These are the frames that remote control my 2016 Volt, in the order you need them. My car is second generation, the adapter is a Macchina M2 running M2RET on a Raspberry Pi, and everything here was run against the car rather than copied from a forum. Where the existing documentation disagrees with the car, I say so in a short note.
1. Make the adapter actually receive
It turns out there is a bug in the stock M2 firmware, and it makes everything else impossible. Here's the fix: set CAN0EN=1 and then press the reset button on the board. The console command calls set_baudrate() and enable(), but the setup that opens the channel only runs in setup() at boot, so a channel enabled at runtime reports itself enabled and receives nothing at any bitrate.
The console implies that
CAN0EN=1is enough. It isn't, and silence from a channel that was never initialised is indistinguishable from silence from a sleeping car.
After the reset mine read 1921 frames in ten seconds. Before it, that channel had shown zero for months.
2. Wake the car
It turns out the wake is just the sequence the car broadcasts itself when you press the fob, on the single-wire bus at OBD pin 1. Send these about 150 ms apart:
0x100 len 0
0x620 01 40 00 00 00 00 00 00
0x621 01 40 00 00 00 00 00 00
0x622 01 40 00 00 00 00 00 00
0x627 01 40 00 00 00 00 00 00
0x62C 01 40 00 00 00 00 00 00
0x62D 01 40 00 00 00 00 00 00
The body network comes up and stays up for roughly half a minute, so the command has to go out inside that window. Sending only 0x100 produces a burst of frames with IDs starting 13FFE0, which looks like a reply but is an announcement. The car is awake when IDs shaped like 10xxxxxx arrive hundreds of times a second.
The forum answer is a high voltage transceiver wake mode. On this board it wedges transmission until you reset, and the six frames above are the whole trick.
Deep sleep is real. If the first attempt produces nothing, repeat it or send a burst of 0x100 on its own.
3. Send the command
Commands are Device Control, service 0xAE, addressed to the body control module at 0x241 on the high speed bus at pins 6 and 14, 500 kbps. The single-wire bus carries status only, which is why injecting there does nothing.
| Function | Frame |
|---|---|
| tester present | 0x241 01 3E 00 00 00 00 00 00 |
| lock all | 0x241 07 AE 01 01 01 00 00 00 |
| driver door unlock | 0x241 07 AE 01 04 04 00 00 00 |
| passenger doors unlock | 0x241 07 AE 01 02 02 00 00 00 |
| park lights on | 0x241 07 AE 07 FF 2F FF 2F FF |
| windows down | 0x241 07 AE 3B FF 01 01 01 01 |
| windows up | 0x241 07 AE 3B FF 02 02 02 02 |
The module answers on 0x641 when it accepts a request and 0x642 when it rejects one, which is worth filtering while you test. The window rows are untested on my car and come from the OVMS source. None of these arm or disarm the alarm.
These frames sent on the single-wire bus do nothing. The firewall theory comes from exactly that. There is no firewall.
4. Precondition, partially
My car uses 0x10390040, one byte, on the single-wire bus, and drives it 00 then 02 then 03 when you hold the fob button. Replaying that and holding 03 powered up the infotainment system but never started the climate fan, so either the sequence is incomplete or the BCM overwrites my value with its own.
The OVMS OnStar path,
0x1024E097, never appears in this car's traffic. I spent time on it before checking.
5. Finding more commands
Log the car in listen-only mode with host timestamps while you press buttons and say out loud what you are doing. Each action shows up as a cluster, and the clusters are unambiguous. That is how I found the wake sequence and how I learned that the frame most of the prior art attributes to remote start is not the one my car sends.
The most linked Volt bus material is derived from a first generation car and never says so at the top.
6. Safety
Never send programming, flash, VIN learn, key learn or immobiliser routines. Locks, windows and climate requests are low stakes, and a twelve volt disconnect recovers a module that has gone strange. Keep the car on a tender so it does not brown out during a session.
Where the useful source is
The OVMS Volt and Ampera module is the only documentation worth reading closely, mostly because it admits uncertainty in comments, and the approach that works is sometimes sitting in a block somebody commented out. The Macchina pin mapping tells you which OBD pins feed which channel, and guessing at that wastes evenings.
github.com/openvehicles/Open-Vehicle-Monitoring-System-3
vehicle/OVMS.V3/components/vehicle_voltampera/src/
vehicle_voltampera.cpp
va_ac_preheat.cpp
docs.macchina.cc/product-documentation/m2-docs/detailed-reference/installation
In the end my car was never protecting itself from me. One missing function call, one reset button, and one bus I had been ignoring.
Verified on a 2016 Chevrolet Volt (Gen 2) with a Macchina M2 running M2RET. Another model year may differ.

No comments:
Post a Comment