Hello.
We use jetson tx2 board with 12V 5A power supply.
It has been work fine for a while, but recently problems have emerged.
It can reboot or shutdown some times.
After that fault I read pmc/pmic controller status from dtb and got:
as i remember 0x40 is watchdog in the PMIC
I think about voltage regulators with high resistance in feedback(100KOhm), but not researched it.
We replace jetson’s with this problem to new from the box and it solve issue.
yes. in our cases after jetson stall(or poweroff according to leds and fan state), after reset in /proc/device-tree/chosen/reset/pmic-reset-reason/register-value we got 0x10 or 0x50 value
After some time of work, the network connection becomes down without any reason. SOC, PCIE Power and PWR leds are off. For make Jetson work again users unplug-plug power cable(we are install resistor for start on power appears).
WayneWWW,
i’m not co-working with florentin.vonfrankenberg.
I have ~15 devices based on TX2 devkit with custom camera board.
Usually new device not have problems. They are work fine over some months before fault first time. Then time between faults decreases from one fault per week to some fails per day. Replacing devkit(not camera board) to new solves issue for device.
Now I have 2 devkits with this problem, but I can not reproduce it on the table.
I not run any tests with pure BSP. What you cat suggest? stress + some CUDA example will okay?
We use custom camera-board (max9286 + ov10640-ov490).
Kernel built from sources with added support of our cameras.
BSP 28.2 and 28.2.1 with installed ROS.
All my tests in lab environment not reproduce issue. May be problem in EMC compability with other devices in usage environment.