hello WayneWWW,
Thank you very much for your suggestions. I believe we have progressed quite a bit in solving this issue. Here is a breakdown of our tests on the devkit carrier:
-
Disable all of the pins unused by our application
Result: Fastboot mode always initiated, debug log stops just between
Kernel DTB loaded at 0x83100000
DeviceTree Init done
and
Pinmux applied successfully -
Disable all of the pins unused by our application, except debug serial terminal
Result: Fastboot mode always initiated, debug log shows
Power button long press detected
Entering fastboot mode..
and after some lines it just repeats
Key Code 8 not handled -
Disable all of the pins unused by our application, except debug serial terminal and BUTTON_POWER_DOWN pin (also enabling its internal pull-up)
Result: Boots successfully on the devkit carrier (no fastboot)
We managed to get the TXD serial pin out of the custom carrier and repeated the third test with 3 production modules, each one (when connected to our custom carrier) presented one of the different behaviors I listed on the initial post.
-
A module that always worked properly
Result: Boots properly.
1_success.txt (28.6 KB) -
A module that booted properly as long as it temperature was low
Result: Boots properly while cool
2_Succes.txt (26.9 KB)
and when booting after heating-up it stopped a bit after
Hit any key to stop autoboot:
2_Fail.txt (19.3 KB) -
A module that never booted properly
Result: Actually managed to boot properly once
3_Succes.txt (32.2 KB)
but on every other attempt it just stopped a bit after
Hit any key to stop autoboot:
3_Fail.txt (18.1 KB)
We found a post regarding the stop autoboot message where the absence of the pull-up resistor on the RXD line caused the issue and believe that might be the case for us as well.
The pinmux use case that causes the devkit to enter fastboot seems to be marking the BUTTON_POWER_DOWN as unused.
Thanks again for the help. We will report back to you once we test the RXD pin hypothesis.