Hi all,

Could you please list a few possible reasons for OAK-SoM-Pro module failure and how I can test for them? I've had two separate modules fail on me.

Things started off with intermittent detection via USB from a Raspberry Pi CM4 on a custom baseboard. I later figured out that there was a significant voltage drop on the 5V rail delivered to the baseboard and VPU, causing undervoltage warnings on the CM4.

To compensate for this, I increased the voltage to 5.2V (still well below the max for he OAK-SoM-Pro and CM4).

Upon testing again, I noticed the OAK-SoM-Pro PGOOD LED was no longer burning and the module started drawing a significant amount of current (~3.6W with only the OAK-SoM-Pro plugged in), with the Movidius chip getting pretty hot.

Is it possible that the unstable power might have corrupted the firmware on the device somehow? If so, how could I go about checking if this is the case?

Note: On one of the failed devices, the PGOOD does get asserted. So there is at least some life remaining in it.

    MernokAdriaan
    I recall we had issues with the USB bridge on the OAK-CM4 which caused the SOM to stop working (still working, just the USB bridge was dead). Can you check that?

    Thanks
    Jaka

        jakaskerl

        If you mean to say the OAK-SoM-Pro has a USB bridge onboard, I'm not sure how I would test this.

        If you are referring to the USB hub on the OAK-CM4, my custom baseboard has no USB hub, so this is, unfortunately, not the case.

        Other OAK SoMs work with the same CM4, so also not the USB interface on the CM4 that is the issue.

        Note my comment on the OAK SoM pulling a lot of power and getting quite hot.

            MernokAdriaan
            It was the PCI-USB bridge on the CM4 POE baseboard. But not applicable in this case.

            Do you have another board (FFC-4P or similar) to test if SOM is the issue? I don't see how unstable power would corrupt the firmware, but perhaps the device could be in an unbootable state. Module RST/powercycle should clear that out. You can try re-flashing the bootloader when you boot into recovery boot mode if that works.

            Thanks,
            Jaka

                jakaskerl

                Unfortunately, no, not at this time.

                I have, however, tested the two modules mentioned in various baseboards in which known-good modules seem to work without any issues.

                What configuration of boot resistors is required to boot into recovery mode? The SoM's datasheet does not seem to provide this information.