Wouldn't boot -- it wasn't the sd card's fault

So, after much heartache trying to make an “acceptable” sd card, I’ve found out it wasn’t the SD card’s fault.

What I saw:

During boot the Nvidia logo would appear and then everything would go dark. No obvious activity of any kind. I could not ping the enet; there was no connection light on the enet. Nothing.

I bought a new jetson nano, which behaved the same way out of the box. So either both are bricked or there is something else.

It turns out that my – till now – fabulous brilliant and utterly wonderful in all ways 5V supply was not putting out sufficient voltage. I still have to find out why. A 2nd supply was putting out 5.34V (as measured from the GPIO pins) and that worked. My original supply appears to be putting out 5.04V. My surmise is that I have finally plugged in too many devices to my 5V supply, though proving that will take some doing (to position the power meter).

Hi, it’s better to measure the 5V on board to make sure if the voltage is correct. Also too many devices will really draw too much current and so can’t power on. If possible, a oscilloscope is best to observe if any voltage drop on 5V input.

Wasn’t the power supply either.

The beloved power supply now has a power meter attached. The beloved power meter is putting out 5.29 +/- .02 V like a champion. However, the board will still not boot using the beloved power supply.

It boots just fine with a “stock” power supply. I measured the voltage from the GPIO pins, and it’s real darn close to 4.96 V. So my guess is the heroic wonderful power supply is putting out slightly too much V to make the board’s power supervisor chip happy, and it’s locking the board down.

So when I get a chance I will adjust downward and I think all will be well. The issues with this theory that I can see: (1) the GPIO pins are lying-- they’re telling me what the board has decided to put on the pins, (2) the “stock” power supply is putting out something more than the label says – I don’t have a graceful way to measure that directly right now.

It looks like power supply capability issue. Better to use oscilloscope to probe the input voltage on board to check if any deep drop when power on.