First off, awesome job Tony! I just tuned my Allroad and your flashing tool is extremely handy, I do have some issues though.

My car is 2001 Allroad. I have to ECUs at this time, one original from 2001 Allroad and one from 2002 A6. They both flash ok on the bench without a single issue ever.

Every time I try to flash in the car though, I am having extremely hard time pushing the flash through successfully (that's with K module fuse out). I believe that there is some issue with Radio interfering with the flash albeit this is just a theory based on the fact that sometimes when transfer visibly stalls the radio display flashes really quick when it is off, and pops up "DIAG" message if the radio happens to be on. Now this is issue is not NEfmoto Flasher specific at all, I used to have Maestro and the same issues were plaguing it as well although their software was re-trying each failed block so 99% of time the flash was successful after all in one short session.

With Nefmoto, for one successful flash I have about 20-30 failed ones anywhere from first to last flashable block (60%ish). I just spent last 3h trying to push a file through and I am still failing so I am taking a break. I am at work and can't bench load it, luckily I have the second ECU with previous version of tune so no big deal swapping it in.

I attached a log from failed session. The important observation of mine is that after failing of the flash, if I don't disconnect from ECU, I can re-try the flash again... almost unlimited times at that.

My request would be to add automatic block re-try on error. Given that the session is still open it should not be a biggie? This would help me and other Allroad (I believe the issue is Allroad specific) owners/user.

Now if the premium version of your software was available, it would also help me somewhat as there would be much less to flash and fail on, but the re-try would help everyone.


Thanks in advance!

Hi Tony, the log file is attached. I forgot to attach it initially and uploaded it about 10 minutes after the original post time.

I noticed that I might have had chosen wrong chip type, I had 29F800 instead of 29F800BB... I chose the new layout and failed only twice, third time I finally flashed ok. Maybe proper layout made the error recovery you mentioned work properly.


Thanks.

I understand.

I am not asking to do miracles and figure out the impossible (what was received and what wasn't) but re-trying the current block/sector starting with flash erase for current address range should be sufficient and would do the trick?

I figured something is interfering with transfer as logging on my car is also problematic sometimes due to timeouts etc albeit EcuX and VCDS automatically reconnect so it is not that bothersome only resulting with gaps in data.  Again, this is independent of ECU (two separate ECUs having the same issue in my car) but most likely something else making mess on K-Line. .

Big thanks!

I just wanted to point one thing again. When it aborts like that, I can re-initiate the transfer in the same session immediately without any issues but if I power off the ignition the ECU will brick and I have to pull fuses/disconnet battery for at least 1minute for it to reset and allow new session. If it is possible for you to write any arbitrary memory address in the ECU, as it appears to be, it should be possible to jump to beginning of memory address range that failed and re-try?

Apologies if I am drilling you about something that is not easily doable.

Thanks!

I understand that by saying "turning off ignition" you mean backing the key into accessory position? I always just backed it all the way...

Not sure as I had no need to do that. If I connect with Fast Init the session is active even after the flashing fails. I can immediately re-try flashing whole flash memory without doing anything else than just pressing that one button.

I can experiment with slow init just to have an answer for you next time I try flashing the car.



Re-trying only during current session is all that would be needed. Storing the state and resuming in next session would be an accident waiting to happen and I understand that.


Keep us posted about premium feature, I even got a proper cable to be able to use it

Yup, it's a real pain in the ass. You RequestDownload for a specific address and size, then you TransferData until you finish sending all of the data. The TransferData messages have no address or offset information inside them. Also the TransferDataPositiveResponse have no address, offset, or size in it. So if you fail to send one TransferData message, you have no way of guaranteeing that you are still in synch. If you get out of synch, there is no way to end the transfer of data either. The ECU will reject all requests until it gets all of the data it expects. The way I handle this is by just sending TransferData messages of a 0xFF until the ECU is happy.

The last project I was on I was doing client server networking, and yes, the KWP2000 flashing protocol sucks.

Using the special version I was able to flash my car three times already without a single hiccup. Very nice, thank you!

Log is attached.