diff --git a/handoff.md b/handoff.md index 89ea73c..ede7577 100644 --- a/handoff.md +++ b/handoff.md @@ -15,17 +15,27 @@ Pick-up notes for diagnosing the truck in the cab. Repo: `git.jpaul.io/justin/fo - Working **baud: 38400** (default in the tool). If you get garbage, try `9600`. - It's an `AT@2 = ?` clone (no stored serial) — normal, works fine for generic OBD-II. -## What the tool does / doesn't -- DOES: read stored/pending/permanent codes, decode them (6.0 codes flagged as - no-start suspects), clear codes (`--clear`), show battery voltage + key live PIDs. -- EXPERIMENTAL (`--ford`): tries a small table of community-sourced Ford 6.0 - Mode-22 PIDs (ICP, IPR%, FICM main/sync/logic V, EBP, EOT). **PID numbers and - scaling are TENTATIVE** — raw bytes are always printed; cross-check against a - meter or FORScan before trusting a value. -- `--pid XXXX`: one-shot probe of any 16-bit Mode-22 PID; prints raw bytes plus - a few common decodings so you can eyeball the right scaling. -- DOESN'T (yet): a verified Ford-enhanced PID set the way FORScan has. We're - building toward that — see follow-ups below. +## What the tool does +- **DTCs**: stored / pending / permanent (modes 03/07/0A), with 6.0-relevant + codes flagged as no-start suspects. `--clear` erases (mode 04). +- **Generic live PIDs**: RPM, ECT, IAT, MAP, engine load, module voltage, ATRV. +- **`--ford`**: Mode-22 read of the **VERIFIED** Ford 6.0 PID set — ICP, EBP, + MAP, BARO, MGP (boost), EOT, gear, TSS, plus the documented FICM trio + (`09D0` Main, `09CF` Logic, `09CE` Vehicle V) and `09CD` Sync. Full PID + research + sources in + [diagnostics/2026-06-29-no-start/pid-research.md](diagnostics/2026-06-29-no-start/pid-research.md). +- **`--crank`**: dedicated cranking monitor — large ICP readout + trace of ICP, + FICM Main, battery, RPM. **This is the no-start mode.** +- **`--dash`**: real-time gauge dashboard. Presets: `--dash crank` (cranking), + `--dash full` (everything). Add `--dash-log run.csv` to capture a CSV trace. +- **`--pid XXXX`**: one-shot probe of any 16-bit Mode-22 PID; prints raw bytes + plus a few candidate decodings. +- **`--scan AAAA-BBBB`** (with optional `--scan-log PATH`): brute-force a PID + range. Use for hunting new PIDs. + +What still needs FORScan: **desired ICP** (`ICP_DES`) and a labeled FICM-Sync +yes/no gauge — neither has a published Mode-22 DID. Everything else for +no-start diagnosis is reachable through this tool now. ## 6.0 no-start priority checklist (from the triage) The 6.0 needs, to fire: **good batteries → FICM ~48V → ICP ~500 psi → fuel → cam/crank signal.** @@ -55,35 +65,45 @@ The 6.0 needs, to fire: **good batteries → FICM ~48V → ICP ~500 psi → fuel time** and runs/idles fine until shut off — then needs ether again. That signature ⇒ **high-pressure oil (ICP) bleed-off during cranking** — STC fitting / oil rail O-rings / HPOP-related. Compression, FICM, CMP/CKP, - fuel supply all confirmed good. Next: visual under valve covers + STC - fitting inspection. -- Pushed to `git.jpaul.io/justin/ford-obd`, branch `main`. Files: `obd_reader.py`, - `RUN_OBD.bat`, `README.md`, `README.txt`, `handoff.md`, `diagnostics/`. + fuel supply all confirmed good. +- **2026-06-30 desktop session:** multi-agent PID research workflow corrected + the Mode-22 table — the original 12xx guesses were wrong addresses, not a + bus problem. Verified PID set now in `obd_reader.py`; reasoning + sources + in [pid-research.md](diagnostics/2026-06-29-no-start/pid-research.md). + Added `--dash` (real-time gauges) and `--crank` (no-start monitor). +- Pushed to `git.jpaul.io/justin/ford-obd`, branch `main`. ## To resume with Claude from the cab Mention: "6.0 Power Stroke no-start, using the ford-obd tool (git.jpaul.io/justin/ford-obd)". Then **paste the tool's full output** (codes + live values). Useful to also say: cranks vs. no-crank, hot vs. cold, what changed before it died, and FICM/ICP readings if you metered them. -## To resume from the desktop (after the 2026-06-29 session) -Read [diagnostics/2026-06-29-no-start/README.md](diagnostics/2026-06-29-no-start/README.md) -first — that's the full state. Top open items, in order: -1. **Visual under valve covers + STC fitting** — looking for the high-pressure - oil bleed source. That's where the diagnosis lands. -2. Re-probe the clean scan hits (`1310`, `1440-1451`, `11Bx`) with `--pid XXXX` - for uncontaminated data; map them to known engine conditions to identify - ICP / IPR. -3. FORScan via the CyanLabs mirror once available — ICP and IPR live data - would confirm a high-pressure leak in seconds. +## To resume at the truck (2026-06-30 state) -## Open follow-ups (when off the truck) -- FORScan from the CyanLabs mirror once forscan.org is back — useful as the - ground truth for verifying our Mode-22 PID table. -- **Verify** every entry in `FORD_60_PIDS` (obd_reader.py) against either a - meter reading or FORScan, one at a time. As each is confirmed, drop the - "TENTATIVE" warning for that row. -- **Option 3**: brute-force scan of the Mode-22 PID space (e.g. `0x1000`- - `0x14FF`) and log which PIDs return data, so we can fingerprint what this - PCM exposes without external references. Hook it in as `--scan`. -- Add `--watch` mode that streams ATRV + PID 0142 (and the FICM PIDs once - verified) at ~5 Hz so we can watch the voltage sag during cranking. +The diagnosis stands: high-pressure oil (ICP) bleeds off during cranking; +HPOP outruns the leak once running. The toolchain can now confirm this +directly from the OBD port. Top actions, in order: + +1. **`python obd_reader.py --crank`** during a short crank. A healthy 6.0 + builds ~500+ psi ICP in 1–2 seconds. If ours stalls below 500 psi or + builds very slowly, that confirms the bleed-off and is the smoking gun. +2. **Probe the FICM PIDs that were outside the scan window** to validate + the documented `09xx` family on this truck: + - `python obd_reader.py --pid 09D0` (FICM Main, expect ~48V) + - `--pid 09CF` (Logic, ~12V), `--pid 09CE` (Vehicle V), `--pid 09CD` (Sync) +3. **Visual under the valve covers + STC fitting** — the likely physical + source of the leak. STC fitting is the #1 6.0 culprit for this pattern. +4. **Widen the brute scan** to round out the verified set — + `python obd_reader.py --scan 0900-09FF --scan-log scan-09xx.txt` and + `--scan 1600-16FF --scan-log scan-16xx.txt`. + +## Open follow-ups +Detailed action list lives in +[pid-research.md §5](diagnostics/2026-06-29-no-start/pid-research.md). Highlights: +- Validate the documented-but-untested PIDs (`09D0/09CF/09CE/09CD`, `16AD`, + `1434`, `1172`, `1672`, `096D`, `099F`) by reading them on-vehicle. +- Sanity-check IPR (`1434`) against expected ranges: KOEO ~14–15 %, + cranking ~30–40 %, warm idle ~25–30 %. +- Calibrate `16C1` fuel-level scaling per-truck (published divisors misread). +- FORScan via CyanLabs mirror — only path to `ICP_DES` and a labeled FICM + Sync gauge.