ from s3_analyzer import (  # noqa: E402
    write_claude_export,
 )
 from frame_db import FrameDB  # noqa: E402
 from minimateplus.binary_annotate import annotate_blastware_binary  # noqa: E402
 # ── colour palette ────────────────────────────────────────────────────────────
 BG       = "#1e1e1e"
@@ -2675,6 +2676,95 @@ class DownloadPanel(tk.Frame):
        self._on_capture_ready(bw_path, s3_path, label)
 # ─────────────────────────────────────────────────────────────────────────────
 # Inspector panel — annotated hex view of a Series-3 binary
 # ─────────────────────────────────────────────────────────────────────────────
 class InspectorPanel(tk.Frame):
    """Load any Series-3 waveform binary and read it as an annotated hex dump.
    Regions the decoder understands (header, STRT, per-channel sample records,
    footer) are labelled and colour-coded; everything the decoder cannot account
    for is flagged UNKNOWN, so undecoded bytes stand out for hand-inspection.
    """
    _KIND_COLOR = {
        "header":  ACCENT,
        "strt":    YELLOW,
        "sample":  COL_S3,
        "footer":  FG_DIM,
        "unknown": RED,
    }
    def __init__(self, parent: tk.Widget, initialdir=None, **kw) -> None:
        super().__init__(parent, bg=BG, **kw)
        self._path = None
        self._initialdir = initialdir
        self._build()
    def _build(self) -> None:
        bar = tk.Frame(self, bg=BG2)
        bar.pack(side=tk.TOP, fill=tk.X)
        tk.Button(bar, text="Open binary…", command=self._open, bg=BG3, fg=FG,
                  relief=tk.FLAT, font=MONO, activebackground=ACCENT).pack(side=tk.LEFT, padx=6, pady=6)
        self._path_var = tk.StringVar(value="(no file loaded)")
        tk.Label(bar, textvariable=self._path_var, bg=BG2, fg=FG_DIM, font=MONO).pack(side=tk.LEFT, padx=6)
        self._summary_var = tk.StringVar(value="")
        tk.Label(bar, textvariable=self._summary_var, bg=BG2, fg=FG, font=MONO).pack(side=tk.RIGHT, padx=10)
        legend = tk.Frame(self, bg=BG2)
        legend.pack(side=tk.TOP, fill=tk.X)
        tk.Label(legend, text="legend:", bg=BG2, fg=FG_DIM, font=MONO).pack(side=tk.LEFT, padx=(8, 2))
        for kind, color in self._KIND_COLOR.items():
            tk.Label(legend, text=f"■ {kind}", bg=BG2, fg=color, font=MONO).pack(side=tk.LEFT, padx=5, pady=2)
        self._text = scrolledtext.ScrolledText(
            self, bg=BG, fg=FG, insertbackground=FG, font=MONO, wrap=tk.NONE, borderwidth=0)
        self._text.pack(side=tk.TOP, fill=tk.BOTH, expand=True)
        for kind, color in self._KIND_COLOR.items():
            self._text.tag_configure(kind, foreground=color)
        self._text.tag_configure("label", foreground="#ffffff", font=("Consolas", 9, "bold"))
        self._text.tag_configure("dim", foreground=FG_DIM)
        self._text.configure(state=tk.DISABLED)
    def _open(self) -> None:
        p = filedialog.askopenfilename(title="Open a Series-3 binary", initialdir=self._initialdir)
        if p:
            self.load(Path(p))
    def load(self, path: Path) -> None:
        try:
            raw = path.read_bytes()
            spans = annotate_blastware_binary(raw)
        except Exception as e:  # noqa: BLE001 — surface any read/annotate failure to the user
            messagebox.showerror("Inspector", f"Failed to read/annotate:\n{path}\n\n{e}")
            return
        self._path = path
        self._path_var.set(str(path))
        self._render(raw, spans)
    def _render(self, raw: bytes, spans) -> None:
        t = self._text
        t.configure(state=tk.NORMAL)
        t.delete("1.0", tk.END)
        unknown = sum(s.end - s.start for s in spans if s.kind == "unknown")
        pct = 100 * unknown / max(1, len(raw))
        self._summary_var.set(f"{len(raw)} B · {len(spans)} regions · {pct:.1f}% unknown")
        for s in spans:
            t.insert(tk.END, f"\n── {s.label}  [0x{s.start:04x}:0x{s.end:04x}]  {s.end - s.start} B ──\n", ("label",))
            self._insert_hex(t, raw, s.start, s.end, s.kind)
        t.configure(state=tk.DISABLED)
    def _insert_hex(self, t: tk.Text, raw: bytes, start: int, end: int, kind: str) -> None:
        for off in range(start, end, 16):
            row = raw[off:min(off + 16, end)]
            hx = " ".join(f"{b:02x}" for b in row).ljust(16 * 3 - 1)
            txt = "".join(chr(b) if 32 <= b < 127 else "." for b in row)
            t.insert(tk.END, f"  0x{off:04x}  ", ("dim",))
            t.insert(tk.END, hx, (kind,))
            t.insert(tk.END, f"  {txt}\n", ("dim",))
 # ─────────────────────────────────────────────────────────────────────────────
 # Main application window
 # ─────────────────────────────────────────────────────────────────────────────
@@ -2730,6 +2820,9 @@ class SeismoLab(tk.Tk):
        )
        nb.add(self._download_panel, text="  Download  ")
        self._inspector_panel = InspectorPanel(nb)
        nb.add(self._inspector_panel, text="  Inspector  ")
        self._nb = nb
        self.protocol("WM_DELETE_WINDOW", self._on_close)
@@ -0,0 +1,133 @@
 """USBM RI8507 / OSMRE blasting compliance chart.
 Renders the velocity-vs-frequency compliance scatter Blastware draws on its Event
 Report: each channel's significant waveform cycles as ``(frequency, peak
 velocity)`` points on log-log axes against the regulatory limit curve(s). A point
 below the curve passes; above fails.
 Two pieces, kept separate so both can be reused/extended:
  * ``limit_at`` / ``limit_curve`` — the regulatory limit curve(s), as data.
  * ``channel_compliance_points`` — the per-cycle (freq, velocity) scatter, by
    the zero-crossing method (matches Blastware: each channel's cloud tops out
    at that channel's PPV).
 Limit curves (USBM RI8507 Figure B-1 / OSM 30 CFR 816.67), drawn CONTINUOUS — a
 constant-displacement bound (sloped, ``v = 2πf·d``) meets a constant-velocity
 plateau at the frequency where they're equal, so there are no vertical steps
 (matching how Blastware draws it). Two lines:
  * **Drywall** (modern gypsum board) — 0.75 in/s plateau (solid).
  * **Plaster** on wood lath (older homes) — 0.50 in/s plateau (dashed).
 Both use a 0.030 in low-frequency displacement bound and rise through a 0.010 in
 displacement bound to a 2.0 in/s high-frequency plateau. Values from USBM RI8507
 (Appendix B) / 30 CFR 816.67; ⚠ confirm the exact shape against a Blastware
 report before trusting for compliance.
 """
 from __future__ import annotations
 import math
 from typing import Dict, Sequence, Tuple
 import numpy as np
 from matplotlib.ticker import FixedLocator, NullLocator
 # curve name → (low-freq "ultimate" displacement in, mid velocity plateau in/s,
 #               high-freq displacement in, high-freq velocity plateau in/s).
 # RI8507 Fig B-1 (p.74): ultimate max displacement 0.030 in (< ~4 Hz), plateau
 # 0.75 (Drywall) / 0.50 (plaster), rising diagonal at 0.008 in displacement up to
 # a 2.0 in/s plateau reached at ~40 Hz.
 _CURVES: Dict[str, Tuple[float, float, float, float]] = {
    "Drywall": (0.030, 0.75, 0.008, 2.00),
    "Plaster": (0.030, 0.50, 0.008, 2.00),
 }
 # how each curve is stroked on the chart
 _CURVE_STYLE = {"Drywall": {"ls": "-", "lw": 1.0}, "Plaster": {"ls": "--", "lw": 0.9}}
 STANDARDS = tuple(_CURVES)
 # Blastware's channel markers/colours on the compliance chart.
 _CHANNEL_STYLE = {
    "Tran": ("+", "#d62728"),   # red +
    "Vert": ("x", "#2ca02c"),   # green x
    "Long": ("o", "#1f77b4"),   # blue o
 }
 def limit_at(freq_hz: float, curve: str = "Drywall") -> float:
    """Max allowed PPV (in/s) at ``freq_hz`` for ``curve`` (continuous)."""
    d_low, v_mid, d_high, v_high = _CURVES[curve]
    f = max(freq_hz, 1.0)
    f_a = v_mid / (2.0 * math.pi * d_low)    # disp_low → vel_mid
    f_b = v_mid / (2.0 * math.pi * d_high)   # vel_mid → disp_high
    f_c = v_high / (2.0 * math.pi * d_high)  # disp_high → vel_high
    if f <= f_a:
        return 2.0 * math.pi * f * d_low
    if f <= f_b:
        return v_mid
    if f <= f_c:
        return 2.0 * math.pi * f * d_high
    return v_high
 def limit_curve(curve: str = "Drywall", fmin: float = 1.0, fmax: float = 100.0, n: int = 400):
    """(freqs, limits) sampled across the band for plotting one curve."""
    freqs = np.logspace(np.log10(fmin), np.log10(fmax), n)
    return freqs, np.array([limit_at(f, curve) for f in freqs])
 def channel_compliance_points(
    samples: Sequence[float], sps: float, fmin: float = 1.0, fmax: float = 100.0,
    vmin: float = 0.0,
 ) -> Tuple[np.ndarray, np.ndarray]:
    """Per-cycle (frequency, peak velocity) scatter for one channel.
    Zero-crossing method: split the trace at sign changes; each half-cycle
    contributes one point at ``(1/(2·half_period), max|amplitude|)``. Matches
    Blastware — the cloud's ceiling is the channel PPV. ``samples`` must be in the
    velocity unit you want plotted (in/s). Points outside ``[fmin, fmax]`` or at
    or below ``vmin`` are dropped.
    """
    x = np.asarray(samples, dtype=float)
    if x.size < 3:
        return np.empty(0), np.empty(0)
    zc = np.where(np.diff(np.signbit(x)))[0]
    freqs, vels = [], []
    for a, b in zip(zc[:-1], zc[1:]):
        half_period = (b - a) / sps
        if half_period <= 0:
            continue
        freqs.append(1.0 / (2.0 * half_period))
        vels.append(float(np.abs(x[a:b + 1]).max()))
    f = np.array(freqs)
    v = np.array(vels)
    keep = (f >= fmin) & (f <= fmax) & (v > vmin)
    return f[keep], v[keep]
 def draw_compliance_chart(ax, channels: Dict[str, Sequence[float]], sps: float) -> None:
    """Draw the compliance chart (both limit curves + per-channel scatter)."""
    for name, style in _CURVE_STYLE.items():
        cf, cv = limit_curve(name)
        ax.plot(cf, cv, color="#333", zorder=3, **style)
    for ch, (marker, color) in _CHANNEL_STYLE.items():
        samples = channels.get(ch)
        if samples is None or len(samples) == 0:
            continue
        f, v = channel_compliance_points(samples, sps)
        ax.scatter(f, v, marker=marker, s=12, c=color, linewidths=0.7, zorder=4, label=ch)
    ax.set_xscale("log")
    ax.set_yscale("log")
    ax.set_xlim(1, 100)
    ax.set_ylim(0.0394, 10)
    ax.set_box_aspect(1)   # square plot box (log-log compliance charts are square)
    xt = [1, 2, 5, 10, 20, 50, 100]
    yt = [0.0394, 0.05, 0.1, 0.2, 0.5, 1, 2, 5, 10]
    ax.xaxis.set_major_locator(FixedLocator(xt)); ax.xaxis.set_minor_locator(NullLocator())
    ax.yaxis.set_major_locator(FixedLocator(yt)); ax.yaxis.set_minor_locator(NullLocator())
    ax.set_xticklabels([str(v) for v in xt])
    ax.set_yticklabels([("%g" % v) for v in yt])
    ax.set_xlabel("Frequency (Hz)", fontsize=7)
    ax.set_ylabel("Velocity (in/s)", fontsize=7)
    ax.tick_params(labelsize=6)
    ax.grid(True, which="both", ls=":", lw=0.4, color="#ccc")
@@ -82,6 +82,7 @@ CREATE TABLE IF NOT EXISTS events (
    record_type             TEXT,                   -- "single_shot" | "continuous"
    false_trigger           INTEGER NOT NULL DEFAULT 0,  -- 0=no, 1=yes (manual flag)
    reviewed_real           INTEGER NOT NULL DEFAULT 0,  -- 0=no, 1=operator-confirmed real (mutually exclusive with false_trigger)
    false_trigger_reason    TEXT,                   -- optional FT cause ("offset", ...); NULL = none. Only meaningful when false_trigger=1.
    blastware_filename      TEXT,                   -- event file within waveform store; extension is per-event (AB0T encodes timestamp)
    blastware_filesize      INTEGER,                -- bytes; NULL if no event file saved
    a5_pickle_filename      TEXT,                   -- "<filename>.a5.pkl" sidecar
@@ -99,6 +100,10 @@ CREATE TABLE IF NOT EXISTS events (
    shape_near_peak_count   INTEGER,                -- samples >= 0.5 * peak (FT: few; real: many)
    shape_sample_count      INTEGER,                -- total samples (to normalize near_peak_count)
    shape_axis              TEXT,                   -- geophone channel measured ("Tran"/"Vert"/"Long")
    shape_offset            INTEGER,                -- 1 = DC-offset false trigger (pre-trigger baseline off zero + flat). Meaningful for waveforms only.
    shape_offset_axis       TEXT,                   -- geo channel the offset was measured on
    shape_offset_pre        REAL,                   -- pre-trigger baseline median (in/s)
    shape_offset_spread     REAL,                   -- max(pre,mid,end) - min(...) in in/s; small = constant/DC
    created_at              TEXT NOT NULL DEFAULT (strftime('%Y-%m-%dT%H:%M:%SZ', 'now')),
    UNIQUE(serial, timestamp)
 );
@@ -225,7 +230,12 @@ class SeismoDb:
            ("shape_near_peak_count", "INTEGER"),
            ("shape_sample_count",    "INTEGER"),
            ("shape_axis",            "TEXT"),
            ("shape_offset",          "INTEGER"),
            ("shape_offset_axis",     "TEXT"),
            ("shape_offset_pre",      "REAL"),
            ("shape_offset_spread",   "REAL"),
            ("reviewed_real",         "INTEGER NOT NULL DEFAULT 0"),
            ("false_trigger_reason",  "TEXT"),
        ):
            if col not in existing_cols:
                log.info("_migrate: events ADD COLUMN %s %s", col, ddl)
@@ -430,9 +440,11 @@ class SeismoDb:
                             tran_zc_above_range, vert_zc_above_range,
                             long_zc_above_range, mic_zc_above_range,
                             shape_crest_factor, shape_near_peak_count,
-                             shape_sample_count, shape_axis)
+                             shape_sample_count, shape_axis,
                             shape_offset, shape_offset_axis,
                             shape_offset_pre, shape_offset_spread)
                        VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?,
-                                ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+                                ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
                        """,
                        (
                            self._new_id(), serial, key, session_id, ts,
@@ -464,6 +476,10 @@ class SeismoDb:
                            rec.get("shape_near_peak_count"),
                            rec.get("shape_sample_count"),
                            rec.get("shape_axis"),
                            rec.get("shape_offset"),
                            rec.get("shape_offset_axis"),
                            rec.get("shape_offset_pre"),
                            rec.get("shape_offset_spread"),
                        ),
                    )
                    inserted += 1
@@ -517,7 +533,11 @@ class SeismoDb:
                               shape_crest_factor    = COALESCE(?, shape_crest_factor),
                               shape_near_peak_count = COALESCE(?, shape_near_peak_count),
                               shape_sample_count    = COALESCE(?, shape_sample_count),
-                               shape_axis            = COALESCE(?, shape_axis)
+                               shape_axis            = COALESCE(?, shape_axis),
                               shape_offset          = COALESCE(?, shape_offset),
                               shape_offset_axis     = COALESCE(?, shape_offset_axis),
                               shape_offset_pre      = COALESCE(?, shape_offset_pre),
                               shape_offset_spread   = COALESCE(?, shape_offset_spread)
                         WHERE serial = ? AND timestamp = ?
