Crazy loud noise fan early before NVIDIA logo display

Robotics & Edge Computing Jetson & Embedded Systems Jetson Orin Nano
1

JetPack 5.1.4 L4T35.6

It should be OK with fresh new BSP as it was ocurred recently. And I think it might be related with software updates, but I don’t know what triggered.

The situation is quite similiar: at the moment of power-on and will end within 2 to 5 seconds.

There is no solution above, I want to know if there is any clue to this?

PS: I don’t think it’s nvfancontrol issue, as it works fine after entering desktop.

EDIT: It might be something to do with the driver as loud fan noise after upgrading to software 9.0

I DID upgrade CUDA from 11.4 to 11.8 and software stack, might be the cause, but no idea what caused it.

3

Is this based a case on NV devkit or custom board?

At what boot stage are you talking about this case? Have you checked it by using serial console?

4

It’s offical Jetson Orin Nano 8GB devkit (SOC moduel+ carrier board)

I didn’t know what I’m looking if serial console connected. The case is quite simliar as I have mentioned above. And I’m quite sure it’s NOT Jetpack 5.1.4 /L4T35.6 issue. As it just caught my attention since CUDA upgrade. But I’m NOT sure what makes the noise.

5

Hi,

I didn’t know what I’m looking if serial console connected.

You mentioned this noise happened before nvidia logo, so you could check at which stage of boot flow that this noise started to happened and them share that log to me. Or point out a rough stage where you saw it.

And I’m quite sure it’s NOT Jetpack 5.1.4 /L4T35.6 issue. As it just caught my attention since CUDA upgrade.

This actually sounds something that do not make sense. What you reported is something before UEFI bootloader stage. But CUDA installation is after kernel.

6

jetson-orin.txt (92.3 KB)

After “Jetson UEFI firmware (version 6.0-37391689 built on 2024-08-28T08:47:11+00:00)” prints, the noise is gone.

7

Previous stage is totally non open source bootloader software and installing CUDA won’t really bring any impact to it theoretically.

8

The current phenomenon is exactly like this. 100% reproducable

Any test I can do with the board?

9

Please totally reflash the board with jetpack6.1 and see if it would be recovered.

10

I think it should be something update the non-open source code, and cause the trouble.

Re-flash the system won’t help for debug.

11

HI Daniel,

我不確定用中文解釋會不會清楚一點

感覺有些觀念上的錯誤

  1. 你講的東西本身是有點矛盾的. 你最早提到 new BSP的狀態是好的. 你覺得是某些軟體更新造成的

It should be OK with fresh new BSP as it was ocurred recently. And I think it might be related with software updates, but I don’t know what triggered.

  1. 你現在又提到懷疑是某些在non-open source的軟體更新造成的

I think it should be something update the non-open source code, and cause the trouble.

那麼我請你把軟體更新回去之前的版本或是最新的版本確認一下這樣問題還在不在這件事情本身為什麼會有問題?

non-open source的東西,你安裝cuda之類的東西後絕對不會影響他們. 你只有整機重燒的時候有機會更新或改變.
如果想確認這是不是軟體問題, 你就只能做整機重燒確保他們跟一開始的情況是一樣的.

12

I think the uart log as you requested I have attached above.
The issue is loud noise fan (100PWM, I think) before log prints “Jetson UEFI firmware (version 6.0-37391689 built on 2024-08-28T08:47:11+00:00)”.

There is no mis-understanding of the expression.

Now, what rootcause or probable issue it is?

As you have said “Previous stage is totally non open source bootloader software and installing CUDA won’t really bring any impact to it theoretically.”

Is it possbile there is any configuration caused trouble?(As we have no idea about NVIDIA’s private code)

EDIT: As update procedure is automatically, I could miss key points that might be related with the fan issue, something like configuration, software (including non-open-souce, such as NVIDIA’s private code)

EDIT2: I don’t think it’s hardware issue, as it’s quite ok when enterning desktop controlled by nvfancontrol. My guess is software/configuration. I hope someone can help to provide debug methods to do the test. Re-flashing the syetem is the last option.

13

Hi,

Just some more clarification.

  1. If this is “software problem”, then reflash the software shall restore the situation, right? That is what I want to confirm for now. Could you help and confirm that?

  2. Is this a NV devkit or custom board?

  3. No one ever reported this on Orin NX/Nano before. I am not able to directly tell you “the root cause” just according to such little information

14

I have no idea about if it’s a software problem or NOT. From my previous tech-guess, it’s NOT like hardware issue.

YES, I have answered that. It’s jetson orin nano 8GB devkit.

I have a link above, there is a reported issue: The fan will make a loud noise when the NX turn on, and will disappear after about 2~5 seconds.

But I also say in the post: There is no solution above, I want to know if there is any clue to this?

15

Hi,

  1. You told you think it is not hardware issue.
  2. You think fresh new BSP is good.

To prove above, I can only suggest you to try “fresh new BSP” for now. Confirm what you told is correct first, ok?

16

This post you shared is based on custom board. The fan spec might be different from NV devkit situation. Thus, the information from it is not sufficient either.

17

@WayneWWW

Thanks for the info of my search link of fan noise issue.

If it’s the only way you can provide for my devkit.

I can conclue, it might be the issure realted to NVIDIA’s private code or configuration or hardware flash issue???.

Hope it will never triggered the condition which might caused the fan issue.

18

Hi,

No conclusion here. I can only provide you the method to check.

If you cannot try, then nothing I can help here. All of my Orin Nano devkits here do not have such issue.

19

Yes, only god knows what’s the rootcause.

Reflash the system/update software might be never met this trigger condition.

20

If you figure out how to trigger this issue, please post it here so that we can try.

21

It’s a black box of mysteries on this fan issue. Hope someone will help to debug.