NVRAM getting scrambled when driving

[BennVenn]

something you could try if you're handy with a soldering iron - re-write the tune to that nvram, then solder a wire from the dallas WE pin to the VCC pin. That will absolutely guarantee it's not a NVRAM failure and is being caused by EMI or something external. If it doesn't corrupt, you can be almost certain its the nvram pcb

[antus]

@someguy360 I have boards and parts, I could probably hand solder one together when I can find the time. Not too many components on that one. Gareth likely has a few but hes also a busy guy. I'll PM you.

[antus]

An update to this story. It turned out someguy360 was local to me, so we agreed to meet up and look at the board. First thing noticed was that oseplugin didnt connect to it as quickly as normal. Then when I started bench logging the first 30 or so packets errored, and then the comms became stable. Not normal. So we had a bit of a play around with it, things seemed stable enough even though connection wasnt right. So I tapped on the nvram board and after a few heavy taps while logging the comms suddenly comms stopped. I rebooted the ecu and couldnt get a connection. So we pulled the nvram IC out and read it. The chip wasnt corrupted all over, but it had scattered bytes here and there corrupted versus the original read. The board looked fine by eye, good solder everywhere it should be etc, but I suspect the header that plugs in to the ecm socket had a slightly loose connection on one of the address pins, hence random looking data written in random looking places under high vibration. I had another board ready so we swapped over, and repeated the test. The new board connected immediately (like normal) and no error data packets. Even tapping it twice as hard as the first one while logging and everything stayed good. So, old board is in the trash and it looks like we have a solution. This is not a common fault (in fact its the first time ive seen it) and im sorry that happened but hopefully its all good now.

[immortality]

Good news

[someguy360]

Big thanks to Antus for taking time out of your weekend to troubleshoot with me.

At least now I know I'm not going insane and it's not a fault with the car.

I'll try and get it out for a drive at some stage next week once the weathers cleared up and report back but I have full confidence that you found the issue as at least we were able to replicate it on the bench.

[In-Tech]

Great job guys, I'm not familiar with the NVRAM but if it uses a memcal adapter type "thingy", I have seen those fail from Hypertech and JET many times.