                        """,
                        (
@@ -549,6 +569,10 @@ class SeismoDb:
                            rec.get("shape_near_peak_count") if rec else None,
                            rec.get("shape_sample_count")    if rec else None,
                            rec.get("shape_axis")            if rec else None,
                            rec.get("shape_offset")          if rec else None,
                            rec.get("shape_offset_axis")     if rec else None,
                            rec.get("shape_offset_pre")      if rec else None,
                            rec.get("shape_offset_spread")   if rec else None,
                            serial,
                            ts,
                        ),
@@ -603,61 +627,116 @@ class SeismoDb:
            ).fetchall()
        return [dict(r) for r in rows]
-    def find_twins(self, event_id: str, *, window_seconds: int = 300) -> list[dict]:
+    def find_twins(self, event_id: str, *, window_seconds: int | None = None) -> list[dict]:
        """
-        Find this event's histogram/waveform twin(s): rows sharing the same
+        Find this event's histogram/waveform twin(s): the SAME physical event
-        serial and an identical peak_vector_sum, whose timestamp falls
+        recorded both as a scheduled histogram and as a triggered waveform.
        within ``window_seconds`` of this event's timestamp.  Excludes the
        event itself.  Returns [] if the event or any required field
        (serial / peak_vector_sum / timestamp) is missing.
-        Caveat: identical-PVS matching is a proxy for "same physical event
+        A real trigger is captured twice — once as a triggered waveform (stamped
-        recorded twice," not a guarantee. In the rare case where the device
+        at the trigger instant) and once inside the scheduled histogram whose
-        clamps/saturates PVS (clamped to sqrt(3) * geo_range), two distinct
+        interval contains it (stamped at the histogram's interval start, e.g. the
-        saturated events on the same serial within the window can share the
+        7am/7pm call-in). The two can be HOURS apart in time yet report the same
-        same clamped PVS value and be matched as twins even though they are
+        serial and identical peak_vector_sum. Twins are therefore matched by:
-        different events. This is harmless in practice — false_trigger/
+
-        reviewed_real are derived/index columns re-derivable from the
+          * same serial,
-        sidecar source of truth — but worth knowing if twin counts look
+          * identical peak_vector_sum,
-        surprising on a saturated/clamped run.
+          * OPPOSITE record type (one histogram, one waveform), and
          * the waveform's timestamp falls within the histogram's interval —
            from a histogram's timestamp up to the next histogram (same serial).
        This replaces the old ±``window_seconds`` heuristic, which silently
        missed twins more than a few minutes apart (a histogram's interval-start
        stamp and the trigger instant routinely differ by hours). ``window_seconds``
        is still accepted for backward compatibility but is ignored.
        Returns [] if the event or a required field (serial / peak_vector_sum /
        timestamp) is missing.
        Caveat: identical-PVS matching remains a proxy for "same physical event"
        — if the device clamps/saturates PVS (to sqrt(3) * geo_range), two
        distinct saturated events could share a PVS. The added opposite-type and
        interval constraints make a false pairing far less likely than the old
        time-window match, and false_trigger/reviewed_real stay re-derivable from
        the sidecar source of truth.
        """
        def _parse(ts):
            if not ts:
                return None
            try:
                return datetime.datetime.fromisoformat(str(ts).replace(" ", "T"))
            except ValueError:
                return None
        def _is_hist(rt):
            return str(rt or "").lower().startswith("hist")
        row = self.get_event(event_id)
        if not row:
            return []
-        serial = row.get("serial"); pvs = row.get("peak_vector_sum"); ts = row.get("timestamp")
+        serial = row.get("serial"); pvs = row.get("peak_vector_sum")
-        if serial is None or pvs is None or not ts:
+        t_target = _parse(row.get("timestamp"))
        if serial is None or pvs is None or t_target is None:
            return []
-        try:
+        target_hist = _is_hist(row.get("record_type"))
            t = datetime.datetime.fromisoformat(ts.replace(" ", "T"))
        except ValueError:
            return []
        lo = (t - datetime.timedelta(seconds=window_seconds)).isoformat()
        hi = (t + datetime.timedelta(seconds=window_seconds)).isoformat()
        with self._connect() as conn:
            rows = conn.execute(
                "SELECT * FROM events WHERE serial=? AND id!=? AND peak_vector_sum=? "
                "AND timestamp BETWEEN ? AND ?",
                (serial, event_id, pvs, lo, hi),
            ).fetchall()
        return [dict(r) for r in rows]
-    def propagate_review_to_twins(self, event_id: str, *, window_seconds: int = 300) -> list[str]:
+        with self._connect() as conn:
            cand_rows = [dict(r) for r in conn.execute(
                "SELECT * FROM events WHERE serial=? AND id!=? AND peak_vector_sum=?",
                (serial, event_id, pvs)).fetchall()]
            hist_ts = [r["timestamp"] for r in conn.execute(
                "SELECT timestamp FROM events WHERE serial=? AND lower(record_type) LIKE 'hist%'",
                (serial,)).fetchall()]
        # Histogram interval-start times for this serial, sorted, to bound intervals.
        starts = sorted(x for x in (_parse(t) for t in hist_ts) if x is not None)
        def _interval_end(h_start):
            # The next histogram strictly after h_start bounds the interval; else open-ended.
            for x in starts:
                if x > h_start:
                    return x
            return None
        def _covers(h_start, w_time):
            end = _interval_end(h_start)
            return h_start <= w_time and (end is None or w_time < end)
        twins = []
        for c in cand_rows:
            if _is_hist(c.get("record_type")) == target_hist:
                continue  # twins are strictly cross-type (one histogram, one waveform)
            c_time = _parse(c.get("timestamp"))
            if c_time is None:
                continue
            h_start, w_time = (t_target, c_time) if target_hist else (c_time, t_target)
            if _covers(h_start, w_time):
                twins.append(c)
        return twins
    def propagate_review_to_twins(self, event_id: str, *, window_seconds: int | None = None) -> list[str]:
        """
-        Copy this event's `false_trigger`/`reviewed_real` columns onto each
+        Copy this event's `false_trigger`/`reviewed_real`/`false_trigger_reason`
-        of its histogram/waveform twins (see `find_twins`), so flagging one
+        columns onto each of its histogram/waveform twins (see `find_twins`), so
-        twin flags both.  Returns the list of twin ids updated.
+        flagging one twin flags both.  Returns the list of twin ids updated.
        ``window_seconds`` is accepted for backward compatibility but ignored;
        twin matching is now interval-based (see `find_twins`).
        """
        row = self.get_event(event_id)
        if not row:
            return []
        ft = 1 if row.get("false_trigger") else 0
        real = 1 if row.get("reviewed_real") else 0
-        twins = self.find_twins(event_id, window_seconds=window_seconds)
+        # The reason is a subtype of the FT flag — carry it only when the source
        # is actually a false trigger, so a confirmed-real twin never keeps one.
        reason = row.get("false_trigger_reason") if ft else None
        twins = self.find_twins(event_id)
        moved = []
        with self._connect() as conn:
            for tw in twins:
-                conn.execute("UPDATE events SET false_trigger=?, reviewed_real=? WHERE id=?",
+                conn.execute(
-                             (ft, real, tw["id"]))
+                    "UPDATE events SET false_trigger=?, reviewed_real=?, false_trigger_reason=? WHERE id=?",
                    (ft, real, reason, tw["id"]))
                moved.append(tw["id"])
        return moved
@@ -678,7 +757,7 @@ class SeismoDb:
                )
            else:
                cur = conn.execute(
-                    "UPDATE events SET false_trigger=0 WHERE id=?",
+                    "UPDATE events SET false_trigger=0, false_trigger_reason=NULL WHERE id=?",
                    (event_id,),
                )
        return cur.rowcount > 0
@@ -772,7 +851,8 @@ class SeismoDb:
            return False
        has_ft = "false_trigger" in review
        has_real = "reviewed_real" in review
-        if not has_ft and not has_real:
+        has_reason = "false_trigger_reason" in review
        if not has_ft and not has_real and not has_reason:
            # Nothing derived to update; just confirm the row exists.
            with self._connect() as conn:
                row = conn.execute(
@@ -785,11 +865,19 @@ class SeismoDb:
            sets["false_trigger"] = 1 if review.get("false_trigger") else 0
        if has_real:
            sets["reviewed_real"] = 1 if review.get("reviewed_real") else 0
        if has_reason:
            reason = review.get("false_trigger_reason") or None
            sets["false_trigger_reason"] = reason
            if reason:                      # a reason is a subtype of FT → implies FT
                sets["false_trigger"] = 1
        # mutual exclusivity: a true in one forces the other column to 0
        if sets.get("false_trigger") == 1:
            sets["reviewed_real"] = 0
        if sets.get("reviewed_real") == 1:
            sets["false_trigger"] = 0
        # the reason is only meaningful while flagged FT — clear it if FT ends up 0
        if sets.get("false_trigger") == 0:
            sets["false_trigger_reason"] = None
        assign = ", ".join(f"{k}=?" for k in sets)
        params = list(sets.values()) + [event_id]
        with self._connect() as conn:
@@ -121,6 +121,11 @@ class ReportData:
    t0_ms:                  Optional[float] = None
    dt_ms:                  Optional[float] = None
    # Sensor self-check traces — {ch: [samples]} in raw decode units, decoded
    # from the binary's trailing block (see minimateplus.sensor_check). The
    # little waveforms BW draws in its "Sensor Check" strip. Empty when absent.
    sensor_check_waveforms: dict = field(default_factory=dict)
    # Record-type discriminator
    record_type:            Optional[str] = None
    is_histogram:           bool = False
@@ -246,6 +251,8 @@ def gather_report_data(
                "peak_accel_g":       ch.get("peak_accel_g"),
                "peak_disp_in":       ch.get("peak_disp_in"),
                "sensor_check":       sc_ch.get("result"),
                "sc_freq_hz":         sc_ch.get("freq_hz"),
                "sc_ratio":           sc_ch.get("ratio"),
                "peak_date":          peak_date,
                "peak_time":          peak_time,
            })
@@ -290,6 +297,19 @@ def gather_report_data(
        except Exception as exc:
            log.warning("gather_report_data: hdf5 read failed: %s", exc)
    # ── Sensor self-check traces — decoded from the retained raw binary ──
    # The .h5 holds only the main waveform; the sensor-check traces live in the
    # binary's trailing block, so decode them straight from the kept BW file.
    # Waveform events only (histograms have no sensor-check strip).
    if not rd.is_histogram:
        try:
            from minimateplus.sensor_check import decode_sensor_check
            bw_path, _a5 = store.paths_for(serial, filename)
            if bw_path.exists():
                rd.sensor_check_waveforms = decode_sensor_check(bw_path.read_bytes())
        except Exception as exc:
            log.warning("gather_report_data: sensor-check decode failed: %s", exc)
    # ── Histogram aggregation ──
    # Codec emits ~N per-block samples (typically 1/sec); BW reports
    # one bar per configured interval (1 min / 5 min / etc.).  When
@@ -396,9 +416,34 @@ def _render_waveform_layout(fig, rd: ReportData) -> None:
    ax_stats = fig.add_subplot(gs[2]); ax_stats.axis("off")
    _draw_channel_stats_waveform(ax_stats, rd)
    _draw_compliance_panel(fig, rd)
    _draw_waveform_subplot(fig, gs[3], rd)
 # Compliance-chart placement, in figure fractions.  Measured directly off a
 # Blastware Event Report PDF (ref-stuff/n844lqhbzt0w_bw_pdf.pdf) so the chart
 # matches BW's size and position: it spans from just under the header down
 # through the stats band, hard against the right page margin.  The left edge
 # leaves room for the y-axis tick labels + "Velocity (in/s)" title, which the
 # compacted stats table (see _draw_channel_stats_waveform) is sized to clear.
 _COMPLIANCE_BOX = (0.489, 0.502, 0.951, 0.867)   # x0, y0, x1, y1
 def _draw_compliance_panel(fig, rd: ReportData) -> None:
    """Large USBM RI8507 compliance chart in the upper-right, sized and
    positioned to match Blastware's Event Report (see _COMPLIANCE_BOX)."""
    x0, y0, x1, y1 = _COMPLIANCE_BOX
    fig.text((x0 + x1) / 2, y1 + 0.006, "USBM RI8507 And OSMRE", fontsize=9,
             weight="bold", color="#333", ha="center", va="bottom")
    if rd.channels and rd.sample_rate_sps:
        from sfm.compliance import draw_compliance_chart
        ax = fig.add_axes([x0, y0, x1 - x0, y1 - y0])
        draw_compliance_chart(ax, rd.channels, rd.sample_rate_sps)
    else:
        fig.text((x0 + x1) / 2, (y0 + y1) / 2, "(no waveform data)", fontsize=8,
                 color="#bbb", ha="center", va="center", style="italic")
 def _render_histogram_layout(fig, rd: ReportData) -> None:
    """Histogram layout: header / mic-only / per-channel stats / bar plot.
