When I run GameWorks, VisionWorks, or CUDA samples included with Jetpack 3.1, I hear a noise from the processor.
It is not the fan noise. I can’t explain it very well, but it’s a subtle, very high-frequency noise. It sounds like the processor is operating beyond its capacity.
I’m certain it’s not the fan noise. I’ve applied new thermal grease and removed any dust.
This is an 8-year-old Jetson TK1 board.
Another strange thing happens: when I plug in the LAN cable to the Jetson TK1, it suddenly turns off.
I have a rubber plug, and when I attach it to the LAN cable, the board also turns off suddenly.
Due to the device is a bit old, it’s not easy to dig how it happened and why.
You may try to capture some logs as reference to see if can find some clues.
FYI, sometimes a multilayer ceramic capacitor can make an audible noise if it has sufficient current going through it at an audio frequency. Any coil or transformer can do this as well, although I don’t think TK1 has such a coil. Capacitor is a possibility.
If the cable connecting or disconnecting causes a shutdown, then that is likely from a grounding issue. Perhaps the cable and the unit have some sort of ground loop.
Your network equipment can be at a different ground point than other parts, e.g., the power plug/“wall wart”. If current flows, there can be a problem. If the power is rather high frequency, then just capacitance is enough. Static discharge can matter. I don’t know the exact circumstances you are working with, and it might not even be a ground loop, but by far I think what you described as happening is related a ground loop or static electricity.
What is the exact act which produces the reboot? That rubber plug makes me think you are speaking of a shutdown from something more than inserting the ethernet plug. If it is ethernet, then most likely something from ethernet has a ground loop to the Jetson power. This is especially true if you are using a higher end shielded ethernet (I don’t mean that it has shielding; this is a type of termination there the connector itself has a metal exterior, not plastic; in this video part of the plug is metal, but what most people are used to is entirely plastic): https://www.youtube.com/watch?v=GzWXMCFm-AU
I would have to wonder if there is a static electricity issue. Another possibility is that the plug acts as a capacitor dielectric, and one of the nearby signals starts coupling where it shouldn’t (but I have not heard of that happening on a Jetson, it would be something new).
Is the board physically held solidly, or is there any wiggling when putting the plug in? Are two hands used, or just one? I ask the latter because if you have to hold the unit while putting the plug in, then the electrical effect is different compared to if you stand back and there is no contact other than the plug.
I am always cautious with static electricity. I held the power cable adapter with one hand and inserted the RJ45 plug, believing that there was no way for static electricity to be generated.
To prevent a short circuit, I used a glue gun to apply glue to the four corners of the Jetson TK1 board, allowing it to float in the air.
I don’t doubt that it is doing what it does. Without some pretty expensive testing equipment though, it is really a lot of guessing. Take for example the glue to the four corners: This won’t ground, and leaves the Jetson “floating”, which sometimes helps and sometimes does the opposite. I’m kind of grasping to see what might have a possibility. If we talk about the system as a whole, then we have to consider the ground of the Jetson, locally, the ground of the power supply, the cabling of any networking, the ground of the power to any networking, so on.
Can you say more exactly if the RJ45 is always plugged in or always unplugged? What is the reason for unplugging it and using the insulated cap?
Normally one would mount a board like the TK1 using the mounting holes to a ground back plane. Is it possible you could at least temporarily mount this via the mounting holes to a metal back plane? I suppose even cardboard would do if you were to wrap it in aluminum foil.
Also, note that insulators do have a dielectric value. Static includes charge, and not just discharge. There isn’t enough information to know why it has an effect, but obviously it is doing something. The trick is to try to change one thing at a time and see if it differs.
