Strange behaviour with bench power supply


I experience a strange behaviour of the Jetson Nano when powering it with a bench supply via the barrel jack. We tested two different bench supplies with three different Nano boards. All boards turn off ( ie. the power LED goes off and the screen goes blank ) some time after booting, usually a second after showing the login screen. The bench top supplies are rated 3A or even 5A and there where no other peripherals connected to the Nano. I monitored the power line with an oscilloscope and there was not even the slightest power level drop visible.

When I power the Nano with a really cheap wall wart rated at 2A, it just starts normally and stays on as expected, even though this supply drops to about 4.9V.

Has anybody an idea what could cause this behaviour?

Hi, the wall wart is connected to DC jack too? It still looks like a current limit or voltage drop issue…

Hi Trumany,

Yes, the wall wart is connected to the DC jack.

The current limit of the bench supply is set to max, which would be 3A at least. I would expect to see a drop of the power level when the current limit is reached, but I do not see that.

Does the oscilloscope has enough bandwidth/sampling rate to capture voltage drop? Are you probing the voltage at the input end of DC Jack? Or do you try other cables? Per your description, I have no other suspects except current limit…

The scope has 200 MhZ bandwidth and 1 GS sampling rate and I was measuring at the DC Jack directly. We will continue to investigate.

Will the brown-out detection of the Nano turn off the board, instead of a reset? The LED turns off but the board still draws 0.6 amps, is this expected for a brown out?

The DC jack has a reverse connection protection diode, so you need about 5.3V for that jack. You also need to install the barrel jack shunt to short the pin header to enable it (but presumably you’ve already done this.)

Another option is to provide 5V straight on the 5V rail through the 40-pin GPIO header. Pin 2 is GND, Pin 4 and 6 are +5V. (There are more GND pins to additionally connect to if you want to draw > 3A)

I’ve used bench supplies with 5.0-5.1V on the 40-pin header, and it generally works well. When it first booted, it had some start-up problems, which I chalk up to first-boot shenanigans? The draw for a Nano + display + Ethernet is about 1.4A briefly during boot, and then settles between 0.4A and 0.6A depending on the activity of the screen saver I installed on the device.

It’s also worth it making sure that the DIMM connector is firmly connected to the Jetson module; attempt to wiggle the module a bit (with the power off) and firmly push it towards the connector, just to make sure.