I have a broken TX2 8GB WiFi module on my desk right now that I would like to fix. Without schematics that’s pretty hard, but maybe someone from nvidia can help me here.
The module is pretty much dead. It turns on, bot doesn’t boot. It does show up in recovery mode though, boots a recovery MB1, but then fails when downloading the next stage and verifying its checksum:
[0164.557] W> Rail ID 7 is not found on MB1 BCT [0164.561] W> MB1 PMIC CFG is version 1.2 [0164.568] W> MB1 PMIC CFG is version 1.2 [0164.575] W> MB1 PMIC CFG is version 1.2 [0164.579] I> sdram slot : 1 [0164.581] I> Initializing SDRAM [0164.588] I> mss encrypt status check successful [0164.592] I> downloaded image bct_mb1 successfully [0165.666] I> Downloading blob at 0x0000000085300000 [0165.785] E> Checksum verification failed. [0165.789] E> Sending nack 0x0b85c11c [0165.793] E> Failed to receive data [0165.796] E> (0b85c11c): Failed to receive data from host
Most of the power rails are present. VDD_SYS_SOC is definitely there with 0.95V, otherwise the BPMP wouldn’t work and it clearly does. 1.1V for the LPDDR4 is also present. DRAM gets initialized by MB1, I just don’t know if it actually tests the memory after initializing it, but it seems to be happy according to the logs - although the checksum error could indicate memory init problems. The eMMC gets initialized according to the boot log as well. The MAX77620 spits out a handful of voltages that look correct.
What doesn’t work are the other two beefy regulators: VDD_SYS_CPU and VDD_SYS_GPU are dead. However, they are not dead because they are shorted. I checked all the rails and their resistances look okay and they aren’t shorted.
I have the feeling that there is an error in the power sequencing somewhere. Is it normal in the recovery mode for these two regulators to be turned off? Which other part enables these regulators? Is it the Parker SoC itself or the MAX77620 PMIC? Are the PG signals of these regulators just chained together?
Are these two regulators switched on later by the BPMP firmware? Any reason why the regulators would be off in this state of the SOM? Maybe this is even normal?
Any help debugging this would be cool, because I’d really like to fix this module instead of just buying another one. It’s a shame that there are no schematics for the module available :(