@@ -477,11 +522,11 @@ def _split_iso_to_date_time(iso: Optional[str]) -> tuple[Optional[str], Optional
        return (None, None)
-def _kv(ax, x, y, label, value, *, label_w=0.18):
+def _kv(ax, x, y, label, value, *, label_w=0.18, fontsize=8):
    """Render a 'Label  Value' row at axes-coordinates (x, y)."""
-    ax.text(x, y, label, fontsize=8, color="#555", ha="left", va="top",
+    ax.text(x, y, label, fontsize=fontsize, color="#555", ha="left", va="top",
            transform=ax.transAxes)
-    ax.text(x + label_w, y, _fmt(value), fontsize=8, ha="left", va="top",
+    ax.text(x + label_w, y, _fmt(value), fontsize=fontsize, ha="left", va="top",
            transform=ax.transAxes, family="monospace")
@@ -544,14 +589,17 @@ def _draw_header_columns(ax, rows_left, rd: ReportData) -> None:
        ("File Name", rd.file_name),
        ("Post Event Notes", rd.post_event_notes),
    ]
    # fontsize 7.5 (BW's header is a touch smaller than our body text) + a
    # tighter right-column value indent so the long serial+firmware line
    # ("BE##### V ##.##-#.## MiniMate Plus") fits without running off the page.
    y = 0.95
    dy = 0.095
    for label, value in rows_left:
-        _kv(ax, 0.0, y, label, value, label_w=0.18)
+        _kv(ax, 0.0, y, label, value, label_w=0.18, fontsize=7.5)
        y -= dy
    y = 0.95
    for label, value in rows_right:
-        _kv(ax, 0.55, y, label, value, label_w=0.20)
+        _kv(ax, 0.55, y, label, value, label_w=0.14, fontsize=7.5)
        y -= dy
@@ -574,19 +622,14 @@ def _draw_mic_and_usbm(ax, rd: ReportData) -> None:
            transform=ax.transAxes, va="top")
    rows = _mic_rows(rd)
    y = 0.80
    # Tighter label indent + slightly smaller font so the long "Channel Test
    # Passed (Freq = … Amp = … mv)" line clears the enlarged compliance chart's
    # left edge (_COMPLIANCE_BOX) instead of running behind it.
    for label, value in rows:
-        _kv(ax, 0.0, y, label, value, label_w=0.18)
+        _kv(ax, 0.0, y, label, value, label_w=0.13, fontsize=7)
        y -= 0.15
-
+    # The USBM compliance chart is drawn as its own large square panel spanning
-    # USBM chart placeholder — upper-right.  Real piecewise compliance
+    # the mic + stats rows on the right — see _draw_compliance_panel().
    # curves are a separate work item; for now this just shows the title
    # + a "see report" message so the layout is correct.
    ax.text(0.72, 0.97, "USBM RI8507 And OSMRE",
            fontsize=9, weight="bold", color="#333", ha="center", va="top",
            transform=ax.transAxes)
    ax.text(0.72, 0.50, "[compliance chart\ncoming soon]",
            fontsize=8, color="#bbb", ha="center", va="center",
            transform=ax.transAxes, style="italic")
 def _mic_rows(rd: ReportData) -> list[tuple[str, Optional[str]]]:
@@ -636,8 +679,18 @@ def _draw_channel_stats_waveform(ax, rd: ReportData) -> None:
        ("Peak Acceleration",    "peak_accel_g",   "g"),
        ("Peak Displacement",    "peak_disp_in",   "in"),
        ("Sensor Check",         "sensor_check",   ""),
        # Sensor-check sub-rows (indented under "Sensor Check", like BW): the
        # geophone ring-down frequency + overswing ratio from the self-check.
        ("  Frequency",          "sc_freq_hz",     "Hz"),
        ("  Overswing Ratio",    "sc_ratio",       ""),
    ]
-    _draw_stats_table(ax, rd, rows_spec)
+    # Compacted to the left half so the enlarged compliance chart (BW-sized,
    # right against the page margin) has room — see _COMPLIANCE_BOX.
    _draw_stats_table(
        ax, rd, rows_spec,
        bbox_width=0.42, fontsize=7.5,
        col_widths=[0.185, 0.065, 0.065, 0.065, 0.040],
    )
    _draw_pvs_summary(ax, rd, n_data_rows=len(rows_spec))
@@ -698,19 +751,39 @@ def _draw_pvs_summary(
    table_bottom_y = getattr(ax, "_stats_table_bottom", -0.10)
    pvs_y = table_bottom_y - 0.04   # small gap below the table border
-    # Centered for visual balance — looks intentional rather than offset.
+    # Centered under the stats table for visual balance — looks intentional
-    # The original BW-replica had a "NA: Not Applicable" caption below
+    # rather than offset.  When the table is compacted (waveform layout), it
-    # this line; dropped because we use "—" for missing values and the
+    # occupies only the left portion of the axes, so center on the table's
-    # legend was always squished against the PVS line.
+    # width rather than the full axes (which would push the line under the
    # compliance chart).  The original BW-replica had a "NA: Not Applicable"
    # caption below this line; dropped because we use "—" for missing values.
    table_w = getattr(ax, "_stats_table_width", 0.80)
    if table_w < 0.79:
        # Compacted (waveform) layout: left-align under the table, one point
        # smaller, so the line clears the enlarged compliance chart's
        # bottom-left tick labels on the right.
        ax.text(0.0, pvs_y, line, fontsize=8, weight="bold",
                ha="left", va="top", transform=ax.transAxes)
    else:
        ax.text(0.5, pvs_y, line, fontsize=9, weight="bold",
                ha="center", va="top", transform=ax.transAxes)
-def _draw_stats_table(ax, rd: ReportData, rows_spec: list[tuple[str, str, str]]) -> None:
+def _draw_stats_table(
    ax, rd: ReportData, rows_spec: list[tuple[str, str, str]],
    *, bbox_width: float = 0.80, fontsize: float = 8,
    col_widths: Optional[list[float]] = None,
 ) -> None:
    """Render a per-channel stats table (Tran/Vert/Long).
    rows_spec: list of (label, field_name_in_channel_stats, unit_string)
    ``bbox_width`` / ``col_widths`` / ``fontsize`` let a caller compact the
    table (the waveform layout packs it into the left half to clear the
    compliance chart; the histogram layout keeps the wider defaults).
    """
    if col_widths is None:
        col_widths = [0.28, 0.14, 0.14, 0.14, 0.10]
    headers = ["", "Tran", "Vert", "Long", ""]
    ch_lookup = {c["name"]: c for c in rd.channel_stats}
@@ -726,6 +799,8 @@ def _draw_stats_table(ax, rd: ReportData, rows_spec: list[tuple[str, str, str]])
            if field == "zc_freq_hz":
                prefix = ">" if ch_rec.get("zc_freq_above_range") else ""
                return f"{prefix}{val:.0f}"
            if field in ("sc_freq_hz", "sc_ratio"):
                return f"{val:.1f}"       # BW shows 1 decimal (7.5 Hz, 3.6)
            return f"{val:.3f}"
        return str(val)
@@ -750,16 +825,17 @@ def _draw_stats_table(ax, rd: ReportData, rows_spec: list[tuple[str, str, str]])
    table_bottom = 1.0 - table_height
    tbl = ax.table(
        cellText=table_data,
-        colWidths=[0.28, 0.14, 0.14, 0.14, 0.10],
+        colWidths=col_widths,
        cellLoc="left", edges="open",
-        bbox=[0.0, table_bottom, 0.80, table_height],
+        bbox=[0.0, table_bottom, bbox_width, table_height],
    )
    tbl.auto_set_font_size(False)
-    tbl.set_fontsize(8)
+    tbl.set_fontsize(fontsize)
    for j in range(5):
        tbl[(0, j)].set_text_props(weight="bold", color="#555")
-    # Stash the bottom Y so _draw_pvs_summary can position itself below.
+    # Stash the bottom Y + width so _draw_pvs_summary can position itself.
    ax._stats_table_bottom = table_bottom
    ax._stats_table_width = bbox_width
 def _channel_axis_color(ch: str) -> str:
@@ -769,27 +845,59 @@ def _channel_axis_color(ch: str) -> str:
 def _draw_waveform_subplot(fig, gridspec_cell, rd: ReportData) -> None:
    """4-channel stacked waveform plot — Instantel printout order
    (MicL on top, Tran on bottom), shared x-axis in SECONDS, trigger
-    triangle markers at t=0, '0.0' baseline label on right of each."""
+    triangle markers at t=0, '0.0' baseline label on right of each.
-    inner = gridspec_cell.subgridspec(4, 1, hspace=0.0)
+
    When sensor self-check traces are present (rd.sensor_check_waveforms), a
    narrow "Sensor Check" strip of per-channel mini-plots is drawn to the right,
    aligned to the lanes — matching Blastware's Event Report.
    """
    from matplotlib.ticker import MaxNLocator
    order = ["MicL", "Long", "Vert", "Tran"]
    has_sc = bool(rd.sensor_check_waveforms)
    if has_sc:
        # main lanes + a narrow sensor-check strip column, flush against the
        # main panel (BW shares the border — no gap), with the "0.0" baseline
        # labels moved to the right of the strip.  Proportions match BW's
        # Event Report (main ~0.75 / strip ~0.10 of the panel width).
        inner = gridspec_cell.subgridspec(4, 2, width_ratios=[1.0, 0.13],
                                          wspace=0.0, hspace=0.0)
    else:
        inner = gridspec_cell.subgridspec(4, 1, hspace=0.0)
    sr = rd.sample_rate_sps or 1024
    # Convert ms-based time axis to seconds for the x-axis
    dt_s = (rd.dt_ms or (1000.0 / sr)) / 1000.0
    t0_s = (rd.t0_ms if rd.t0_ms is not None else 0.0) / 1000.0
    # Shared geo scale across Long/Vert/Tran (matches the event modal + BW's
    # single amp/div): all three geo lanes use ONE Y scale = the max |sample|
    # across them (padded, floored), so relative amplitudes stay honest instead
    # of each lane auto-zooming to its own peak.  Mic keeps its own (psi) scale.
    GEO_FLOOR_INS = 0.05
    _geo_amax = 0.0
    for _gch in ("Long", "Vert", "Tran"):
        for _x in (rd.channels.get(_gch) or []):
            _a = abs(_x)
            if _a > _geo_amax:
                _geo_amax = _a
    geo_shared = max(_geo_amax * 1.10, GEO_FLOOR_INS)
    main_axes = []
    sc_axes = []
    last_idx = len(order) - 1
    for i, ch in enumerate(order):
-        ax = fig.add_subplot(inner[i])
+        ax = fig.add_subplot(inner[i, 0] if has_sc else inner[i])
        main_axes.append(ax)
        values = rd.channels.get(ch) or []
        times = [t0_s + j * dt_s for j in range(len(values))]
        if values:
            color = _channel_axis_color(ch)
            ax.plot(times, values, color=color, linewidth=0.5)
-            # Symmetric y-axis for geo; zero-anchored for mic.
+            # Geo: one shared symmetric scale (honest relative amplitudes).
            # Mic: symmetric on its own psi scale (different unit).
            if ch != "MicL":
-                amax = max((abs(v) for v in values), default=0.001)
+                ax.set_ylim(-geo_shared, geo_shared)
                ax.set_ylim(-amax * 1.10, amax * 1.10)
            else:
                amax = max((abs(v) for v in values), default=0.001)
                ax.set_ylim(-amax * 1.10, amax * 1.10)
@@ -797,7 +905,10 @@ def _draw_waveform_subplot(fig, gridspec_cell, rd: ReportData) -> None:
        # Channel label on the LEFT (matches BW)
        ax.set_ylabel(ch, fontsize=8, rotation=0, ha="right", va="center",
                      color=_channel_axis_color(ch), weight="bold", labelpad=14)
-        # "0.0" on the RIGHT (BW convention)
+        # "0.0" baseline label on the RIGHT (BW convention).  With the sensor-
        # check strip attached, it goes to the right of the STRIP (drawn below);
        # otherwise just outside the main lane.
        if not has_sc:
            ax.text(1.005, 0.5, "0.0", transform=ax.transAxes,
                    fontsize=7, color="#555", va="center", ha="left")
@@ -814,23 +925,53 @@ def _draw_waveform_subplot(fig, gridspec_cell, rd: ReportData) -> None:
        else:
            ax.tick_params(axis="x", labelsize=7)
        ax.tick_params(axis="y", labelsize=6)
        # Stacked lanes touch, so the top/bottom y-tick labels of adjacent lanes
        # would overprint at the shared boundary.  Prune the extreme ticks so
        # each boundary shows clean interior ticks (0.5 / 0.0 / -0.5) only.
        ax.yaxis.set_major_locator(MaxNLocator(nbins=4, prune="both"))
        # Sensor self-check mini-plot in the right strip (aligned to this lane).
        if has_sc:
            scx = fig.add_subplot(inner[i, 1])
            sc_axes.append(scx)
            sc_vals = rd.sensor_check_waveforms.get(ch) or []
            if sc_vals:
                _col = _channel_axis_color(ch)
                # Faint zero baseline (BW draws the channel baseline through the
                # strip) — reference for the one-sided geophone ring-downs.
                scx.axhline(0.0, color=_col, linewidth=0.3, alpha=0.4)
                scx.plot(range(len(sc_vals)), sc_vals, color=_col, linewidth=0.5)
                # Fit the trace to the box (BW-style) rather than a symmetric
                # scale: the geo self-checks are one-sided dips, so a symmetric
                # scale would strand them in the bottom half with an empty top.
                _lo, _hi = min(sc_vals), max(sc_vals)
                _pad = 0.10 * ((_hi - _lo) or 1.0)
                scx.set_ylim(_lo - _pad, _hi + _pad)
            scx.set_xticks([]); scx.set_yticks([])
            for _s in scx.spines.values():
                _s.set_linewidth(0.4); _s.set_color("#999")
            # "0.0" baseline label to the RIGHT of the strip (BW convention)
            scx.text(1.10, 0.5, "0.0", transform=scx.transAxes,
                     fontsize=7, color="#555", va="center", ha="left")
    # Trigger triangle marker ▼ above the top channel at t=0
-    top_ax = fig.axes[-4]  # MicL is the first added in this gridspec
+    top_ax = main_axes[0]  # MicL
    top_ax.plot([0], [top_ax.get_ylim()[1]], marker="v", color="black",
                markersize=8, clip_on=False, zorder=10)
    # "Sensor Check" caption under the strip (BW convention)
    if has_sc and sc_axes:
        pos = sc_axes[-1].get_position()
        fig.text((pos.x0 + pos.x1) / 2, pos.y0 - 0.012, "Sensor Check",
                 fontsize=7, color="#555", ha="center", va="top")
    # Compute scale-per-division for the footer (10 divs across the chart)
    # and find peak geo amplitude for the geo amp/div setting.
    total_s = times[-1] - times[0] if values else 0
    div_s = total_s / 10 if total_s > 0 else 0
-    geo_amp_div = "—"
+    # Footer div value reflects the SHARED geo scale (so it's correct for all
-    for ch in ("Tran", "Vert", "Long"):
+    # three lanes, not just whichever one happened to be checked first).
