Accesibility of the 2 unused connections on LDU Ampseal?

[jon volk]

Are pins 19 and 20 in the ampseal connector accessible at the logic board level? It seems the connector that brings the rest of the signals out is fully populated. I would like to bring out the swd connections to be externally accessible. I had a wifi upload failure and appear to be dead in the water without doing an ST-Link udpate. I'm in a particularly good mood since that means dropping the motor out of the car to access the board and I would like to never have to do that again for this reason.

I assume I can also solder two wires to the ampseal pcb and route those to the appropriate header.

The ideal scenario would be a CAN bootloader, but I dont think that's in the cards near term on this one.

[muehlpower]

I made an extra molex connector on my LDU, where the serial and the programming port are routed out. The wifi module is mounted on a plug and is now outside the housing.

[johu]

Are pins 19 and 20 in the ampseal connector accessible at the logic board level? It seems the connector that brings the rest of the signals out is fully populated. I would like to bring out the swd connections to be externally accessible. I had a wifi upload failure and appear to be dead in the water without doing an ST-Link udpate. I'm in a particularly good mood since that means dropping the motor out of the car to access the board and I would like to never have to do that again for this reason.

I assume I can also solder two wires to the ampseal pcb and route those to the appropriate header.

The ideal scenario would be a CAN bootloader, but I dont think that's in the cards near term on this one.

Whats the exact failure mode? Which version bootloader? There are a few recovery mechanisms.

[jon volk]

Whats the exact failure mode? Which version bootloader? There are a few recovery mechanisms.

Uploaded a .bin as usual and it stuck around 66% for almost a minute. It then completed and said update successful but there were no parameters loading, only the spinning load icon. There was also no closing of the contactors when I tried the key.

Tried uploading a .bin again and this time the wifi connection dropped out when it was at 98%.

Cycled power again then this time once selecting the .bin file, it would say file uploaded, but it does not begin the progress bar of updating the firmware. At this point no number of power cycles or upload attempts will do anything and the parameters still do not load.

[jon volk]

I made an extra molex connector on my LDU, where the serial and the programming port are routed out. The wifi module is mounted on a plug and is now outside the housing.
20210713_182547.jpg

I like that! I didnt think there was enough real estate behind there to add another connector. Do you happen to have the part number handy? That would be a nice addition since I can mod the cover off my spare LDU.

[johu]

Try rewriting the SPIFS with the sketch data uploader via Wifi.
Had a corrupted FS in my VCU just 2 weeks ago. Basically I deleted a file (say stm32_sine.bin) and re-uploaded it with curl. Then noticed it hadn't uploaded (http://192.168.4.1/list). Rewriting the FS fixed it. Maybe helps here, too.

[jon volk]

Just stumbled across this thread.
Will investigate as I am guilty of renaming files often.
viewtopic.php?f=5&t=1756

[jon volk]

Hahah, I see we're on the same page. Looks like I have 7 .bin files stored WIll delete and report back. That would certainly be better than dropping the drive.

[jon volk]

Try rewriting the SPIFS with the sketch data uploader via Wifi.
Had a corrupted FS in my VCU just 2 weeks ago. Basically I deleted a file (say stm32_sine.bin) and re-uploaded it with curl. Then noticed it hadn't uploaded (http://192.168.4.1/list). Rewriting the FS fixed it. Maybe helps here, too.

I removed all the extra .bin files, re-ran /list to confirm deletion. Uploaded the new file, verified it loaded, and still will not enter into a firmware update.

Ill try rewriting the whole FS and see if that works.

[jon volk]

Unless there are any other idea, looks like the drive unit needs to come out. I reloaded all the files and it still will not enter into a firmware update.

I think the most reversible, least hacky way to get SWD signals out of the case is going to be to re-purpose the cruise input, and brake press input by removing R26/R27 and R6, then jumping off the resistor pad closest to the input over to the the SWDATA and SWCLK pins. Im doing brake pressure, cruise and brake input all over CAN now so I can live without those.

[EV_Builder]

There are mods possible so that the wifi adapter can use the SWD signals for an update.

That should be pretty bomb proof.

[johu]

Unless there are any other idea, looks like the drive unit needs to come out. I reloaded all the files and it still will not enter into a firmware update.

That sucks. Would really like to know what the problem is. When you have the drive unit out, can you connect a TTL <> USB adapter and confirm the board is spitting out the "2D"?

There are mods possible so that the wifi adapter can use the SWD signals for an update.

That should be pretty bomb proof.

Yes, dima did that and apparently it worked for him. I could only get it to read out memory but never had any success in writing or even deleting the flash.

[jon volk]

I’ll see if I can find any definitive failure. Hopefully it comes back over SWD and I don’t find myself desoldering a Nucleo board ha. Investigations my spare LDU, the Molex connector only has a 20pin output off the pcb and it’s fully covered in goo so directly accessing any extra pins that way is a headache.

EDIT: Question on the board itself. Is the accel 5V power connected to the full 5v supply net/plane? For example if I set up a spare Molex plug to an STLink with the above mentioned connections to SWDIO, SWCLK, and then feed feed the board 5v and gnd off the STLink to power things up? That way I can just unplug the connector, pop the new programming connector on, flash the board to recover in the event of a similar event, then plug the main connector back in.

[muehlpower]

I’ll see if I can find any definitive failure. Hopefully it comes back over SWD and I don’t find myself desoldering a Nucleo board ha. Investigations my spare LDU, the Molex connector only has a 20pin output off the pcb and it’s fully covered in goo so directly accessing any extra pins that way is a headache.

EDIT: Question on the board itself. Is the accel 5V power connected to the full 5v supply net/plane? For example if I set up a spare Molex plug to an STLink with the above mentioned connections to SWDIO, SWCLK, and then feed feed the board 5v and gnd off the STLink to power things up? That way I can just unplug the connector, pop the new programming connector on, flash the board to recover in the event of a similar event, then plug the main connector back in.

If you look at my picture of the back of the board, you can see that I have put SWDIO, SWCLK, GND and 5V on unused pins of the WiFi module. So you could also supply the board with 5V via my external connection, the 3.3V is generated from the 5V

[jon volk]

So I finally had a chance to drop the drive unit and pull the board. Connected via SWD and did a memory compare to both the previous good running bin and the one I was attempting to flash. There was a mismatch on both. Reflashed via SWD and parameters reappeared on the WiFi. I was also able to flash over WiFi as well. Did the below mods to bring data and clock out to the connector. Now if I get stuck, I can attach gnd, 5v, data, and clock to the stlink with a spare connector.