12-OBD-Speed: read-only UDS DID scan of the car's modules

- Log page now runs a diagnostic scan instead of CAN listening (the Fiesta's OBD
  port carries no broadcast traffic): unfiltered check, module discovery
  0x700-0x7EF (22F188, then 3E00 with 1 s timeout), all 65536 DIDs per module,
  progress in NVS, speed/voltage polled every 2 s in between
- Result (drives 2026-11-12..18): only the PCM 0x7E0 answers; 389 DIDs supported
  (logs in docs/canlog/scan, scan2)
- Reset scan button, BLE connection interval request, progress lines
- CLAUDE.md: screen-has-priority rule, web UI future idea, findings

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
mangavalkandClaude Opus 5.5 committed 2026-11-18 16:06:31 +01:00
1 parent 60cd7cf262
commit c9ad39c8ed
27 files changed
+15622 -380

No files matched your search

+35 -19
View File
@@ -16,9 +16,9 @@ Uses the shared board library and LVGL 9.6 (`../shared`), NimBLE-Arduino 2.5.1.
dimensions shown; not 360×360 in orange). Tap one: it is copied into the device's flash (progress shown) and
played from there, so the SD card is not needed afterwards. At boot the device checks (in the background)
whether the chosen file on the SD card changed (size/date) and re-copies it. **Rescan** re-reads the card.
- **Swipe left once more → CAN log** (temporary, to find the cruise/ACC set speed and speed-sign messages, see below).
**Start/Stop** (remembered: logging resumes by itself after power-up), **Mark**, the last frame of the three Ford
candidate IDs, and the file on the SD card. While logging, the speed page shows a red `REC <marks>` pill and
- **Swipe left once more → CAN log** (temporary: diagnostic scan to find the cruise/ACC set speed and speed limit, see below).
**Start/Stop** (remembered: logging resumes by itself after power-up), **Mark**, scan progress, **Reset scan**,
and the file on the SD card. While logging, the speed page shows a red `REC <marks>` pill and
**a tap on the speed page adds a mark** (pill flashes `MARK n`).
- **Idle** — after that time without speed data (ignition off, adapter gone) the screen goes dark, the GIF pauses and
OBD polling stops; only `ATRV` every 30 s (answered by the adapter itself, doesn't wake the car's bus).
@@ -33,34 +33,50 @@ Service `FFF0`, notifications + commands on `FFF1` (verified with the user's "OB
Init: `ATZ`, `ATE0`, `ATL0`, `ATS0`, `ATH0`, `ATSP0`, `0100`; then `010D` (speed) is polled continuously.
Serial (115200) prints status changes as `[obd] ...`.
## CAN logger (finding ACC set speed / speed sign)
Neither value is a standard OBD PID; on Fords they are broadcast CAN messages. Candidates from opendbc
(`ford_lincoln_base_pt.dbc`, made for newer Fords; **not yet confirmed on the Fiesta Mk8**):
`0x165` EngBrakeData `Veh_V_DsplyCcSet` (set speed as shown) + `CcStat_D_Actl`, `0x202` `Veh_V_RqCcSet` (0.5 km/h),
`0x3D9` IPMA_Data2 `IsaVLim_D_Rq` (speed limit, 254 = no data) + `IsaVLimUnit_D_Rq` (1 km/h, 2 mph).
## Log page: diagnostic scan (finding ACC set speed / speed sign)
Neither value is a standard OBD PID. **First test drive (2026-11-02, `docs/canlog`)**: listening on the OBD port
(ELM327 `ATMA`, ID survey with wildcard filters) saw **no broadcast frames at all**, only diagnostic replies: the
Fiesta Mk8's gateway passes only diagnostics to the OBD socket. (Broadcast candidates from opendbc for newer Fords,
for reference: `0x165` `Veh_V_DsplyCcSet`, `0x202` `Veh_V_RqCcSet`, `0x3D9` `IsaVLim_D_Rq`; the Mk7.5 Fiesta sheet in
github.com/consp/ford-can-stuff lists `0x3CD` "Traffic Recognition Data" on the infotainment bus.)
While logging, the OBD task (after `ATCSM1` silent monitoring, `ATH1`, `ATCAF0`):
1. **Survey**: listens to the ID ranges `0XX`..`7XX` (`ATCRA` with wildcards + `ATMA`, 1.3 s each); a range that
overflows the adapter (`BUFFER FULL` / 300 lines) is split into 16 smaller ones. Repeated every 10 min.
2. **Capture**: one frame of every ID found, in turn; the three candidates after every 8 others.