-        v = rd.channels.get(ch) or []
+    geo_amp_div = f"{(geo_shared * 2) / 10:.3f}" if _geo_amax > 0 else "—"
        if v:
            amax = max(abs(x) for x in v)
            geo_amp_div = f"{(amax * 1.1 * 2) / 10:.3f}"
            break
    fig.text(
        0.11, 0.030,
        f"Time(Seconds) {div_s:.2f} sec/div   Amplitude Geo: {geo_amp_div} in/s/div   Mic: 0.001 psi(L)/div",
@@ -67,6 +67,7 @@ from minimateplus.blastware_file import write_blastware_file, blastware_filename
 from minimateplus.client import _decode_a5_metadata_into, _decode_a5_waveform, _decode_event_count
 from minimateplus.framing import build_bw_write_frame, SESSION_RESET, POLL_PROBE, POLL_DATA
 from minimateplus.protocol import SUB_STOP_MONITORING
 from minimateplus.event_file_io import TOOL_VERSION as SFM_VERSION  # single source for the service version (release-bumped)
 from sfm import event_hdf5
 from sfm.cache import SFMCache, get_cache
 from sfm.database import SeismoDb
@@ -90,7 +91,7 @@ app = FastAPI(
        "Implements the minimateplus RS-232 protocol library.\n"
        "Proxied by terra-view at /api/sfm/*."
    ),
-    version="0.26.0",
+    version=SFM_VERSION,
 )
 # Allow requests from the waveform viewer opened as a local file (file://)
@@ -371,7 +372,7 @@ def _backfill_events(events: list, info: "DeviceInfo") -> None:
@app.get("/health")
 def health() -> dict:
    """Service heartbeat.  No device I/O."""
-    return {"status": "ok", "service": "sfm", "version": "0.1.0"}
+    return {"status": "ok", "service": "sfm", "version": SFM_VERSION}
@app.get("/", response_class=FileResponse)
@@ -47,6 +47,60 @@ def shape_from_samples(chans: dict) -> dict | None:
    return s
 # ── Offset (DC-baseline) detection ────────────────────────────────────────────
 # A DC offset is a false trigger where the geophone baseline sits at a constant
 # non-zero floor (sensor bumped / settled / drifted) instead of oscillating
 # around zero. Brian's method (validated in scratch/offset_scan3.py): the
 # pre-trigger window is definitionally quiet, so a true offset shows |pre| off
 # zero AND stays flat across the record (pre ≈ mid ≈ end). A transient moves one
 # third relative to the others and is rejected by the spread test.
 # Thresholds are in in/s (the .h5 samples are already range-scaled); validated at
 # Normal range (10 in/s) — the only range in the fleet.
 OFFSET_FLOOR = 0.025        # |pre| at/above this reads as an off-zero baseline (5 A/D counts)
 OFFSET_MAX_SPREAD = 0.02    # max(pre,mid,end) - min(...) at/below this reads as flat/constant
 def _channel_offset(x, pretrig_n):
    """Return (pre, spread, is_offset) for one channel, or None if unusable."""
    x = np.asarray(x, dtype=float)
    n = x.size
    if n < 3:
        return None
    t = n // 3
    pre = x[:pretrig_n] if (pretrig_n and 0 < pretrig_n < n) else x[:t]
    mid, end = x[t:2 * t], x[2 * t:]
    if pre.size == 0 or mid.size == 0 or end.size == 0:
        return None
    vals = [float(np.median(seg)) for seg in (pre, mid, end)]
    spread = max(vals) - min(vals)
    is_offset = abs(vals[0]) >= OFFSET_FLOOR and spread <= OFFSET_MAX_SPREAD
    return vals[0], spread, is_offset
 def offset_from_samples(chans: dict, pretrig_n) -> dict | None:
    """Detect a DC-offset false trigger across the geophone channels.
    An event is offset if ANY geo channel's pre-trigger baseline is off zero and
    flat across the record. Reports the tripping axis (or, if none trips, the
    most-offset-like axis) with its ``pre``/``spread`` for transparency + tuning.
    Returns None when no geo channel is usable.
    """
    results = []
    for ax in _GEO_CHANNELS:
        x = chans.get(ax)
        if x is None:
            continue
        r = _channel_offset(x, pretrig_n)
        if r is not None:
            results.append((ax, r[0], r[1], r[2]))
    if not results:
        return None
    offenders = [r for r in results if r[3]]
    ax, pre, spread, _ = max(offenders or results, key=lambda r: abs(r[1]))
    return {"offset": bool(offenders), "axis": ax,
            "pre": round(pre, 6), "spread": round(spread, 6)}
 def shape_from_h5(path) -> dict | None:
    import h5py
    try:
@@ -56,3 +110,18 @@ def shape_from_h5(path) -> dict | None:
    except Exception:
        return None
    return shape_from_samples(chans)
 def offset_from_h5(path) -> dict | None:
    """offset_from_samples fed from an event's .h5 (float32 in/s geo samples +
    the pretrig_samples attribute)."""
    import h5py
    try:
        with h5py.File(path, "r") as f:
            chans = {ax: f[f"samples/{ax}"][:] for ax in _GEO_CHANNELS
                     if f"samples/{ax}" in f}
            pretrig_n = f.attrs.get("pretrig_samples")
    except Exception:
        return None
    pretrig_n = int(pretrig_n) if pretrig_n is not None else 0
    return offset_from_samples(chans, pretrig_n)
@@ -32,6 +32,7 @@ from __future__ import annotations
 import datetime
 import logging
 import pickle
 import re
 import shutil
 from pathlib import Path
 from typing import Optional, Union
@@ -41,7 +42,7 @@ from minimateplus.blastware_file import blastware_filename, write_blastware_file
 from minimateplus.framing import S3Frame
 from minimateplus.models import Event
 from sfm import event_hdf5
-from sfm.shape_metrics import shape_from_h5
+from sfm.shape_metrics import shape_from_h5, offset_from_h5
 log = logging.getLogger("sfm.waveform_store")
@@ -270,6 +271,13 @@ class WaveformStore:
            "shape_sample_count":    _shape["sample_count"],
            "shape_axis":            _shape["axis"],
        } if _shape else {}
        _offset = offset_from_h5(hdf5_path) if hdf5_filename else None
        _offset_rec = {
            "shape_offset":        1 if _offset["offset"] else 0,
            "shape_offset_axis":   _offset["axis"],
            "shape_offset_pre":    _offset["pre"],
            "shape_offset_spread": _offset["spread"],
        } if _offset else {}
        return {
            "filename":            filename,
            "filesize":            filesize,
@@ -278,6 +286,7 @@ class WaveformStore:
            "hdf5_filename":       hdf5_filename,
            "sidecar_filename":    sidecar_path.name,
            **_shape_rec,
            **_offset_rec,
        }
    def save_imported_bw(
@@ -371,8 +380,16 @@ class WaveformStore:
        # Resolve serial.  blastware_filename derives a 4-char prefix from
        # the numeric serial (e.g. BE11529 → M529); we go the other way
-        # via the source filename if a hint wasn't given.
+        # if a hint wasn't given.  The filename carries only the NUMBER,
-        serial = serial_hint or _serial_from_bw_filename(source_path.name) or "UNKNOWN"
+        # so read the family prefix out of the body first — a BlastMate
        # ("BA") filed as "BE" is a unit that does not exist.  The
        # filename-only decoder stays as the last resort.
        serial = (
            serial_hint
            or _serial_from_bw_bytes(bw_bytes, source_path.name)
            or _serial_from_bw_filename(source_path.name)
            or "UNKNOWN"
        )
        # Use the source filename verbatim — it already encodes timestamp
        # + record type per BW's AB0T scheme, and we want to preserve it
@@ -461,6 +478,13 @@ class WaveformStore:
            "shape_sample_count":    _shape["sample_count"],
            "shape_axis":            _shape["axis"],
        } if _shape else {}
        _offset = offset_from_h5(hdf5_path) if hdf5_filename else None
        _offset_rec = {
            "shape_offset":        1 if _offset["offset"] else 0,
            "shape_offset_axis":   _offset["axis"],
            "shape_offset_pre":    _offset["pre"],
            "shape_offset_spread": _offset["spread"],
        } if _offset else {}
        return ev, {
            "filename":           filename,
            "filesize":           filesize,
@@ -470,6 +494,7 @@ class WaveformStore:
            "sidecar_filename":   sidecar_path.name,
            "serial":             serial,
            **_shape_rec,
            **_offset_rec,
        }
    def save_imported_idf(
@@ -570,8 +595,19 @@ class WaveformStore:
                )
        # Binary-derived peaks fill in when the .txt didn't supply them.
-        # They're ~3% low vs the device-authoritative .txt values (residual
+        #
-        # codec drift), so .txt always wins when present.
+        # The old justification for this precedence -- "binary peaks are ~3%
        # low vs the .txt" -- was a decoder bug (geo LSB 0.0003 instead of
        # 0.000310308) and was fixed 2026-09-10; the binary now agrees with
        # Thor's own export per-sample.  The .txt still wins when present
        # because it is what the operator sees in Thor's report.
        #
        # ⚠ One case where the .txt is the *less* accurate of the two:
        # Thor floors displayed histogram PPV at 0.0050 in/s, so on quiet
        # IDFH events the .txt reports 0.0050 while the binary decodes the
        # true ~0.0025.  41.4% of prod IDFH sidecars carry a component PPV
        # larger than their own vector sum because of it.  Left as-is
        # deliberately, so stored peaks keep matching Thor's report.
        if binary_peaks is not None:
            if binary_peaks.transverse_ips and not report_dict.get("tran_ppv"):
                report_dict["tran_ppv"] = binary_peaks.transverse_ips
@@ -751,6 +787,13 @@ class WaveformStore:
            "shape_sample_count":    _shape["sample_count"],
            "shape_axis":            _shape["axis"],
        } if _shape else {}
        _offset = offset_from_h5(hdf5_path) if hdf5_filename else None
        _offset_rec = {
            "shape_offset":        1 if _offset["offset"] else 0,
            "shape_offset_axis":   _offset["axis"],
            "shape_offset_pre":    _offset["pre"],
            "shape_offset_spread": _offset["spread"],
        } if _offset else {}
        return ev, {
            "filename":           filename,
            "filesize":           filesize,
@@ -760,6 +803,7 @@ class WaveformStore:
            "sidecar_filename":   sidecar_path.name,
            "serial":             serial,
            **_shape_rec,
            **_offset_rec,
        }
    def load_a5(self, serial: str, filename: str) -> Optional[list[S3Frame]]:
@@ -816,20 +860,24 @@ class WaveformStore:
 # ── helpers ─────────────────────────────────────────────────────────────────────
-def _serial_from_bw_filename(name: str) -> Optional[str]:
+def _serial_number_from_bw_filename(name: str) -> Optional[int]:
    """
-    Reverse of `blastware_filename`'s serial-prefix encoding.
+    Reverse of `blastware_filename`'s serial-prefix encoding — the NUMBER only.
    BW filename format (V10.72): `<P><serial3><stem4>.<ext>`
    where P = chr(ord('B') + floor(serial // 1000))
    and serial3 = f"{serial % 1000:03d}".
    Examples (from CLAUDE.md verification archive):
-      P036... → BE14036       H907... → BE6907
+      P036... → 14036       H907... → 6907
-      M529... → BE11529       T003... → BE18003
+      M529... → 11529       T003... → 18003
      L895... → 10895
-    Returns the inferred BE-prefix serial (e.g. "BE11529") or None when
+    ⚠ The filename encodes **only the number**.  The two-letter family
-    the filename doesn't match the expected pattern.
+    prefix is NOT in it — "BE" is a MiniMate Plus, "BA" a BlastMate — so
    the prefix has to come from the file body (`_serial_from_bw_bytes`)
    or from an explicit hint.  Returns None when the filename doesn't
    match the expected pattern.
    """
    if not name:
        return None
@@ -842,5 +890,43 @@ def _serial_from_bw_filename(name: str) -> Optional[str]:
    if prefix_letter < "B":
        return None
    thousands = ord(prefix_letter) - ord("B")
-    serial_num = thousands * 1000 + int(base[1:4])
+    return thousands * 1000 + int(base[1:4])
-    return f"BE{serial_num}"
+
 _BW_SERIAL_RE = re.compile(rb"[A-Z]{2}\d{3,6}")
 def _serial_from_bw_bytes(data: bytes, name: str) -> Optional[str]:
    """
    Read the real serial — prefix included — out of a BW file body.
    The body carries the serial as a plain ASCII string ("BE9558",
    "BA10895").  We accept a candidate only when its numeric part matches
    the number the filename encodes, which keeps a stray byte sequence in
    the sample stream from being mistaken for a serial.
    Returns None when the filename number can't be derived or no
    candidate in the body agrees with it — the caller then falls back.
    """
    num = _serial_number_from_bw_filename(name)
    if num is None or not data:
        return None
    for match in _BW_SERIAL_RE.findall(data):
        candidate = match.decode("ascii", errors="replace")
        if candidate[2:].lstrip("0") == str(num):
            return candidate
    return None
 def _serial_from_bw_filename(name: str) -> Optional[str]:
    """
    Best-effort serial from the filename alone.
    ⚠ The family prefix is a **guess** — the filename does not carry it.
    "BE" is right for every MiniMate Plus but wrong for a BlastMate, whose
    serials start "BA".  Prefer `_serial_from_bw_bytes` whenever the file
    body is at hand; this exists for callers that only have a name
    (log lines, dry-run output).
    """
    num = _serial_number_from_bw_filename(name)
    return None if num is None else f"BE{num}"
@@ -0,0 +1,50 @@
 """Structural annotation of a Series-3 Blastware binary (for the seismo_lab
 Binary Inspector). The annotator maps byte ranges to labelled spans; anything
 the decoder can't account for is a first-class ``unknown`` span, so the whole
 file is tiled and the gaps (candidate FFT/spectral data) are visible.
