I just observed that when un- and replugging the 12V DC plug on the board, the Jetson TK1 starts to boot. But when un- and replugging the wall outlet, the Jetson does not boot up until the power button on the board is pressed. But when leaving the wall outlet unplugged for > 1min and then replugging, it comes up again. This is the same for both of our devices.
Is this expected behaviour, and if not, can there be anything done about this?
It’s not working “perfectly” for me either.
I missed your goal though, so how would you like it to behave?
There are pins for the power and reset and so they can be controlled with external hardware, if needed.
tldr: when unplugging and plugging in the power, we expect the tk1 to turn on.
in our setup the TK1s are all powered from a switchable poweroutlet and if someone switches off the power (intentionally or by mistake) and later on again we expect them to turn on again. We have set them up in a way where pushing a power on button is not an option (they are sitting in a remote place).
Since that works fine when unplugging the 12V DC plug but not when unplugging the 230V plug we assume something is not working correctly with the power good circuit on the board or the power supply.
in a x86 PCs BIOS there is a power failure state settings or power return setting, where you can switch how the pc should behave when the power returns -> run or stay off. is there any such option in the TK1?
I think whether or not the Jetson will turn on after plugging the power depends on how long the board itself was without power. If you disconnect it from the wall, the adapter will give still some power for awhile but when you disconnect the jack from the board, the power is complete stopped.
Afaik, there is no proper way to make it behave consistently. The only approach is to use something external to power it on using the header pins.
I’d say it is expected behavior. There are capacitors and power does not drop completely or instantly if the power supply is still connected to the Jetson, despite power being pulled from the wall. A desktop x86 machine might pull 20 or 100 watts even under lower power conditions…yanking the power when running on x86 would consume so much power compared to any capacitors that internally things would go completely down quite fast. Your Jetson doesn’t consume that much power even when running full…when idle those capactitors should have some life in them. If you really want something which consumes so little power to reboot (which it does when power has gone down completely…including capacitors), you should build a power lost detection and wire it to the reset pins (simply because power didn’t really reset completely in the time given).
As a side note, yanking the power from the wall means yanking the transformer part from the wall…it’s an inductive load. I’m not sure what kind of built in surge protection the power unit has, but I doubt it has much (if any). This is risky to the Jetson’s life.
This should help.
- Remove the C6D4 part (0.014F capacitor on VBAT_BKUP pin for RTC backup power) in the Jetson TK1 schematic if you have a soldering iron to try.
This should help PMIC in Jetson board to recognize power off status much quickly when the wall outlet is unplugged, so replugging the wall outlet will trigger power on again.
Tested and working. Thank you youngk.
To remove the Capacitor, I recommand using pliers and only using mechanical force, which works well here. I killed one TK1 board trying to unsolder the capacitor using either reworking station or soldering iron…
Are there any non destructive bypasses – perhaps done via software?
Hello tzaitsev, thanks for your inquiry. The only workaround is in hardware, if you are planning to create a product based on the TK1 you could integrate this workaround into the product.