Files
tpms-schrader-eg53ma4-sim/docs/PROTOCOL.md
T
czcionka e183d23950
check / check-without-hardware (push) Successful in 6m45s
Docs and code: translate Polish comments, messages and documentation to English
- README, CHANGELOG, RELEASE_NOTES, THIRD_PARTY_NOTICES and docs/*.md
  translated to English; cross-markdown anchors updated
- C/C++/Python/shell comments and user-facing messages translated
- identifiers, protocol names, commands and literals unchanged
- native tests re-verified: 24/24 PASS; tpms_frame.py self-test PASS
2026-10-30 18:01:32 +00:00

110 lines
5.2 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# Schrader EG53MA4 protocol — facts and unknowns
## Pinned sources
| Source | Version used for validation | Scope |
|---|---|---|
| `rtl_433/src/devices/schraeder.c` | commit `406f9a33550a2b6b3d0d8a8a486417ce5b3f13ab` | decoder and slicer parameters |
| `rtl_433_tests/tests/Schrader_EG53MA4_TPMS/01` | commit `1aa8c29d6b9383739d1aeed495dbc5069b704392` | four CU8 recordings + JSON |
| PlatformIO | 6.1.18 | build/test |
| `espressif32` | 8.0.0 | toolchain/ESP32 platform |
| SmartRC-CC1101-Driver-Lib | `de6757ef633b3e626dae5c7835cdb66ab53f9a1d` (3.0.2) | CC1101 configuration |
| Practical receiver analysis | <https://milek7.pl/tpmsreceiver/> | context; not a source of frame fields |
| ESP32+CC1101 integration pattern | <https://github.com/merbanan/rtl_433/issues/2978> | integration, not the EG53MA4 protocol |
We do not use the unlabelled `master` as a reproducible reference. The
commercial emulator <https://spooftpms.com/us-eu-gm-1-2-family-tpms-simulator/>
only confirms that such a class of device exists, not this project's
compatibility.
## Confirmed format after Manchester decoding
The decoder requires exactly 120 bits: 40 bits of synchronisation and 80 data
bits.
```text
0 40 72 96 104 112 120
| 40 × sync (0) | flags 4 B | ID 3 B | P 1 B | T 1 B | sum 1 B |
4C 90 00 70 2F C8 71 68 49 65
```
- `flags`: 32 bits; the field meanings are not fully known.
- `ID`: 24 bits; the firmware rejects 0, `FFFFFF`, and duplicates between
wheels.
- `P`: 2.5 kPa/bit **according to the rtl_433 decoder, but the scale is not
confirmed** — all four captures have `P = 0`, so no real pressure verified
the conversion (in the decoder source it is "probable"). A car measurement:
`P = 104` (260 kPa on the rtl_433 scale) shows up as **284 kPa**, so the
firmware corrects the scale with the constant
`TPMS_PRESSURE_KPA100_PER_RAW` (`tpms_frame.h`, default 273 = 2.73 kPa/bit;
`P = 95` for 260 kPa). Quantisation: `(kPa×100 + 136) / 273`. Calibration
confirmed on hardware: 260 kPa → 260 kPa on the car display (`AUDIT.md`).
- `T`: byte in °F per the decoder, not °C. 23 °C → 73 °F.
- `sum`: the sum of the nine preceding bytes modulo 256; it is not CRC-8.
- The value `4c900070` occurs in all 64 records of the four captures. That
does not prove that every sensor/car mode always uses this value.
## Polarity, preamble, and tail
The recordings and the `OOK_PULSE_MANCHESTER_ZEROBIT` slicer confirm G.E.T.
Manchester: `0 → 01`, `1 → 10`. The slicer prepends the first zero bit. The
CC1101 buffer contains:
```text
1 starting chip + 39 sync cells + 80 data cells + 1 OFF padding chip
= 240 chips = 30 bytes, MSB first.
```
The last padding chip is an OOK zero (silence). The test checks exactly this
length and polarity and rejects a truncated/corrupted buffer.
## Timing and repetitions
- The decoder has `short_width=123 µs`, `reset_limit=300 µs`.
- Analysis of `gfile001` shows a typical pulse+gap period of about 236 µs,
i.e. a chip of about **118 µs**; pulse widths differ from gap widths. The
firmware uses **120 µs** experimentally, which decodes on an independent
`rtl_433`. The exact clock of a physical CC1101 requires an HIL measurement.
- Each of the four recordings contains **16 valid repetitions**. In
`gfile001` the gaps between decode starts vary roughly 115–172 ms; they are
not constant.
- The defaults of **3 frames, a 50 ms gap, a 30 s cycle** are experimental
emulator settings, not a confirmed OE schedule. The ACC/parking intervals
also have no OE-compatibility status.
## Carrier and modulation
The base profile: **433.92 MHz, ASK/OOK**. The recordings are decoded at this
frequency. The kit's README also contains 315 MHz product data and a mention
of ASK+FSK; that is no evidence of FSK in the analysed files nor of
compatibility with the specific Astra. FSK is not available in the base
firmware/generator.
## CC1101 packet engine
The firmware sets fixed-length 30 B, a normal FIFO, MSB first, and disables
the hardware preamble/sync (`SYNC_MODE=0`), Manchester, whitening, FEC, CRC,
address check, and append status. The whole preamble and the protocol's sum
are inside our 30 bytes. End of TX is expected via `MARCSTATE=IDLE` with a
limit; `TXFIFO_UNDERFLOW` and leftover bytes are errors. This prevents the
tail from being truncated by `TXBYTES=0` alone.
The configuration has been audited, but the synthetic CU8 bypasses the
CC1101. The absence of extra bytes and the exact tail still have to be
confirmed on a logic analyser/RTL-SDR in HIL.
## Unknowns — they must not be "tuned" randomly
1. The meaning of individual `flags` bits, including any sensor low-battery
bit. **`TPMS_FLAGS_LOWBAT_MASK` does not exist in the firmware.**
2. The dependence of flags on learning, motion, LF, and the original sensor's
battery state.
3. The interval and minimal number of repetitions required by a specific ECU.
4. The compatibility of OE 13506028/a related part number with a specific
unit without a VIN, markings, a readout, or a test.
5. The physical accuracy of frequency, power, and the tail of the generated
CC1101 transmission before HIL.
Low emulator supply is only a console warning and does not change any field
of the simulated sensor.