 """
 from pathlib import Path
 from minimateplus.binary_annotate import annotate_blastware_binary, Span
 # A known-good full-3-channel Series-3 waveform binary (the V70 cracking fixture).
 FIXTURE = Path(__file__).parent / "fixtures" / "5-11-26" / "M529LL1L.V70"
 def _raw() -> bytes:
    return FIXTURE.read_bytes()
 def test_spans_tile_the_whole_file():
    raw = _raw()
    spans = annotate_blastware_binary(raw)
    assert spans, "expected at least one span"
    assert spans[0].start == 0
    assert spans[-1].end == len(raw)
    for a, b in zip(spans, spans[1:]):
        assert a.end == b.start, f"gap/overlap between {a!r} and {b!r}"
    for s in spans:
        assert s.start < s.end, f"empty/negative span {s!r}"
 def test_strt_record_is_located():
    raw = _raw()
    spans = annotate_blastware_binary(raw)
    strt = [s for s in spans if s.kind == "strt"]
    assert strt, "expected a STRT region"
    assert raw[strt[0].start : strt[0].start + 4] == b"STRT"
 def test_geo_sample_records_annotated():
    raw = _raw()
    spans = annotate_blastware_binary(raw)
    chans = {s.label.split()[0] for s in spans if s.kind == "sample"}
    # V70 is a full three-geo-channel event.
    assert {"Tran", "Vert", "Long"} <= chans, f"expected geo records, got {chans}"
 def test_footer_is_last():
    raw = _raw()
    spans = annotate_blastware_binary(raw)
    assert spans[-1].kind == "footer"
    assert spans[-1].end - spans[-1].start == 26
@@ -0,0 +1,35 @@
 """USBM/OSMRE compliance curve + scatter logic (sfm.compliance).
 Rendering is verified visually against Blastware reports."""
 import math
 import numpy as np
 from sfm.compliance import limit_at, channel_compliance_points
 def test_osmre_velocity_segments():
    assert abs(limit_at(6.0) - 0.75) < 1e-9      # 3.5–12 Hz flat
    assert abs(limit_at(50.0) - 2.00) < 1e-9     # 30–100 Hz flat
 def test_displacement_segments():
    assert abs(limit_at(2.0) - 2 * math.pi * 2.0 * 0.030) < 1e-9    # low-freq 0.030 in
    assert abs(limit_at(20.0) - 2 * math.pi * 20.0 * 0.008) < 1e-9  # rising diagonal 0.008 in
 def test_limit_clamps_below_1hz():
    assert limit_at(0.1) == limit_at(1.0)
 def test_scatter_ceiling_is_ppv_at_dominant_freq():
    # ~27 Hz blast-like trace whose energy peaks mid-record (inside full cycles,
    # as a real event does): the scatter cloud's ceiling is the trace PPV and the
    # top point sits near the dominant frequency.
    sps, n = 1024.0, 3328
    t = np.arange(n) / sps
    env = np.exp(-((t - 1.5) ** 2) / (2 * 0.3 ** 2))
    x = 0.9 * env * np.sin(2 * np.pi * 27.0 * t)
    f, v = channel_compliance_points(x, sps)
    assert len(f) > 20
    assert v.max() >= 0.99 * np.abs(x).max()
    assert 20.0 < f[int(np.argmax(v))] < 35.0
@@ -1,33 +1,71 @@
-import datetime
+import sqlite3
 from sfm.database import SeismoDb
 from minimateplus.models import Event, Timestamp
-def _ins(db, key, serial, pvs, ts):
+def _ins(db, key, serial, pvs, ts, record_type="Waveform"):
    ev = Event(index=0)
    ev._waveform_key = bytes.fromhex(key)
    ev.timestamp = ts
    # peak_vector_sum comes from peak_values; simplest: insert then UPDATE pvs directly
    db.insert_events([ev], serial=serial)
    row = [r for r in db.query_events(serial=serial) if r["waveform_key"] == key][0]
    import sqlite3
    with sqlite3.connect(db.db_path) as c:
-        c.execute("UPDATE events SET peak_vector_sum=? WHERE id=?", (pvs, row["id"]))
+        c.execute("UPDATE events SET peak_vector_sum=?, record_type=? WHERE id=?",
                  (pvs, record_type, row["id"]))
    return row["id"]
-def test_find_twins_matches_same_serial_pvs_near_time(tmp_path):
+def _ts(hour, minute, second=0, day=25):
    return Timestamp(raw=b"", flag=0x10, year=2026, unknown_byte=0, month=2, day=day,
                     hour=hour, minute=minute, second=second)
 def test_histogram_and_waveform_twin_across_hours(tmp_path):
    # The real UM12947 case: histogram stamped at its 7pm interval start, the
    # triggered waveform 75 min later — same serial + identical PVS. The old
    # ±5-min window missed this; interval matching catches it, both directions.
    db = SeismoDb(tmp_path / "s.db")
-    base = Timestamp(raw=b"", flag=0x10, year=2026, unknown_byte=0, month=2, day=25, hour=20, minute=19, second=5)
+    hist_pm = _ins(db, "01110001", "BE1", 0.4763, _ts(19, 31, 17), "Histogram")
-    twin = Timestamp(raw=b"", flag=0x10, year=2026, unknown_byte=0, month=2, day=25, hour=20, minute=19, second=45)
+    wave    = _ins(db, "01110002", "BE1", 0.4763, _ts(20, 46, 44), "Waveform")
-    far = Timestamp(raw=b"", flag=0x10, year=2026, unknown_byte=0, month=2, day=25, hour=21, minute=0, second=0)
+    _ins(db, "01110003", "BE1", 0.0100, _ts(7, 0, 0, day=26), "Histogram")  # bounds the interval
-    # d needs a timestamp distinct from `twin` (UNIQUE(serial, timestamp) would
+    assert {r["id"] for r in db.find_twins(hist_pm)} == {wave}
-    # otherwise collide with b and UPSERT onto its row instead of inserting a
+    assert {r["id"] for r in db.find_twins(wave)} == {hist_pm}
-    # new one) while staying near `base` in time.
+
-    near = Timestamp(raw=b"", flag=0x10, year=2026, unknown_byte=0, month=2, day=25, hour=20, minute=19, second=44)
+
-    a = _ins(db, "01110001", "BE1", 0.4763, base)
+def test_same_type_not_twinned(tmp_path):
-    b = _ins(db, "01110002", "BE1", 0.4763, twin)   # twin: same serial+pvs, 40s apart
+    # Two waveforms, same serial + PVS, seconds apart → NOT twins (cross-type only).
-    c = _ins(db, "01110003", "BE1", 0.4763, far)    # same pvs but >window away
+    db = SeismoDb(tmp_path / "s.db")
-    d = _ins(db, "01110004", "BE1", 0.9999, near)   # near time but different pvs
+    a = _ins(db, "01110001", "BE1", 0.4763, _ts(20, 19, 5), "Waveform")
-    ids = {r["id"] for r in db.find_twins(a, window_seconds=300)}
+    _ins(db, "01110002", "BE1", 0.4763, _ts(20, 19, 45), "Waveform")
-    assert ids == {b}
+    assert db.find_twins(a) == []
 def test_waveform_matches_only_the_containing_interval(tmp_path):
    # Two overnight intervals with the same PVS; a waveform in the SECOND interval
    # must twin with that histogram, never the first — even though PVS matches both.
    db = SeismoDb(tmp_path / "s.db")
    h1 = _ins(db, "01110001", "BE1", 0.4763, _ts(19, 0, 0, day=25), "Histogram")
    h2 = _ins(db, "01110002", "BE1", 0.4763, _ts(7, 0, 0, day=26), "Histogram")
    w = _ins(db, "01110003", "BE1", 0.4763, _ts(8, 0, 0, day=26), "Waveform")
    assert {r["id"] for r in db.find_twins(w)} == {h2}
    assert w not in {r["id"] for r in db.find_twins(h1)}
 def test_different_pvs_not_twinned(tmp_path):
    db = SeismoDb(tmp_path / "s.db")
    h = _ins(db, "01110001", "BE1", 0.4763, _ts(19, 0, 0), "Histogram")
    _ins(db, "01110002", "BE1", 0.9999, _ts(20, 0, 0), "Waveform")  # different PVS
    assert db.find_twins(h) == []
 def test_open_ended_latest_interval(tmp_path):
    # A waveform after the latest histogram (nothing bounds the interval) still twins.
    db = SeismoDb(tmp_path / "s.db")
    h = _ins(db, "01110001", "BE1", 0.4763, _ts(19, 0, 0), "Histogram")
    w = _ins(db, "01110002", "BE1", 0.4763, _ts(23, 30, 0), "Waveform")
    assert {r["id"] for r in db.find_twins(h)} == {w}
 def test_missing_fields_returns_empty(tmp_path):
    db = SeismoDb(tmp_path / "s.db")
    assert db.find_twins("nonexistent-id") == []
@@ -0,0 +1,103 @@
 import sqlite3
 from sfm.database import SeismoDb
 from minimateplus.models import Event, Timestamp
 def _ev(db, key="0111aaaa", serial="BE1"):
    ev = Event(index=0)
    ev._waveform_key = bytes.fromhex(key)
    ev.timestamp = Timestamp(raw=b"", flag=0x10, year=2026, unknown_byte=0,
                             month=6, day=25, hour=8, minute=0, second=0)
    ev.record_type = "Waveform"
    db.insert_events([ev], serial=serial)
    return [r for r in db.query_events(serial=serial) if r["waveform_key"] == key][0]["id"]
 def test_flag_offset_reason_implies_ft(tmp_path):
    db = SeismoDb(tmp_path / "s.db")
    eid = _ev(db)
    db.update_event_review(eid, {"false_trigger_reason": "offset"})
    row = db.get_event(eid)
    assert row["false_trigger"] == 1          # a reason is a subtype of FT
    assert row["false_trigger_reason"] == "offset"
    assert row["reviewed_real"] == 0
 def test_plain_ft_leaves_reason_null(tmp_path):
    # Reason is OPTIONAL — flagging FT without one records no reason.
    db = SeismoDb(tmp_path / "s.db")
    eid = _ev(db)
    db.update_event_review(eid, {"false_trigger": True})
    row = db.get_event(eid)
    assert row["false_trigger"] == 1
    assert row["false_trigger_reason"] is None
 def test_confirm_real_clears_reason(tmp_path):
    db = SeismoDb(tmp_path / "s.db")
    eid = _ev(db)
    db.update_event_review(eid, {"false_trigger_reason": "offset"})
    db.update_event_review(eid, {"reviewed_real": True})
    row = db.get_event(eid)
    assert row["reviewed_real"] == 1
    assert row["false_trigger"] == 0
    assert row["false_trigger_reason"] is None
 def test_clear_ft_clears_reason(tmp_path):
    db = SeismoDb(tmp_path / "s.db")
    eid = _ev(db)
    db.update_event_review(eid, {"false_trigger_reason": "offset"})
    db.update_event_review(eid, {"false_trigger": False})
    row = db.get_event(eid)
    assert row["false_trigger"] == 0
    assert row["false_trigger_reason"] is None
 def test_set_false_trigger_false_clears_reason(tmp_path):
    db = SeismoDb(tmp_path / "s.db")
    eid = _ev(db)
    db.update_event_review(eid, {"false_trigger_reason": "offset"})
    assert db.set_false_trigger(eid, False) is True
    row = db.get_event(eid)
    assert row["false_trigger"] == 0
    assert row["false_trigger_reason"] is None
 def test_reason_can_be_cleared_without_clearing_ft(tmp_path):
    # Setting reason to None removes the reason but leaves the FT flag intact.
    db = SeismoDb(tmp_path / "s.db")
    eid = _ev(db)
    db.update_event_review(eid, {"false_trigger_reason": "offset"})
    db.update_event_review(eid, {"false_trigger_reason": None})
    row = db.get_event(eid)
    assert row["false_trigger"] == 1
    assert row["false_trigger_reason"] is None
 def _ts(h, m, d=25):
    return Timestamp(raw=b"", flag=0x10, year=2026, unknown_byte=0,
                     month=2, day=d, hour=h, minute=m, second=0)
 def test_offset_reason_propagates_to_twin(tmp_path):
    # Flag a waveform as offset → its histogram twin also becomes FT with reason=offset.
    db = SeismoDb(tmp_path / "s.db")
    def ins(key, ts, rt):
        ev = Event(index=0); ev._waveform_key = bytes.fromhex(key); ev.timestamp = ts
        db.insert_events([ev], serial="BE1")
        rid = [r for r in db.query_events(serial="BE1") if r["waveform_key"] == key][0]["id"]
        with sqlite3.connect(db.db_path) as c:
            c.execute("UPDATE events SET peak_vector_sum=0.4763, record_type=? WHERE id=?", (rt, rid))
        return rid
    hist = ins("01110001", _ts(19, 31), "Histogram")   # interval start
    wave = ins("01110002", _ts(20, 46), "Waveform")     # trigger inside the interval
    db.update_event_review(wave, {"false_trigger_reason": "offset"})
    db.propagate_review_to_twins(wave)
    row = db.get_event(hist)
    assert row["false_trigger"] == 1
    assert row["false_trigger_reason"] == "offset"
@@ -0,0 +1,17 @@
 """The /health version must track the release, not a stale literal.
 terra-view's SFM Admin page displays whatever `/health` reports. It was
 hardcoded to "0.1.0" and never bumped, so the page showed 0.1.0 while the
 service was actually 0.26.0. These guard against that regression — and run
 without httpx (they call the endpoint function directly, no TestClient).
 """
 from minimateplus.event_file_io import TOOL_VERSION
 from sfm.server import app, health
 def test_health_reports_current_tool_version():
    assert health()["version"] == TOOL_VERSION
 def test_openapi_version_matches_tool_version():
    assert app.version == TOOL_VERSION
@@ -643,3 +643,38 @@ def test_multi_interval_matches_blastware_ascii_exactly():
                assert hz is None
            elif not cell.startswith("<"):
                assert hz is not None and abs(hz - float(cell)) <= max(0.55, float(cell) * 0.02)
 def test_partial_final_block_is_not_disqualified_by_missing_third_header():
    """A body can exceed two strides yet hold only two real blocks.
    Regression for BE18193 `T193L0XM.CI0H` — 51 intervals at 2 s = one full
    30-interval block plus a 21-interval remainder, in a body long enough to
    demand a third block header at ``2 * stride`` that does not exist.  The
    third-block confirmation used to be mandatory whenever the body was long
    enough, so the correct stride was discarded and the file decoded to
    nothing.  A missing third header means end-of-stream, not disqualification;
    the block-counter check is the decisive anti-false-positive test.
