Cool. Thanks for posting this.

Does the interference only occur on a particular year of car?

Does slow init work any better?

Here are some more traces of communication, this time started with slowinit.
For all logs, channel 0 is at the OBD connector side and channel 1 is at the ECU's side of the K-line.
Several effects of the instrument cluster are visible in the traces:
- disturbance of the connection setup, making necessary a second connect attempt.
- disturbances during running communication session (single bytes disturbed, no
   further data passed to ecu).

There are certain baud rates that are unreachable to the c167 running at 24MHz. I investigated this back when I was writing my eeprom programmer and found that in boot mode, at 38400bps, the BR deviation is greater thar 2.5% which is the maximun allowed for a reliable communication.
Using the virtual com port I'm limited to use one of the UART's standard baud rates, but the FTDI libraries allows a much larger number of baud rates to choose from. I wonder at how much higher than 56700bps you can establish communication, that speed doesn't make sense to write the 95040 but would be nice to squeeze a few more sps from the data logger or the flash sw.

Premium features are great!! flashing takes 52s with the fast flash. I'll give v1.9.2 a try soon, I'm still using v1.9.1.2.

I can just say fastinit never works in my configuration due to instrument cluster.
Slowinit 0x11 works most times, but communication then is really unstable.
What you can see in the logs is happening always when you have some gaps more than a few 100ms in communication.
No way to have e.g. 2seconds idle, then send testerPresent.
The last logfile of slowinit shows what also often happens with communication - cluster just stops transmitting to ecu.

All these things happen independent of a certain baudrate, so neither staying at 10400 nor going down to 9600 or up
to whatever rate helps in this case.

Only way to have stable communication with the cluster in the way is after double slowinit to 0x01.
A slowinit 0x01 seems to tell the cluster to be in pass-through-mode.

Thanks, now I'm running me7logger at 125kbps. Is there any way to reduce driver's latency? That's my problem now, going from 57.6k to 125k only gives 8% more sps with 20 bytes message length, but change is noticeable with longer messages, with 249 byte msg len I get a 50% improvement.

Increasing the baud rate speeds up data transmission speed, but it does not decrease time between messages. The biggest bang for your buck with high data rates, is when you send large data packets. To decrease the time between messages, you need to get the ECU to do less work, and you need send messages to the ECU with the minimum allowed time between responses.