Powering Jetson Nano Through GPIO (J41)

Hi all. I am trying to power the Jetson Nano through the GPIO (J41) on pins 2,4 (5V) and 9,14 (GND). For some reason, the board powers on long enough to show the NVIDIA logo then shuts down completely. Here’s what I’ve tried so far:

  1. Using an anker power bank I’ve powered the Jetson Nano successfully through the uUSB port (also proving that the supply is capable of powering the board). I have tried powering the Jetson from the same anker supply through the GPIO but still the board shuts down after the NVIDIA logo.

  2. Powering the Jetson Nano through the GPIO using a bench power supply. Just like before, I see the logo screen followed by a power down.

I took a few measurements with a multimeter and noticed that when I initially power the board through the GPIO, the 3V3 LDO seems to be doing its job properly. Once the board shuts down even the 3V3 rail on GPIO goes out as well, however, 5V is still present on pins 2,4 on the GPIO.

Is there something fundamentally different between powering the Jetson through the GPIO vs the barrel jack or uUSB? The only difference I see is that we completely bypass the selection circuit on page 11 of the schematic. Is there a signal I’m missing that gets triggered when you go through the uUSB or barrel jack that allows all of the regulators and Jetson to stay on?

Hi, how did you connect the GND of power source to Nano board? Do you have oscilloscope to check the 5V input when power down happen? Each pin of 5V of J41 can be used at up to 2.5A. Better to check that to make sure if it is caused by drew current or voltage droop on power route.

Hi Trumany, I both cases I connected the Jetson Nano carrier board ground (pins 9 and 14 on J41) to the ground of my power supply (the supply is able to provide 5V=6A). I can check the 5V input during power down with a scope in a few hours. It seems strange to me that this would be an issue considering that the Anker power supply is able to power the Jetson carrier board through the uUSB connector but not through the GPIO pins. Could it have something to do with the U58/U67 (TPS25944L) on page 11 of the schematic for the carrier board?

In theory it should have nothing to do with the load switch. Besides the scope measurement, please check the connection quality of power cable to J41 pins and also you can try other shorter cable.

I was able to put a scope on the supply and also monitor the total amount of voltage sag experienced by the supply during the Jetson Nano’s power up.

When the Jetson Nano pulled around 1.1A during startup it was causing my supply to sag down from 5V to around 4.8V. When this happened, the Jetson Nano promptly shut down. One solution was to minimize contact resistance by using multiple pins for GND on J41 and connecting those all to the GND from my Anker power supply. Another solution seemed to be using shorter cables. In other words, reducing resistance introduced by the cable and improving the quality of the connection.

Ultimately what I think I will do will be to use a buck-boost converter to maintain a steady 5V during operation. I will also create a resistor divider network to fake a signal to the Anker power supply on the D+ and D- pins to output the maximum possible current at all times (since I am getting power over USB ports).

The end goal with all of this is to create a sister-board which mounts on J41 (imagine a raspberry-pi shield) in order to interface some sensors and peripherals. My goal with all of this is to power the Jetson Nano through the GPIO and minimize cables in the final design.

I will continue to do some tests and report back on how the boost-buck converter affected powering the jetson through J41. The goal there would be to reliably power the jetson through GPIO regardless of potential voltage sags from any power supply during spikes in current draw.

I would like to hear how this works for you armandop4597. We have created just a shield in house that functions the same way, but we have not had the chance to assemble the components on the PCB yet to verify this power supply configuration. After testing your new configuration please let me know how it does and I’ll send you what we came up with once hour setup has been tested.