Flashing fails on system_b

[robbederks]

This is on Windows with Zadig. Couldn't get it to connect at all on Ubuntu (even with the command to unbind)

[incognitojam]

I didn't reproduce this on my Windows laptop... But @andiradulescu mentioned a similar thing happening on mac, when not providing power to the other port. #80 (comment)

[incognitojam]

I can reproduce this consistently with a comma 3X but not with a comma 3. The issue does not occur when power is provided to the OBD-C port. Flashing is only interrupted once it reaches the system partition.

I will add logging for any responses we receive from the device to QDL.js to see if I can get more information.

[andiradulescu]

From what I tested today:

• comma 3 - flashed fine twice - when plugging the cable there is no light from panda
• comma 3X - got the error everytime - when plugging the cable there is a red light from panda

@robbederks any idea why panda is getting power while 2nd usb is connected? is it "a bug or a feature"?

@incognitojam in #95 you say this happened with a C3, also? is there a light from panda?

My assumption is that panda is messing up the flashing process. Plugging power in main port, makes panda flash blue (as it should). And flashing completes successfully.

Please correct me if I got something wrong.

IMG_2907.mov

[robbederks]

If you want to look further into this, you can try modifying the panda firmware to do nothing (or just erase it) to see if that fixes anything.

Seems unlikely that it's related, but this is an easy way to tell for sure

[andiradulescu]

Ok, so I bricked my panda intentionally (return 0; in main.c). This made the panda not appear or blink any led. Was this what you meant? Meanwhile, I fixed it back.

The qdl flashing error persists. Even with panda in this "nothing" state.

Anything else I should try? I still assume that something is drawing current in C3X (while in C3 it doesn't happen). Any ideas what, besides panda?

I would naively ask how to turn panda completely off, but I assume it's not possible. I also don't understand why panda is getting power via the 2nd usb port.

Thanks!

[dzid26]

any idea why panda is getting power while 2nd usb is connected?

On C3X, the 4.2V output from SOM backfeeds into 5V rail and is enough to be stepped down to 1.8V to power up the panda.

is it "a bug or a feature"?

Could be unrelated, but it looks like the 4.2V from the SOM is browning out periodically after 2s and resets the panda - this coincides with the LED pulses. @robbederks Is this known behaviour? I wonder what causes it.

[robbederks]

The backfeeding is somewhat undefined, but with no firmware running on the panda (except for the bootstub which shouldn't do much) this pretty much proves that it isn't related to the flashing bug.

What might be more relevant is the fact that the 3X likely has a LightningHard SOM and the 3 is Thundercomm based, with differences in the UFS etc.

I'll see if I can find some time to briefly look into this.

Anything useful in the serial output from the SOM while flashing?

[andiradulescu]

I doubt we can get any logs since panda is resetting, thus no som_debug.sh. - I will test tomorrow. There is a possibility that C3X with Thundercomm might've worked fine with one cable, but I think the whole point is to get this fixed somehow for the current revisions / latest product.

Otherwise flashing a C3x will need two cables... #113