I have a Jetson Orin Nano (bought new from Amazon in August 2025). Since then I have bought 17 of these for use with a course in next few weeks. I just started using the GPIO on my Jetson Orin Nano yesterday for the first time (I worked with Jetson Nano GPIO last year in summer before transitioning to Jetson Orin Nano). In my quick tests, I found that while the physical pins 2 and 4 (5V rail) provide 5V measurement on a multimeter (with ground on physical pin 9), in all likelihood that 5V can not drive even any tiny load.
When I connect a 1K (or 10K, 100K resistor across GND and 5V and measure across Gnd and 5V pins, I get 0.06V (or 0.56V, 2.86V, respectively). Also, if I connect one end of a 1K (or 2K, 10K) to ground and place the other end to one probe of a multimeter and other probe of multimeter to 5V, I measure 0.6mA (or 0.6mA to 0.7mA) current. All these tests suggest that the 5V rail is not able to deliver any power whatsoever to light up even an LED. So instead of using 5V from pin2 or pin 4, I am having to use an external 5V supply to toggle LED with PWM32/PWM33. I know I have to use external 5V supply for servos and such, but it is problematic that even LEDs will not light up (LED with 470ohm in series).
Can someone from Tech Support help or perhaps the board was already in this defective state and can be replaced? Any help will be appreciated. Many thnx.
There are several third party carrier board vendors, and those ship with the SD card slot on the carrier board. The developer’s kit has the SD card on the bottom of the module itself. Can you verify where the SD card is? If it is on the bottom of the module, which is a developer’s kit, then there also won’t be any eMMC. This is the model which supports the NVIDIA software (third party versions are similar, but they’d have some minor changes and you’d have to note them when working on a Jetson). This can then be compared to a carrier board or schematic.
This particular forum is for the original/older Jetson Nano. If this is a Jetson Orin Nano, then it goes with this forum:
https://forums.developer.nvidia.com/c/robotics-edge-computing/jetson-systems/jetson-orin-nano/632
I’m sure you can get help here in this forum, but it might take longer in the wrong forum, and might result in people looking at the wrong schematic.
Actual documentation, including things like carrier board information, is here (you’d have to enter the correct model):
https://developer.nvidia.com/embedded/downloads
Note that a 5V rail is different than a GPIO, and I do not know what the actual 5V rail makes available for current. The GPIO though goes through a converter to change internal 1.8V to 3.3V, and the available current requires buffering. You’d never be able to power an LED, and probably TTL logic, directly off of GPIO (not in all cases, but most cases).
If you’ve identified the specific model, and if you can describe which header you are speaking of, then this would be easier to answer (there are multiple headers).
I have NVIDIA Jetson Orin Nano Developer Kit. The SD Card is on the bottom of the Jetson Orin Nano Core Module. Here is the model number from the original box: P3766. From the box, the part # is: 945-13766-0000-00.
The 5V that I am referring to is from physical pins 2 and 4 on the 40-pin GPIO. I believe one should be able to drive a small LED (in series with 470 Ohm resistor) connected between the 5V (Pin 2 or 4) and Gnd (Pin 9). Once again, these pin numbers refer to physical pin numbers on the 40-pin GPIO. Usually, I should be able to connect an LED like this: Pin 2 of GPIO (5V) connected to one pin of 470 Ohm; other pin of 470 Ohm connected to the Anode of LED; Cathode of LED connected to Collector of 2N222 BJT (all a series connection). Then the Emitter of 2N222 goes to Pin 9 of GPIO (GND). Finally, Base of 2N222 is connected with a 10K resistor to GPIO Pin 32 (or Pin33) configured as PWM. Then under program control the LED can be controlled to brighten and dim. I have similar things done with Jetson Nano (different 5V pin numbers and Gnd pin numbers on 40 pin GPIO, Gnd was Pin 6 and 5V Pin 2). Currently this does not work on Jetson Orin Nano Developer kit, I had to put 5V connection on an external supply and then do common ground with that supply for GPIO Pin 9 of Jetson Orin Nano Developer Kit. I saw on forums others complaining of same issue and someone from Tech support indicated that likely a MOSFET that provided the output for GPIO pins 2 and 4 was dead and simply outputting 5v without any way to provide current. Thnx for your help. Are you able to move the discussion to more relevant forum or should I repost?
For reference, documentation on that hardware can be found here:
https://developer.nvidia.com/embedded/downloads#?tx=$product,jetson_orin_nano
If you go to that above URL, and then filter for the word “schematic”, then you should be able to find “Jetson Orin Nano Developer Kit Carrier Board Reference Design Files”. Note that this is named after the carrier board model, not the module number you gave above. Page 23 lists the 40-pin GPIO header. This has symbol VDD_5V_GPIO for pins 2 and 4.
I can’t say for certain, but this does have protection circuitry to block reverse power flow there. I don’t know how much current this is supposed to make available. Someone from NVIDIA would be needed to answer that.
However, because the actual power source, without the intervening circuitry is VDD_5V_SYS, there would be no chance of the intervening circuitry blocking output if you go to the 12-pin button header, pin 2. See page 24 of that document. Would you be able to switch to using that header for your power output?
If VDD_5V_GPIO (pins 2 and 4 of the 40-pin header) have insufficient power, and if your circuitry is correct, then I wonder if your design is triggering the back power blocking. A purely passive resistor and LED would not do that, so either there is a circuit failure (for this case of purely passive power draw), or it is just more than allowed, and I cannot answer which case this is.
I have almost nothing on the 40pin GPIO header when I do the test for Pin2/Pin4 5V. So I am quite certain that this is a failed product. I have been using 5V now from external supply and have been able to do digital inputs, outputs, pwms, etc., by doing custom .dts and .dtbo. But the 5V does not want to play from Pin2/Pin4. I would like nVIDIA to provide a replacement since the very first time I tested the 40 pin header the 5V did not behave as it should.
