High failure rate of NX Devkit carrier board

3 out of 5 NX Devkit carrier boards have failed now. In each case, they’re essentially presenting a dead short to the 19v power supply. Anyone else?

Nothing special about the setups. No USB peripherals, nothing connected to the 40 pin header, using the supplied 19v power supply. Each time though, the module was doing heavy compiles using all 6 cores at 100% and with the fan spinning at 100%. I’m wondering of the carrier board input regulators are shorting out from the load. The modules themselves are fine. I can swap them to one of the two remaining carrier boards with no issues. Unfortunately though, I now have 3 perfectly good NX modules without anything to put them on.

The carrier board specification lists the input regulators as TPS53015 for both 5v and 3.3v but the only info I can find on them says they’re rated for only 120ma so that can’t be right. Anyone know that they really are?

hello gtj,

are you running a use-case with heavy GPU loads also?
please check Topic 173571 for reference, this is a reproducible issue with 5A OC limit on Xavier-NX.

There was no gpu activity at all and no OC was involved. The power model was “2 - MODE_15W_6CORE”.

hello gtj,

we may have a test locally, by using CPU stress utility to make CPU fully loaded.
may I know the software release version you’re working with? thanks

Currently running L4T 4.5.0 with the accompanying 4.9.201-tegra kernel but this also happened with L4T 4.3. Was doing a make -j6 of the geany editor at the time.

The output current of TPS53015 could be much higher than 120mA which can be 10A or more. Not sure why your boards failed, if it is boar defect you can run RMA for them to find out root cause.

Thanks @Trumany. I didn’t actually find the datasheet for the TPS53015 so the 120ma might have been just me reading something wrong. I’ll do RMAs, thanks.