    """
    full = [(1, 1, 2, 2, 3, 3, 4, 4)] * 30
    partial = [(5, 5, 6, 6, 7, 7, 8, 8)] * 21
    body = (_mk_multi_block(full, ctr=256)
            + _mk_multi_block(partial, ctr=257)
            + b"\xff" * 700)      # trailing padding past 2 * stride
    stride = 12 + 20 * 30
    assert 2 * stride + 6 <= len(body), "padding must reach past two strides"
    # the whole point: a third header is absent, and that must not disqualify
    assert detect_multi_interval_stride(body) == stride
    recs = walk_multi_interval_blocks(body)
    assert len(recs) == 51
    assert recs[0]["t_peak"] == 1
    assert recs[-1]["t_peak"] == 5
 def test_third_block_still_rejects_a_mismatched_counter():
    """The corroboration must still bite when a third block IS present."""
    ivs = [(1, 1, 2, 2, 3, 3, 4, 4)] * 4
    body = (_mk_multi_block(ivs, ctr=256)
            + _mk_multi_block(ivs, ctr=257)
            + _mk_multi_block(ivs, ctr=999))   # counter jumps — not consecutive
    assert detect_multi_interval_stride(body) != 12 + 20 * 4
@@ -0,0 +1,322 @@
 """Per-sample verification of the Thor / Micromate (series-4) IDF binary codec.
 Ground truth is Thor's own CSV export, written next to each binary by the
 Thor desktop application.  For waveforms the export carries a per-sample
 block of four columns (Tran, Vert, Long, Mic) in in/s and psi -- the
 series-4 equivalent of Blastware's ``_ASCII.TXT`` exports.
 The full-corpus harness is ``scratch/verify_thor_against_csv.py``; these
 tests pin the two constants that harness established so they cannot
 regress silently.
 """
 from __future__ import annotations
 import csv
 import os
 import sys
 from pathlib import Path
 import pytest
 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
 from micromate.idf_file import (
    _GEO_LSB_IPS,
    geo_count_to_ips,
    read_idf_file,
 )
 FIXTURES = Path(__file__).parent / "fixtures" / "thor-idf"
 IDFW = FIXTURES / "UM11719_20231219162723.IDFW"
 IDFH = FIXTURES / "UM11719_20231219162648.IDFH"
 GEO_CHANNELS = ("Tran", "Vert", "Long")
 # tests/fixtures/ is gitignored, so a fresh checkout has no sample data.
 # Skip rather than fail, matching test_idf_ascii_report.py.  To populate:
 #
 #   B="<thor-watcher>/example-data/THORDATA_example/THORDATA_example/UPMC Presby"
 #   mkdir -p tests/fixtures/thor-idf
 #   for f in UM11719/UM11719_20231219162723.IDFW \
 #            UM11719/UM11719_20231219162648.IDFH \
 #            UM13981/UM13981_20220207084555.IDFW \
 #            UM13981/UM13981_20220207183102.IDFH \
 #            UM13981/UM13981_20221202063059.IDFH; do
 #     cp "$B/$f" tests/fixtures/thor-idf/
 #     cp "$B/$(dirname $f)/CSV/$(basename $f).csv" tests/fixtures/thor-idf/
 #   done
 pytestmark = pytest.mark.skipif(
    not FIXTURES.is_dir() or not any(FIXTURES.glob("*.IDFW")),
    reason=f"Thor IDF fixtures not present under {FIXTURES}",
 )
 def _parse_export(path: Path):
    """Split a Thor CSV export into (header dict, per-sample rows)."""
    header, rows = {}, []
    with path.open(newline="", encoding="utf-8", errors="replace") as fh:
        for rec in csv.reader(fh):
            if len(rec) == 2:
                header[rec[0].strip()] = rec[1].strip()
            elif len(rec) >= 3:
                try:
                    rows.append([float(x) for x in rec])
                except ValueError:
                    pass
    return header, rows
 def _header_float(header, key):
    return float(header[key].split()[0])
@pytest.fixture(scope="module")
 def idfw_export():
    return _parse_export(IDFW.with_suffix(".IDFW.csv"))
 # ─── The geo scale constant ────────────────────────────────────────────────
 def test_geo_lsb_matches_thor_quantisation():
    """Thor's own export quantises geo samples to this LSB.
    Derived by maximising exact-match count over 1,046,016 paired samples
    (454 channel-events, 2 units); independently corroborated on 8
    production units via their device-reported PPV.  The historical value
    0.0003 read every series-4 geophone sample 3.3% low.
    """
    assert _GEO_LSB_IPS == pytest.approx(0.000310308, rel=1e-6)
 def test_geo_lsb_is_not_the_legacy_value():
    # Guards against a revert to the truncated 0.0003 constant.
    assert abs(_GEO_LSB_IPS - 0.0003) > 1e-6
 # ─── Per-sample fidelity ───────────────────────────────────────────────────
 def test_waveform_channel_lengths_match_export(idfw_export):
    _header, rows = idfw_export
    result = read_idf_file(IDFW)
    for channel in GEO_CHANNELS:
        assert len(result.samples[channel]) == len(rows), (
            f"{channel} truncated: decoded {len(result.samples[channel])} "
            f"samples, export has {len(rows)}"
        )
 def test_waveform_samples_match_export_exactly(idfw_export):
    """Every geo sample must reproduce Thor's exported value to 4 dp."""
    _header, rows = idfw_export
    result = read_idf_file(IDFW)
    for index, channel in enumerate(GEO_CHANNELS):
        decoded = result.samples[channel]
        expected = [row[index] for row in rows]
        mismatches = [
            (i, geo_count_to_ips(c), v)
            for i, (c, v) in enumerate(zip(decoded, expected))
            if abs(geo_count_to_ips(c) - v) >= 5e-5
        ]
        assert not mismatches, (
            f"{channel}: {len(mismatches)} of {len(expected)} samples differ; "
            f"first three {mismatches[:3]}"
        )
 def test_waveform_ppv_matches_export(idfw_export):
    header, _rows = idfw_export
    result = read_idf_file(IDFW)
    for channel, attr in (
        ("Tran", "transverse_ips"),
        ("Vert", "vertical_ips"),
        ("Long", "longitudinal_ips"),
    ):
        decoded = getattr(result.event.peaks, attr)
        assert decoded == pytest.approx(
            _header_float(header, f"{channel}PPV"), abs=5e-5
        ), f"{channel} PPV disagrees with Thor's export"
 # ─── Histogram path shares the same scale ──────────────────────────────────
 def test_histogram_peaks_match_export():
    header, _rows = _parse_export(IDFH.with_suffix(".IDFH.csv"))
    result = read_idf_file(IDFH)
    assert result.intervals, "IDFH decoded no intervals"
    for channel, attr in (
        ("Tran", "transverse_ips"),
        ("Vert", "vertical_ips"),
        ("Long", "longitudinal_ips"),
    ):
        decoded = getattr(result.event.peaks, attr)
        expected = _header_float(header, f"{channel}PPV")
        # Histogram peaks are stored per-interval, so the export's PPV is
        # reproduced within one quantisation step rather than exactly.
        assert decoded == pytest.approx(expected, abs=2 * _GEO_LSB_IPS), (
            f"{channel} histogram peak {decoded} vs export {expected}"
        )
 # ─── Regressions found 2026-09-10 ──────────────────────────────────────────
 IDFH_LONG = FIXTURES / "UM13981_20220207183102.IDFH"      # 719 intervals
 IDFH_SENTINEL = FIXTURES / "UM13981_20221202063059.IDFH"  # holds an unwritten slot
 IDFW_RAW16 = FIXTURES / "UM13981_20220207084555.IDFW"     # segment 0 is MODE_RAW16
 def test_histogram_decodes_past_250_intervals():
    """The segment validator must not require a zero counter high byte.
    The interval counter is a uint16 cumulative index.  Requiring its high
    byte to be zero rejected every segment past interval 255, capping each
    histogram at 250 intervals and truncating any run longer than ~4 hours —
    frequently discarding the part that held the peak.
    """
    result = read_idf_file(IDFH_LONG)
    header, _rows = _parse_export(IDFH_LONG.with_suffix(".IDFH.csv"))
    expected = float(header["NumberOfIntervals"])
    assert len(result.intervals) == 719
    assert len(result.intervals) == pytest.approx(expected, abs=1.0)
 def test_histogram_ignores_unwritten_interval_slot():
    """A never-written interval keeps its ±full-scale seed and must be dropped.
    Counting it fabricates a 10.0 in/s peak on every channel, which then wins
    the max-over-intervals and poisons the whole file's PPV.
    """
    header, _rows = _parse_export(IDFH_SENTINEL.with_suffix(".IDFH.csv"))
    result = read_idf_file(IDFH_SENTINEL)
    for channel, attr in (
        ("Tran", "transverse_ips"),
        ("Vert", "vertical_ips"),
        ("Long", "longitudinal_ips"),
    ):
        decoded = getattr(result.event.peaks, attr)
        assert decoded < 1.0, f"{channel} peak {decoded} looks like the ±FS seed"
        assert decoded == pytest.approx(
            _header_float(header, f"{channel}PPV"), abs=2 * _GEO_LSB_IPS
        )
 def test_waveform_raw16_segment_zero_is_decoded():
    """Segment-0 records can be raw int16 (MODE_RAW16, 10-byte header).
    That mode was absent from the dispatch, so the record fell through
    unhandled and the channel silently lost its first 512 samples.
    """
    rows = _parse_export(IDFW_RAW16.with_suffix(".IDFW.csv"))[1]
    result = read_idf_file(IDFW_RAW16)
    for index, channel in enumerate(GEO_CHANNELS):
        decoded = result.samples[channel]
        assert len(decoded) == len(rows), f"{channel} lost segment 0"
        expected = [row[index] for row in rows]
        bad = sum(
            1 for c, v in zip(decoded, expected)
            if abs(geo_count_to_ips(c) - v) >= 5e-5
        )
        assert bad == 0, f"{channel}: {bad} samples differ from Thor's export"
 def test_body_offset_search_is_not_quadratic():
    """The body scan must stay cheap enough for bulk ingest.
    MODE_RAW16 is (0x00, 0x00), so scanning for candidate *preambles* treats
    every run of three zero bytes as a body start and trial-decodes each one
    (~0.5 s/file measured).  The search anchors on record headers instead.
    """
    import time
    start = time.perf_counter()
    for _ in range(3):
        read_idf_file(IDFW_RAW16)
    elapsed = (time.perf_counter() - start) / 3
    assert elapsed < 0.15, f"body-offset search took {elapsed*1000:.0f} ms/file"
 # ─── Mic-disabled (3-channel) units, found 2026-09-10 ──────────────────────
 IDFW_3CH = FIXTURES / "UM20147_20250531135901.IDFW"   # body head below old floor
 IDFH_3CH = FIXTURES / "UM20147_20250330070110.IDFH"   # 56-byte interval records
 def test_three_channel_waveform_decodes_all_geo_channels():
    """A mic-disabled unit's shorter header moves the record chain head.
    Its head sits at 0x0dba, below the old ``_BODY_SCAN_FLOOR`` of 0x0E00, so
    the scan could not see it and fell through to the *Vert* segment-0 record
    — decoding a body shifted one position around the channel rotation, which
    surfaced as Vert being exactly 512 samples short.
    """
    rows = _parse_export(IDFW_3CH.with_suffix(".IDFW.csv"))[1]
    result = read_idf_file(IDFW_3CH)
    for index, channel in enumerate(GEO_CHANNELS):
        decoded = result.samples[channel]
        assert len(decoded) == len(rows), (
            f"{channel}: {len(decoded)} samples, export has {len(rows)}"
        )
        expected = [row[index] for row in rows]
        bad = sum(
            1 for c, v in zip(decoded, expected)
            if abs(geo_count_to_ips(c) - v) >= 5e-5
        )
        assert bad == 0, f"{channel}: {bad} samples differ from Thor's export"
    # Mic is genuinely absent on these units, not merely undecoded.
    assert not result.samples.get("MicL")
 def test_three_channel_histogram_uses_56_byte_intervals():
    """Interval stride is 16 bytes per channel + an 8-byte tail, not a constant.
    A mic-disabled unit packs 56-byte records, so assuming 72 read 7 intervals
    out of every 10-interval segment and then walked off alignment into
    garbage, which decoded as ~10 in/s peaks.  The true count comes from the
    segment's cumulative interval counter.
    """
    header, _rows = _parse_export(IDFH_3CH.with_suffix(".IDFH.csv"))
    result = read_idf_file(IDFH_3CH)
    expected_intervals = float(header["NumberOfIntervals"])
    assert len(result.intervals) == pytest.approx(expected_intervals, abs=1.0)
    assert {iv.n_channels for iv in result.intervals} == {3}
    for channel, attr in (
        ("Tran", "transverse_ips"),
        ("Vert", "vertical_ips"),
        ("Long", "longitudinal_ips"),
    ):
        decoded = getattr(result.event.peaks, attr)
        assert decoded < 1.0, f"{channel} peak {decoded} looks like walked-off garbage"
        assert decoded == pytest.approx(
            _header_float(header, f"{channel}PPV"), rel=0.02
        )
 # ─── `40 NN` blocks with NN > 8, verified 2026-09-11 ───────────────────────
 IDFW_WIDE40 = FIXTURES / "UM12947_20250806134504.IDFW"
 def test_wide_forty_nn_block_does_not_truncate_channels():
    """Loud events use `40 NN` blocks with NN well above the old cap of 8.
    ``data_block_len()`` rejected NN > 0x08, which halted the block walk
    part-way through a record.  The walker stops at the first unrecognised
    tag instead of raising, so this surfaced as silently short channels —
    here Tran 1812 / Vert 2132 / Long 2324 where the export has 2324 for all
    three.  The affected files use NN of 12, 16, 20 ... up to 196.