Someone from NVIDIA would have to look at that point on the schematic and see if the protection circuitry there allows using those particular pins are intended to provide that much current (maybe @WayneWWW can answer that). My thought though is that any kind of GPIO on a Jetson has very little current output capability, and the power rails themselves might have different limitations depending on where you are pulling that 5V from.
If that location is intended to have enough power for an L.E.D., then that seems to be an obvious failure, but perhaps that location is not designed for that much output current.
Hi botz.mecha,
Are you using the devkit or custom board for Orin Nano?
How do you power the board? with adapter?
As you have bought 17 boards, is the issue happening on each of them?
Hello KevinFFF: Sorry for the delayed response. I have been preoccupied. I am using Jetson Orin Nano Development Kit, not any custom kits. The board is powered using the power supply that came with the development kit. This is experienced on the first board so I told other users not to use the 5V on the remaining kits.
Thnx
Please get another devkit and check if there’s similar issue.
I don’t think 5V from PIN2/PIN4 can not power on a LED. There may be HW damage on your devkit.
In fact this answer is incorrect. See Jetson-Orin-Nano-DevKit-Carrier-Board-Specification_SP-11324-001_v1.3.pdf pp. 26, Table 3.3 (see snippet below). P2 and P4 should be able to provide up to an Amp. I have yet to connect any LED, I simply did voltage and amp tests that I provided above. Even I/O pins are listed to have current capability of +/- 20 uA. P2 and P4 spec. are 1A but I do not get it to go beyond 0.6mA as reported earlier. I have used power pins on the 40-pin header of Jetson Nano Carrier board for all kinds of experimentation without any issue. This carrier board should have worked in same manner based on documentation. Any way, NVIDIA is handling RMA now so we shall check it out.
Okay, please perform RMA if the issue is specific to this board.
Thnx Kevin. Yes, that RMA is proceeding slowly. I have another issue. One new Jetson Orin Nano Developer Kit is not turning on new out of the box. An SD card was prepared for Jetpack 6.2.1 and installed on the kit. When the NVIDIA-supplied power supply is used on the barrel connection and turned on, there is no activity on the kit, no LED, no fan. Can you suggest what debugging steps should be followed? Many thnx.
Please try using the SDK manager to flash the devkit with SD installed.
You can also check if there’s any serial console log output when you plug the power or it can be recognized in force recovery state.
So unfortunately with the NVIDA Power Supply plugged in, the LED next to USB-C port does not turn on and the fan does not turn on (usual power-on behavior). I checked the voltages at the 40-pin GPIO header using P9 for Gnd. I found that both P2 and P4 show 5V. However, when I check for 3.3V on Pin P1 or P3, I do not get 3.3V. On P1, I get 2.1V and on P3 I get 2.8V. I also removed the SD card and still the same behavior. Then I removed the Jetson Orin Nano Module from the Carrier Board, just in case if it was not sitting in SODIMM connector properly. Without the module connected to Carrier board, 5V is seen across P2-P9 but there is no output on P1 or P3. Based on my online searches, the 5V to 3.3V regulator is on carrier board, so unclear why this behavior. After reinstalling the module, still could not get 3.3V on P1 [or P2]. I also did forced recovery mode by putting a jumper across Pins 9 and 10 on J14 header. Connecting on USB C header also does not do anything. Basically, without steady 3.3V, I believe the module will not turn on. This is a dead on arrival Jeston Orin Nano Developer kit – very disappointing. I have bought and used over 200 R-Pis since 2014 without encountering a single dead on arrival unit. Out of 18 or so Jetson Orin Nano Developer Kits I bought in last 8 months (16 in last two months) – now two have showed out of the box problems. Any way, I had a spare new kit that we opened and installed the SD card that we had prepared and tried on the dead on arrival kit. The spare new kit powered up and the LED next to USB C turned on, fan turned on, and the boot up started without any issue bringing to the initial setup screen. See enclosed images of faulty 3.3V pins. I would like NVIDIA to replace this device and have some reasonable quality control. It is quite disappointing to receive faulty units that delay teaching and learning for classroom and such bad experience will leave future customers with negative experience about NVIDIA.
Is there any serial console log output at this moment?
Incidentally, the serial console uses a level shifter from 1.8V internal to 3.3V on the header pins. If serial console works, then the 3.3V rail has only partially failed.
Unfortunately there is noting on serial console. I now have same behavior right out of the box on 3 kits out of 16 that we bought couple of months ago and gave out to students two weeks ago. The SD cards worked right away when I gave to those teams some spare kits I had. Replacement units turned on, LED was on, fan was on, and initial setup screens for ubuntu started. Three dead out of 18 is not a positive experience.
Actually, I mis-spoke earlier. On GPIO 40-pin connector, P3 is not a 3.3V pin. The 3.3V is on P1 and P17. In one kit, both of those give about 2V and two kits, just 0V. I just realized that I must have mistyped. Thnx.
It isn’t likely, but sometimes there will be a revision level change to the carrier board which could affect this. It is unlikely this is the cause, but it is possible, especially if the device tree has to change with revision level. The default Jetson flash software would normally just do the right thing with this though, and not all revision changes have to change device tree, so I say that is very low in probability (but it isn’t zero).
Can you verify your serial console hardware/software combination works with some of the dev kits, but not all?
Yes, PIN1 and PIN17 of J12 are 3.3V output.
However, PIN2 and PIN4 of J12 are 5V output.
From Jetson Orin Nano Super: 5V rail not working as expected - #7 by botz.mecha, it seems you have submitted the RMA for these problematic devkit boards.
