Jetson Nano not Turning On

My board is dead too. The green light comes on but there is nothing on the screen. I have tried two different power supplies so far. I will go through all the steps including an effort to monitor serial debugging info with a usb/ftdi adapter.

This the the fourth board in a row that Nvidia has shipped to us that does not work. We even took it down to Best Buy and tried to boot on a brand new monitor hoping it was a content protection/hdmi encryption issue.

@vreppeto,
You need to make sure that you are reading the instructions correctly. Are you using USB or DC Barrel jack input? Because those have different jumper settings. How are you turning on your board? Momentary switch? Have you double checked that the pins from your switch are actually shorting when the switch is closed? What about your power supply? Is says 5V on it, but have you confirmed with a DMM? Next, your HDMI cable – do you know that it is a properly functioning cable? How about the SD card? Can you swap that out and load a fresh image?
Okay – let say all of those things are in order – and everything you’re doing is a-okay. Then the next step is to monitor the UART during bootup. Make sure that the card is seated and the jetson boots.
If it doesn’t, you have to start looking at the possibility that something in your setup is damaging the boards.
And sometime after that, you have to start looking at the supply chain. It’s unlikely that NVidia shipped you four bad boards. But not impossible.
I do have a working theory that the carrier board was never put through a proper DFM analysis prior to manufacture. That could have left a variety of potential latent defects that are brought to light if stored in high-temperature or highly-variable temperature storage environments. But my first guess is that there’s some tiny little thing that you’re doing wrong.

We made the mistake of expecting something to display on the monitor to appear even if the sd card was not inserted. This is a false assumption.

I purchased a new TTL/USB adapter and I am observing a boot log. The old TTL adapter was apparently no longer working. This device is attempting to boot but without an image on the sd card nothing appears on the monitor.

Previously I used DD to copy the image to an sd card. However, the boot log still reports a failure to see the formatted partition, and even though the board is attempting to boot the monitor remains blank. The boot log ends with:
[0000.642] Invalid GPT Partition
[0000.832] Using BFS PT to query partitions

The instructions do not mention that the monitor will remain blank until a kernel image is successfully loaded. Nor did I find such a remark in the extensive research I performed. Questions directed to tech support on a chat session regarding these issues were ignored. They only wanted to know if I had performed basic setup procedures before they proceeded with an RMA.

All the concerns you have listed have been taken into consideration. We have several HDMI cables, power supplies and monitors we have tried. We are now using a battery and a 5v linear regulator as a power supply to guarantee the cleanest possible signal. We have another with built in OSD and yes we tested with a DMM and an oscilloscope.

I have new sd card I will try tomorrow with a fresh image burned via Etcher on Ubuntu 18.04. I will not be using DD this time.

Even though it is not mentioned in the setup instructions somewhere I read that a extFAT partition should be created before the image is flashed. This makes no sense to me but since no partition type is specified I will try it if the normal instructions fail.

If you use a uSD card 64Gb or greater you will need to format it with ExFat.

Ex:

$sudo apt-get install exfat-fuse exfat-utils
$sudo mkfs.exfat -n sdxc64 /dev/sdh  
mkexfatfs 1.2.3  
Creating... done.  
Flushing... done.  
File system created successfully.

Launch Etcher
You might have to select Select Drive to get to the uSD card.

Once the uSD is burnt, then run GParted and select Fix to clean the uSD card from the Etcher write.

When this is done, close the program and eject the uSD card, remove it from the PC and then place it in the Jetson Nano again.

During boot, try hitting the ‘Esc’ key while the board boots to see if it will show the boot sequence.

Yeah – a missing / improperly loaded card will stop you dead in your tracks.

I’m somewhat surprised that DD didn’t do the trick – probably some malformed/missing flag/option. Anyways – I think you’re on the right track. I hope you have a working Nano sometime tomorrow!

Is this thread still hot? I caught up with my busted board at a PCB assembly shop in Colorado. I followed it through optical and 5-axis x-ray inspection. I accidentally left the USB drive behind, but it was mailed and should catch up with me sometime this week.

Here’s the thing, all of the issues found could have been picked up on the optical inspection, and before AOI, they should have been picked up during ERC or DFM. There was conductive detritus captured between the pins of the Jetson Nano Computer and the Jetson Nano Carrier Board connector. Whether it was copper from the etching bath or confetti from somewhere else, that was the problem.

Now, you should look at the outer-layer gerbers and you’ll see <1 mil net spacing. That’s a no-bueno. Anyhow, if anyone is still paying attention, my USB drive should be here within a week. I can give images then.

Hi mjhughes, sure, please share by then.

Sorry, it took this long for the USB drive to finally catch up. It literally finally came today (after going through quite a few hoops in the lost mail office at USPS)


This is IC U27 – IPC allows up to 50% voiding – so these are passable. But it’s a good example of why you should fill/cap vias or at the very least retract the paste layer from near the vias. There’s a good chance that this board has low production yields that can be attributed to voiding. Here’s a recent blog post on the issue. https://www.royalcircuits.com/blog/all-about-solder/


I’ve posted this one before – but the annular rings are way, way too close to the adjacent trace. If your fabricator doesn’t have their etch tanks dialed in to within a “nuthair”, those will fail to separate during etch, decreasing your yield. Or worse, those traces are close enough for electromigration or a high-resistance short later on. That should not have passed DFM checks.

The detrius picture is focus-blurred beyond recognition – but I still have the board and the detrius is still there. So I’ll try to find a way to take a photo of that for you sometime this week.

Also – I distinctly remember recording videos – but I don’t see them on my flash drive – I’ll get in touch and see if they got saved on the computer instead.

I bought jetson nano , just with usb-c type power connector .
I use 5V, 3.5A power supply for the board , 32G SD card , imaged by windows 10 machine , follow the steps .
the board was power up, green led was on , the NVIDIA logo was on the display, but it stuck there , never booting
show : invalid format for 1 seconds , then went to dark screen, then show logo , repeating…

what is the problem? is the image format wrong ?

1 Like

Hi sunnywang2003,

Please help to open a new topic for your issue. Thanks