    """
    rows = _parse_export(IDFW_WIDE40.with_suffix(".IDFW.csv"))[1]
    result = read_idf_file(IDFW_WIDE40)
    for index, channel in enumerate(GEO_CHANNELS):
        decoded = result.samples[channel]
        assert len(decoded) == len(rows), (
            f"{channel}: {len(decoded)} samples, export has {len(rows)}"
        )
        expected = [row[index] for row in rows]
        bad = sum(
            1 for c, v in zip(decoded, expected)
            if abs(geo_count_to_ips(c) - v) >= 5e-5
        )
        assert bad == 0, f"{channel}: {bad} samples differ from Thor's export"
@@ -0,0 +1,94 @@
 import numpy as np
 import h5py
 from sfm.shape_metrics import offset_from_samples, offset_from_h5
 def test_flags_constant_dc_floor():
    # A geophone channel sitting at a constant +0.05 in/s across the whole record
    # is a DC offset: baseline off zero AND flat across pre/mid/end thirds.
    n = 300
    chans = {"Tran": np.full(n, 0.05), "Vert": np.zeros(n), "Long": np.zeros(n)}
    r = offset_from_samples(chans, pretrig_n=50)
    assert r["offset"] is True
    assert r["axis"] == "Tran"
    assert abs(r["pre"] - 0.05) < 1e-6
    assert r["spread"] < 0.02
 def test_transient_rejected_by_spread():
    # Off-zero pre-trigger but the baseline SETTLES back over the record — a
    # transient, not a constant offset. The spread test must reject it.
    x = np.concatenate([np.full(100, 0.05), np.full(100, 0.025), np.zeros(100)])
    chans = {"Tran": x, "Vert": np.zeros(300), "Long": np.zeros(300)}
    r = offset_from_samples(chans, pretrig_n=100)
    assert r["offset"] is False
 def test_clean_oscillation_not_offset():
    t = np.arange(300)
    x = 0.4 * np.sin(2 * np.pi * t / 20)   # oscillates around zero — baseline IS zero
    chans = {"Tran": x, "Vert": np.zeros(300), "Long": np.zeros(300)}
    r = offset_from_samples(chans, pretrig_n=50)
    assert r["offset"] is False
 def test_below_floor_not_offset_but_reports_pre():
    # A flat baseline below the floor is not an offset; still report the axis/pre
    # for tuning transparency.
    n = 300
    chans = {"Tran": np.full(n, 0.01), "Vert": np.zeros(n), "Long": np.zeros(n)}
    r = offset_from_samples(chans, pretrig_n=50)
    assert r["offset"] is False
    assert r["axis"] == "Tran"
    assert abs(r["pre"] - 0.01) < 1e-6
 def test_none_when_no_geo_channels():
    assert offset_from_samples({"MicL": np.full(300, 0.05)}, pretrig_n=50) is None
 def test_pretrig_fallback_when_invalid():
    # pretrig_n of 0 (missing/unusable) falls back to the first third.
    n = 300
    chans = {"Tran": np.full(n, 0.05), "Vert": np.zeros(n), "Long": np.zeros(n)}
    r = offset_from_samples(chans, pretrig_n=0)
    assert r["offset"] is True
 def test_flags_offset_on_any_axis():
    # Offset on Vert alone still flags the event, and Vert is reported.
    n = 300
    chans = {"Tran": np.zeros(n), "Vert": np.full(n, -0.06), "Long": np.zeros(n)}
    r = offset_from_samples(chans, pretrig_n=50)
    assert r["offset"] is True
    assert r["axis"] == "Vert"
 def _write_h5(path, chans, pretrig_n):
    with h5py.File(path, "w") as f:
        g = f.create_group("samples")
        for k, v in chans.items():
            g.create_dataset(k, data=np.asarray(v, dtype="float32"))
        if pretrig_n is not None:
            f.attrs["pretrig_samples"] = pretrig_n
 def test_offset_from_h5_reads_pretrig_attr(tmp_path):
    p = tmp_path / "ev.h5"
    n = 300
    _write_h5(p, {"Tran": np.full(n, 0.05), "Vert": np.zeros(n), "Long": np.zeros(n)},
              pretrig_n=50)
    r = offset_from_h5(str(p))
    assert r["offset"] is True and r["axis"] == "Tran"
 def test_offset_from_h5_missing_pretrig_attr_falls_back(tmp_path):
    p = tmp_path / "noattr.h5"
    n = 300
    _write_h5(p, {"Tran": np.full(n, 0.05), "Vert": np.zeros(n), "Long": np.zeros(n)},
              pretrig_n=None)
    assert offset_from_h5(str(p))["offset"] is True   # falls back to first-third
 def test_offset_from_h5_missing_file_is_none(tmp_path):
    assert offset_from_h5(str(tmp_path / "nope.h5")) is None
@@ -0,0 +1,64 @@
 from __future__ import annotations
 from pathlib import Path
 import numpy as np, h5py
 from sfm.database import SeismoDb
 from sfm.waveform_store import WaveformStore
 from scripts.backfill_event_shape import backfill_shape
 from minimateplus.models import Event, Timestamp, PeakValues
 _FIX = Path(__file__).parent / "fixtures/histogram-extension-re/events-5-21-26/K558LL8B.7I0W"
 def _event(waveform_key="0111abcd"):
    ev = Event(index=0)
    ev._waveform_key = bytes.fromhex(waveform_key)
    ev.timestamp = Timestamp(raw=b"", flag=0x10, year=2026, unknown_byte=0,
                             month=6, day=25, hour=8, minute=50, second=0)
    ev.record_type = "Waveform"
    ev.peak_values = PeakValues(tran=0.075, vert=0.220, long=0.045,
                                 peak_vector_sum=0.231, micl=0.01)
    return ev
 def test_insert_stores_offset_from_record(tmp_path: Path):
    db = SeismoDb(tmp_path / "s.db")
    ev = _event()
    rec = {ev._waveform_key.hex(): {
        "filename": "F.CE0W", "filesize": 10,
        "shape_offset": 1, "shape_offset_axis": "Tran",
        "shape_offset_pre": 0.05, "shape_offset_spread": 0.001}}
    db.insert_events([ev], serial="BE1", waveform_records=rec)
    row = db.query_events(serial="BE1")[0]
    assert row["shape_offset"] == 1
    assert row["shape_offset_axis"] == "Tran"
    assert abs(row["shape_offset_pre"] - 0.05) < 1e-6
    assert abs(row["shape_offset_spread"] - 0.001) < 1e-6
 def test_save_imported_bw_attaches_offset(tmp_path: Path):
    store = WaveformStore(tmp_path / "waveforms")
    ev, rec = store.save_imported_bw(_FIX.read_bytes(), source_path=_FIX, serial_hint="BE9558")
    assert rec["shape_offset"] in (0, 1)
    assert rec["shape_offset_axis"] in ("Tran", "Vert", "Long")
    assert "shape_offset_pre" in rec and "shape_offset_spread" in rec
 def test_backfill_updates_offset(tmp_path: Path):
    db = SeismoDb(tmp_path / "s.db")
    store = WaveformStore(tmp_path / "waveforms")
    ev = Event(index=0); ev._waveform_key = bytes.fromhex("0111abcd")
    db.insert_events([ev], serial="BE1",
                     waveform_records={ev._waveform_key.hex(): {"filename": "F.CE0W", "filesize": 10}})
    p = store.hdf5_path_for("BE1", "F.CE0W")
    with h5py.File(p, "w") as f:
        g = f.create_group("samples")
        g.create_dataset("Tran", data=np.full(300, 0.05, "float32"))
        g.create_dataset("Vert", data=np.zeros(300, "float32"))
        g.create_dataset("Long", data=np.zeros(300, "float32"))
        f.attrs["pretrig_samples"] = 50
    backfill_shape(db, store)
    row = db.query_events(serial="BE1")[0]
    assert row["shape_offset"] == 1
    assert row["shape_offset_axis"] == "Tran"
@@ -0,0 +1,61 @@
 """The event-report PDF must draw the three geo channels on ONE shared Y scale
 (max |sample| across Long/Vert/Tran, floored), not each trace auto-zoomed to its
 own peak — so relative amplitudes are honest and a small channel doesn't fill its
 lane looking as big as a large one.  Mirrors the event-modal waveform behaviour.
 """
 import matplotlib
 matplotlib.use("Agg")
 import matplotlib.pyplot as plt
 import pytest
 from sfm.report_pdf import ReportData, _draw_waveform_subplot
 def _draw(channels):
    rd = ReportData(
        channels=channels,
        sample_rate_sps=1024,
        dt_ms=1000.0 / 1024,
        t0_ms=0.0,
    )
    fig = plt.figure()
    cell = fig.add_gridspec(1, 1)[0, 0]
    _draw_waveform_subplot(fig, cell, rd)
    by_label = {ax.get_ylabel(): ax for ax in fig.axes}
    try:
        yield_ = {k: by_label[k].get_ylim() for k in ("Long", "Vert", "Tran", "MicL")}
    finally:
        plt.close(fig)
    return yield_
 def test_geo_traces_share_one_y_scale():
    # Tran is the biggest geo channel (0.35); Long 0.10, Vert 0.02.
    ylims = _draw({
        "Long": [0.10, -0.10, 0.0],
        "Vert": [0.02, -0.02, 0.0],
        "Tran": [0.35, -0.35, 0.0],
        "MicL": [0.0005, -0.0005, 0.0],
    })
    # Shared scale = max(0.35 * 1.10, floor 0.05) = 0.385, symmetric.
    expected = pytest.approx(0.385, rel=1e-6)
    for ch in ("Long", "Vert", "Tran"):
        lo, hi = ylims[ch]
        assert hi == expected, f"{ch} top ylim {hi} != shared 0.385"
        assert lo == pytest.approx(-0.385, rel=1e-6), f"{ch} bottom ylim {lo}"
    # All three geo lanes identical.
    assert ylims["Long"] == ylims["Vert"] == ylims["Tran"]
    # Mic keeps its own (much smaller) scale — not lumped into the geo max.
    assert ylims["MicL"][1] < 0.01
 def test_geo_shared_scale_has_floor():
    # A tiny event (all geo well under the floor) clamps to the 0.05 floor.
    ylims = _draw({
        "Long": [0.008, -0.008, 0.0],
        "Vert": [0.006, -0.006, 0.0],
        "Tran": [0.010, -0.010, 0.0],
        "MicL": [0.0001, -0.0001, 0.0],
    })
    for ch in ("Long", "Vert", "Tran"):
        assert ylims[ch][1] == pytest.approx(0.05, rel=1e-6), f"{ch} not floored"
@@ -0,0 +1,67 @@
 """Blastware sensor self-check waveform decode (minimateplus.sensor_check).
 Reverse-engineered 2026-09-15 against 7 BE12844 (MiniMate Plus) oracle events.
 After the main waveform record-chain and the trailing metadata / per-channel
 calibration records, a series-3 binary carries four length-prefixed records
 tagged 0x3c-0x3f: the sensor self-check traces the unit records when it pulses
 each sensor before monitoring (Blastware draws these as the little waveforms in
 the "Sensor Check" strip on the right of the Event Report).
  * 0x3c / 0x3d / 0x3e = Tran / Vert / Long geophone ring-downs.
  * 0x3f            = MicL, a pulse train at the mic self-test frequency.
 The self-check injects a fixed pulse, so the response is near-identical across
 events — asserted here as an invariant shape (damped one-sided ring-down for
 the geophones, a multi-pulse train for the mic).
 """
 from pathlib import Path
 import numpy as np
 from minimateplus.sensor_check import decode_sensor_check
 FIXDIR = Path(__file__).parent / "fixtures" / "fft-oracle-2026-09-14"
 EVENTS = sorted(p.name for p in FIXDIR.iterdir())   # 7 BE12844 event binaries
 def _decode(name):
    return decode_sensor_check((FIXDIR / name).read_bytes())
 def test_all_four_channels_present():
    for name in EVENTS:
        sc = _decode(name)
        assert set(sc) == {"Tran", "Vert", "Long", "MicL"}, name
 def test_geo_channels_are_damped_ringdowns():
    # Each geophone self-check is a large one-sided deflection (~-990 raw) that
    # rings back and damps toward a settled value well above the trough.
    for name in EVENTS:
        sc = _decode(name)
        for ch in ("Tran", "Vert", "Long"):
            tr = np.asarray(sc[ch], dtype=float)
            assert 240 <= len(tr) <= 260, f"{name}:{ch} n={len(tr)}"
            assert abs(tr[:3].mean()) < 50, f"{name}:{ch} starts off-baseline"
            assert tr.min() < -800, f"{name}:{ch} min {tr.min()}"
            assert tr.max() < 60, f"{name}:{ch} unexpected positive swing {tr.max()}"
            # damped: settles between the trough and zero, well above the trough
            assert tr.min() < tr[-1] < 0, f"{name}:{ch} end {tr[-1]} not between trough and 0"
            assert abs(tr[-1]) < 0.6 * abs(tr.min()), f"{name}:{ch} not damped, end {tr[-1]}"
 def test_mic_channel_is_a_pulse_train():
    for name in EVENTS:
        tr = np.asarray(_decode(name)["MicL"], dtype=float)
        assert 235 <= len(tr) <= 255, f"{name} mic n={len(tr)}"
        # larger dynamic range than the geo ring-down, and swings both ways
        assert tr.min() < -1500, f"{name} mic min {tr.min()}"
        assert tr.max() > 100, f"{name} mic max {tr.max()}"
        # multiple pulses: several deep local minima
        deep = (tr[1:-1] < tr[:-2]) & (tr[1:-1] < tr[2:]) & (tr[1:-1] < -800)
        assert int(deep.sum()) >= 4, f"{name} mic pulses {int(deep.sum())}"
 def test_returns_empty_when_no_sensor_check_block():
    assert decode_sensor_check(b"not a blastware file") == {}
    assert decode_sensor_check(b"") == {}
@@ -0,0 +1,101 @@
 """The BW filename encodes the serial NUMBER, never the family prefix.
 "BE" is a MiniMate Plus; "BA" is a BlastMate.  Both are Series III and their
 files are byte-compatible — the whole archive's 1,493 BlastMate binaries
 decode through the same codec at 100% — so the only thing that distinguishes
 them downstream is the serial string, and that lives in the file body.
 Synthesising the prefix as "BE" files a BlastMate under a unit that does not
 exist.  Four units in the DL2 archive are affected: BA9229, BA10060, BA10895
 and BA15957.