3. Every 2 s `ATAR` + `ATCAF1` + `010D` (speed, so the display keeps working, a bit laggy) and `ATRV`.
If the adapter can't filter by wildcard it falls back to unfiltered 2 s windows.
So the values have to be asked from the modules. While logging is on, the OBD task runs a **read-only scan**
(only UDS `22` ReadDataByIdentifier in the default session; nothing in the car is changed), progress kept in NVS
(namespace `scan`) so it continues on every drive:
1. **Unfiltered listen check** (once): 2 × 5 s `ATMA` without filter, CAF off and on.
2. **Module discovery**: `22F188` (software number) to every address 0x700..0x7EF (`ATSH` + `ATCRA` addr+8,
`ATST19` = 100 ms timeout). Logged as `D <addr> <reply>`, list as `I`. Repeated if the PCM (0x7E0) didn't answer.
3. **DID scan**: all 65536 DIDs of each module, order 0x706 (IPMA camera), 0x764 (CCM radar), 0x720 (IPC), 0x7E0 (PCM),
then the other modules found; per module the DID ranges D000-DFFF, 4000-4FFF, 0000-3FFF, 5000-CFFF, E000-FFFF.
Positive replies `S <addr> <did> <data>`, other negative responses than 0x31 `R <addr> <did> <nrc>`. A module
without service 22 (NRC 0x11) is skipped; 20 requests without a reply = module asleep: back up and wait.
The `<request>1` suffix (return after the first reply) is used when the adapter supports it.
**Probe** (once, after drive 2 found only the PCM): `3E00` TesterPresent with a 1 s timeout to every address not
found yet. DIDs are asked **3 per request** (`22 D000 D001 D002`, one CAN frame; the module returns only the
supported ones) when the module passes the test `22F1FFF188F1FE`; a hit is re-asked one DID at a time.
`E progress` every 512 DIDs (with ms/DID).
4. Every 2 s `ATSH 7DF` / `ATAR` + `010D` (speed, so the display keeps working, a bit laggy) and `ATRV`.
Page: Start/Stop (remembered), Mark, progress (module, DID, %), modules found, DIDs found, last positive reply,
**Reset scan** (starts over the next time logging runs). While logging the speed page shows `REC <marks>`; a tap on
it adds a mark.
Log: `/canlog/log_NNN.txt` (new file per session, appended every 2 s; kept in RAM while the card is missing).
Lines `<ms> <type> <data>`: `C` adapter, `P` protocol, `V` speed, `B` voltage, `M` mark, `F id data` frame,
`N id` no frame, `I` survey IDs, `E` adapter message/note. Test drive: change the set speed and pass signs, tap
for a mark each time (passenger, or only when safe).
Lines `<ms> <type> <data>`: `C` adapter, `P` protocol, `V` speed, `B` voltage, `M` mark, `D`/`I` modules,
`S`/`R` DID replies, `F` raw frame, `E` adapter message/note.
**Download over USB** (no card reader needed): serial commands `ls` and `get <file>` (115200 baud).
## Files
| File | What |
|---|---|
| `src/main.cpp` | LVGL UI (tileview: speed, adapter menu, settings, background, CAN log) |
| `src/obd_link.cpp/.h` | BLE scan, connect/reconnect, ELM327 init and speed polling, CAN logging steps (background task on core 0) |
| `src/obd_link.cpp/.h` | BLE scan, connect/reconnect, ELM327 init and speed polling, diagnostic scan steps (background task on core 0) |
| `src/font_speed_120.c` | Montserrat Medium 120 px, digits and `-`, 4 bpp (generated with `lv_font_conv`, see header) |
| `src/can_log.cpp/.h` | CAN log buffer + SD writer (background task), marks, values shown on the log page |
| `src/can_log.cpp/.h` | Log buffer + SD writer (background task), marks, info shown on the log page, serial `ls`/`get` |
| `src/sd_card.cpp/.h` | Shared, reference-counted SD_MMC mount (background store + logger) |
| `src/bg_store.cpp/.h` | Background GIF from the SD card: list `/gifs`, copy into the `bggif` flash partition (header written last), mmap for playback, change check at boot (background task) |
| `partitions.csv` | 4 MB app (no OTA) + 11.9 MB `bggif` raw area + NVS |
| `docs/canlog/` | Logs of the test drives |
| `data/bg.gif` | Built-in background (`ezgif-5-cb56986bc7.gif`, 360×360), linked into the firmware via `board_build.embed_files` |
| `sd/gifs/` | (not in git) GIFs converted with `../tools/gif_to_360.py`, ready to copy to the SD card |