 """
 from __future__ import annotations
 import pytest
 from minimateplus.client import _decode_0a_partial_header
 from sfm.waveform_store import (
    _serial_from_bw_bytes,
    _serial_from_bw_filename,
    _serial_number_from_bw_filename,
 )
 # ── the filename gives a number, and only a number ──────────────────────────
@pytest.mark.parametrize("name,num", [
    ("P036L318.C80H", 14036),   # BE14036
    ("H907KWRK.WB0H", 6907),    # BE6907
    ("M529LKIQ.G10",  11529),   # BE11529
    ("T003LQ9K.OE0H", 18003),   # BE18003
    ("L895K63F.GE0W", 10895),   # BA10895 — a BlastMate
    ("K229HGQI.XO0W", 9229),    # BA9229  — a BlastMate
 ])
 def test_number_from_filename(name, num):
    assert _serial_number_from_bw_filename(name) == num
@pytest.mark.parametrize("name", ["", "not_a_bw_file.bin", "AB12", "1234ABCD.XX0W"])
 def test_number_from_filename_rejects_junk(name):
    assert _serial_number_from_bw_filename(name) is None
 def test_filename_only_decoder_is_a_guess():
    """It still answers "BE" — that is why it must not be the first choice."""
    assert _serial_from_bw_filename("L895K63F.GE0W") == "BE10895"
    assert _serial_from_bw_filename("M529LKIQ.G10") == "BE11529"
    assert _serial_from_bw_filename("nonsense") is None
 # ── the body carries the truth ──────────────────────────────────────────────
 def _body(serial: bytes) -> bytes:
    return b"\x00" * 32 + b"STRT" + b"\xff\xfe" + serial + b"\x00Geo: 0.254 in/s\x00"
 def test_body_wins_for_a_blastmate():
    assert _serial_from_bw_bytes(_body(b"BA10895"), "L895K63F.GE0W") == "BA10895"
 def test_body_wins_for_a_minimate():
    assert _serial_from_bw_bytes(_body(b"BE11529"), "M529LKIQ.G10") == "BE11529"
 def test_body_candidate_must_match_the_filename_number():
    """A serial-shaped byte run that disagrees with the filename is ignored."""
    assert _serial_from_bw_bytes(_body(b"XX99999"), "L895K63F.GE0W") is None
 def test_body_tolerates_a_leading_zero():
    assert _serial_from_bw_bytes(_body(b"BA09229"), "K229HGQI.XO0W") == "BA09229"
@pytest.mark.parametrize("data,name", [
    (b"", "L895K63F.GE0W"),          # no bytes
    (_body(b"BA10895"), "junk.bin"),  # no derivable number
 ])
 def test_body_returns_none_when_it_cannot_decide(data, name):
    assert _serial_from_bw_bytes(data, name) is None
 # ── the live monitor-log path ───────────────────────────────────────────────
 def _partial_record(serial: bytes) -> bytes:
    """0x2C partial record: type, prefix, two 9-byte timestamps, then ASCII."""
    ts = bytes([11, 0x10, 4, 0x07, 0xE9, 0, 16, 2, 0])   # 2025-04-11 16:02:00
    return (bytes([0x2C]) + b"\x00" * 10 + ts + ts
            + b"\x00\x00\x00\x00" + serial + b"\x00Geo: 0.254 in/s\x00")
@pytest.mark.parametrize("serial", [b"BE11529", b"BA10895", b"UM11719"])
 def test_monitor_log_reads_any_family_prefix(serial):
    entry = _decode_0a_partial_header(_partial_record(serial), 0, b"\x01\x11\x00\x00")
    assert entry is not None
    assert entry.serial == serial.decode()
 def test_monitor_log_geo_threshold_survives_a_blastmate():
    """The old find(b"BE") skipped the whole block, losing geo too."""
    entry = _decode_0a_partial_header(_partial_record(b"BA10895"), 0, b"\x01\x11\x00\x00")
    assert entry is not None
    assert entry.geo_threshold_ips == pytest.approx(0.254)
@@ -3,31 +3,34 @@ from sfm.database import SeismoDb
 from minimateplus.models import Event, Timestamp
-def _ins(db, key, serial, pvs, ts):
+def _ins(db, key, serial, pvs, ts, record_type="Waveform"):
    ev = Event(index=0)
    ev._waveform_key = bytes.fromhex(key)
    ev.timestamp = ts
    # peak_vector_sum comes from peak_values; simplest: insert then UPDATE pvs directly
    db.insert_events([ev], serial=serial)
    row = [r for r in db.query_events(serial=serial) if r["waveform_key"] == key][0]
    with sqlite3.connect(db.db_path) as c:
-        c.execute("UPDATE events SET peak_vector_sum=? WHERE id=?", (pvs, row["id"]))
+        c.execute("UPDATE events SET peak_vector_sum=?, record_type=? WHERE id=?",
                  (pvs, record_type, row["id"]))
    return row["id"]
-def test_propagate_copies_flags_to_twins(tmp_path):
+def _ts(hour, minute, second=0, day=25):
    return Timestamp(raw=b"", flag=0x10, year=2026, unknown_byte=0, month=2, day=day,
                     hour=hour, minute=minute, second=second)
 def test_propagate_copies_flags_across_hours_apart_twins(tmp_path):
    # Flagging the waveform FT propagates to its histogram twin 75 min earlier
    # (the interval matcher pairs them; the old ±5-min window would have missed it).
    db = SeismoDb(tmp_path / "s.db")
-    base = Timestamp(raw=b"", flag=0x10, year=2026, unknown_byte=0, month=2, day=25, hour=20, minute=19, second=5)
+    hist  = _ins(db, "01110001", "BE1", 0.4763, _ts(19, 31, 17), "Histogram")
-    twin = Timestamp(raw=b"", flag=0x10, year=2026, unknown_byte=0, month=2, day=25, hour=20, minute=19, second=45)
+    wave  = _ins(db, "01110002", "BE1", 0.4763, _ts(20, 46, 44), "Waveform")   # twin, 75 min later
-    other = Timestamp(raw=b"", flag=0x10, year=2026, unknown_byte=0, month=2, day=25, hour=20, minute=19, second=44)
+    other = _ins(db, "01110003", "BE1", 0.9999, _ts(20, 20, 0),  "Waveform")   # different pvs
-    primary_id = _ins(db, "01110001", "BE1", 0.4763, base)
+    db.update_event_review(wave, {"false_trigger": True})
-    twin_id = _ins(db, "01110002", "BE1", 0.4763, twin)        # twin: same serial+pvs, 40s apart
+    moved = db.propagate_review_to_twins(wave)
    non_twin_id = _ins(db, "01110003", "BE1", 0.9999, other)   # near time but different pvs
-    db.update_event_review(primary_id, {"false_trigger": True})
+    assert hist in moved
-    moved = db.propagate_review_to_twins(primary_id)
+    assert db.get_event(hist)["false_trigger"] == 1
-
+    assert db.get_event(other)["false_trigger"] == 0
    assert twin_id in moved
    assert db.get_event(twin_id)["false_trigger"] == 1
    assert db.get_event(non_twin_id)["false_trigger"] == 0
@@ -712,8 +712,27 @@ def test_forty_nn_is_a_data_block_not_a_segment_header():
    """
    assert data_block_len(b"\x40\x02\x00\x01\x00\x02", 0) == (6, 2)
    assert data_block_len(b"\x40\x08" + bytes(16), 0) == (18, 8)
-    # NN > 8 is not a data block
+
-    assert data_block_len(b"\x40\x0c" + bytes(24), 0) == (None, None)
+
 def test_forty_nn_is_not_capped_at_eight():
    """NN > 8 is a perfectly ordinary `40 NN` block.
    This test previously asserted the opposite (`40 0c` -> (None, None)),
    codifying a guard that had no evidence behind it: the only corpora
    available then used NN in {1,2,3,4,8}, so the cap was never exercised.
    Loud UM12947 events use NN of 12, 16, 20 ... up to 196, and rejecting
    them halted the block walk mid-record — surfacing as silently short
    channels, since the walker stops at the first unrecognised tag rather
    than raising.  Lifting the cap took that corpus from 22 length-mismatched
    files to 0, and 1,476,242 of 1,476,249 samples now reproduce Thor's own
    CSV export exactly (the 7 stragglers differ by one 4th-decimal tick).
    Verified 2026-09-11; see docs/idf_protocol_reference.md.
    """
    assert data_block_len(b"\x40\x0c" + bytes(24), 0) == (26, 12)
    assert data_block_len(b"\x40\xc4" + bytes(392), 0) == (394, 196)
    # The real bound is the buffer: a block that cannot fit is not a block.
    assert data_block_len(b"\x40\xc4" + bytes(8), 0) == (None, None)
    assert data_block_len(b"\x40\x00" + bytes(8), 0) == (None, None)
 def test_record_chain_is_followed_by_length_not_by_tag_sniffing():
@@ -0,0 +1,85 @@
 """Blastware-compatible channel FFT (waveform_fft).
 Reverse-engineered 2026-09-14 against 7 BE12844 (MiniMate Plus) events, each with
 a Blastware FFT report as ground truth. The recipe (DC-remove, no window,
 zero-pad to 4096 → 0.25 Hz bins, single-sided 2/N amplitude) reproduces
 Blastware's dominant frequency to the exact bin on all 28 channels and the
 amplitude to report precision.
 """
 from pathlib import Path
 import numpy as np
 from waveform_fft import channel_spectrum, dominant_frequency
 from minimateplus.waveform_codec import decode_waveform_v2
 FIXDIR = Path(__file__).parent / "fixtures" / "fft-oracle-2026-09-14"
 GEO_LSB = 0.005   # 1 decode unit = 16 ADC counts = 0.005 in/s (series-3 Normal range)
 # Blastware FFT-report ground truth: file → {channel: (dominant_hz, amplitude_ips)}.
 # amplitude is None where the channel is at the noise floor (report amp 0.000/0.001)
 # — the dominant frequency still matches exactly, but the amplitude isn't meaningful.
 ORACLE = {
    "N844LPGH.VV0W": {"Tran": (27.00, 0.018), "Vert": (26.75, 0.009), "Long": (26.50, 0.021), "MicL": (2.000, None)},
    "N844LPPR.3S0W": {"Tran": (30.75, None),  "Vert": (46.75, None),  "Long": (26.75, None),  "MicL": (49.50, None)},
    "N844LQHB.ZT0W": {"Tran": (19.75, 0.040), "Vert": (26.50, 0.018), "Long": (26.50, 0.083), "MicL": (2.750, None)},
    "N844LQUE.T50W": {"Tran": (21.50, 0.080), "Vert": (14.25, 0.028), "Long": (28.50, 0.046), "MicL": (5.750, None)},
    "N844LR8W.790W": {"Tran": (31.00, None),  "Vert": (31.00, None),  "Long": (34.00, None),  "MicL": (66.25, None)},
    "N844LRCO.G60W": {"Tran": (32.25, 0.009), "Vert": (32.00, 0.005), "Long": (32.00, 0.008), "MicL": (32.00, None)},
    "N844LRCW.F30W": {"Tran": (21.25, 0.010), "Vert": (42.25, 0.002), "Long": (21.25, 0.014), "MicL": (21.25, None)},
 }
 def test_pure_sine_frequency_and_amplitude():
    # A pure sine at a bin-centre frequency (128 cycles over 4096 samples) has no
    # leakage, so the single-sided 2/N normalisation returns the amplitude exactly.
    sps, n, f0, amp = 1024.0, 4096, 32.0, 0.5
    x = amp * np.sin(2 * np.pi * f0 * np.arange(n) / sps)
    freqs, amps = channel_spectrum(x, sps=sps, nfft=4096)
    fpk, apk = dominant_frequency(freqs, amps)
    assert fpk == 32.0
    assert abs(apk - amp) < 1e-3
 def test_bin_resolution_is_quarter_hz():
    freqs, _ = channel_spectrum(np.zeros(3328), sps=1024.0, nfft=4096)
    assert abs((freqs[1] - freqs[0]) - 0.25) < 1e-9
 def test_empty_input():
    freqs, amps = channel_spectrum([])
    assert len(freqs) == 0 and len(amps) == 0
 def _spectra(fname):
    raw = (FIXDIR / fname).read_bytes()
    dec = decode_waveform_v2(raw[raw.find(b"STRT") + 21:])
    out = {}
    for ch, samples in dec.items():
        ips = np.asarray(samples, float) * GEO_LSB
        out[ch] = channel_spectrum(ips, sps=1024.0)
    return out
 def test_dominant_frequency_matches_blastware_exactly():
    misses = []
    for fname, chans in ORACLE.items():
        spectra = _spectra(fname)
        for ch, (want_hz, _) in chans.items():
            got_hz, _ = dominant_frequency(*spectra[ch])
            if abs(got_hz - want_hz) > 0.25:
                misses.append(f"{fname}:{ch} got {got_hz} want {want_hz}")
    assert not misses, "dominant-frequency mismatches:\n" + "\n".join(misses)
 def test_amplitude_matches_blastware():
    misses = []
    for fname, chans in ORACLE.items():
        spectra = _spectra(fname)
        for ch, (_, want_amp) in chans.items():
            if want_amp is None:
                continue
            _, got_amp = dominant_frequency(*spectra[ch])
            if abs(got_amp - want_amp) > 0.0015:
                misses.append(f"{fname}:{ch} got {got_amp:.4f} want {want_amp:.3f}")
    assert not misses, "amplitude mismatches:\n" + "\n".join(misses)
@@ -0,0 +1,66 @@
 """Blastware-compatible FFT of a decoded seismograph channel.
 Pure numpy; no I/O, no device or DB dependencies. Feed it a channel's decoded
 samples **in the unit you want the amplitudes in** (e.g. in/s) and it returns the
 single-sided amplitude spectrum that Blastware's *FFT Report* draws.
 Reverse-engineered 2026-09-14 against 7 BE12844 (MiniMate Plus) events with
 Blastware FFT reports as ground truth. The recipe reproduces Blastware's
 **dominant frequency to the exact 0.25 Hz bin on all 28 channels** and the
 amplitude to report precision:
  1. remove the DC component (subtract the mean); **no window** — a window
     smears the peak and measurably worsens the match,
  2. zero-pad to ``nfft`` (4096 → 0.25 Hz bins at 1024 sps — Blastware's
     resolution),
  3. single-sided amplitude ``A[k] = 2·|X[k]| / N`` where ``N`` is the real
     sample count (not ``nfft``).
 The compliance chart (USBM RI8507 / OSMRE) is this spectrum's ``(freq, amp)``
 points plotted against the regulatory limit curve; the #10 FFT view is the
 spectrum itself.
 """
 from __future__ import annotations
 import numpy as np
 BW_NFFT = 4096              # 0.25 Hz bins at 1024 sps — Blastware's FFT resolution
 BW_FMIN = 2.0              # dominant-frequency search floor (Hz)
 BW_FMAX = 250.0           # dominant-frequency search ceiling (Hz)
 def channel_spectrum(samples, sps: float = 1024.0, nfft: int = BW_NFFT):
    """Single-sided amplitude spectrum of one channel, Blastware-compatible.
    ``samples`` is a 1-D sequence in the desired amplitude unit (in/s). Returns
    ``(freqs, amps)`` numpy arrays covering ``0 .. sps/2`` in ``sps/nfft`` steps.
    Records longer than ``nfft`` are truncated by the transform — untested
    against Blastware for that case (real MiniMate Plus records are ≤ ~3.3 s,
    well under 4096 samples at 1024 sps).
    """
    x = np.asarray(samples, dtype=float)
    n = x.size
    if n == 0:
        return np.empty(0), np.empty(0)
    x = x - x.mean()                          # DC removal, no window
    mag = np.abs(np.fft.rfft(x, nfft))
    freqs = np.fft.rfftfreq(nfft, 1.0 / sps)
    amps = (2.0 / n) * mag                     # single-sided amplitude
    return freqs, amps
 def dominant_frequency(freqs, amps, fmin: float = BW_FMIN, fmax: float = BW_FMAX):
    """Peak ``(frequency_hz, amplitude)`` of a spectrum within ``[fmin, fmax)``.
    Matches Blastware's "Dominant Frequency" — the largest spectral bin in the
    reportable band (below 2 Hz is baseline/DC drift, above 250 Hz is noise).
    """
    freqs = np.asarray(freqs)
    amps = np.asarray(amps)
    lo = int(np.searchsorted(freqs, fmin))
    hi = int(np.searchsorted(freqs, fmax))
    if hi <= lo:
        return 0.0, 0.0
    k = lo + int(np.argmax(amps[lo:hi]))
    return float(freqs[k]), float(amps[k])