I have a newly designed drone which has a Jetson Nano on top, and a Beitian BN-880 GPS module right beside it (about 2cm away at the same height as the Nano PCB).
With the Jetson Nano turned off, the GPS module works fine but as soon as I turn on the Jetson Nano, the GPS loses all satellites and fails to detect anything. GPS satellites transmit on 1.2-1.5GHz so that must be where the interference is. All my cabling is shielded but the GPS module and Nano are exposed.
Do I need to wrap my Jetson Nano in aluminium foil? I’m not keen to do this because:
a) Cooling problems.
b) Potential for short circuits.
I’d appreciate any help and suggestions. Many thanks.
It could be transmitted noise through a shared power supply, in which case something like ferrite bead might help, but probably you do need to shield the Jetson. The problem may not be one of completely enclosing the Jetson, so perhaps a grounded shield on just one side would work. Holes are also ok in most cases, e.g., screen wire or small visible holes (the size just has to be significantly less than the wavelength…imagine a microwave oven having a see-through door with a mesh in the glass with holes probably a few mm in size…it wouldn’t stop K-band radar, but it will 100% stop the microwave’s output).
Computer components will always have interference issues if not routed correctly. They run with something close to a square wave, and this in turn is the infinite sum of the odd harmonics. If the logic used a sine wave it wouldn’t be a problem.
One thing I suggest is to experiment with placing a single grounded plane in different places. You might find the required coverage is less than 100%.
This has now been solved.
I had removed the metal case of a webcam that was quite heavy in order to reduce weight on the drone. Turns out the case also acted as RF shielding. :-(
The problem was not clear because my Jetson was enabling and disabling the USB camera at different times, making it hard to diagnose.
I was too early in saying this was solved, but it has been diagnosed.
The camera itself emits no interference, however the cable leading to the camera kills the GPS and I see the number of satellites drop from 17 down to zero.
I am wondering if the Jetson Nano USB connectors are properly grounded because the cable I am using is shielded (bought from RS).
USB itself won’t work if not properly grounded. The effectiveness of the grounding and shielding of a sine wave is far far different than when dealing with a square wave. You cannot use the same standards for shielding with a square wave as it is a complicated/complex mix of signals. You would have to put a Faraday shield around the Jetson, and for any other similar system…this is not specific to a Jetson. Bad grounding will be an issue, but this is not what you are experiencing.
There is no reason you could not use a very light weight and vented cage around the Jetson, it is just something you will need to experiment with. Something as thin as aluminum foil with a lot of vent holes on a simple insulator (such as thin paper) would do the job without causing heating issues and without a significant weight. With square waves there is no getting around this, and this is what digital logic is.
Do you think putting the entire Jetson in a Faraday cage would help the issue of interference from the shielded USB 3 cable?
I have tested the cable with my laptop and it also knocks out the GPS on my phone if the cable is too close.
This (shielding the entire Jetson) probably would help, but it is possible the shielded USB3 cable is itself operating as an antenna. In theory the shielding to the ground side will stop this, but between ground loops and the fact that the square wave involved is by nature a lot of noise (it is the infinite sum of the odd harmonics…every harmonic is noise if transmitted as loss instead of shielded), it makes sense to always be suspicious of every wire.
Some cables have better shielding than others, so cheaper cables will have breaks in the weave of the shield braid, while others will use more conductor and not have large “holes” in the weave of the shield wire. What you can do to test is to try without extra shielding other than on the Jetson itself, and if you have issues, then wrap (at least temporarily) the USB cable too. Ground only one end of the shield around the Jetson to avoid ground loops. Then check for results also with both ends of the test shield around the cable grounded to both devices.
If a ground loop occurs, it means power is flowing through the ground of the cable due to (probably) power supply issues. If the GPS is powered by the Jetson, then this really wouldn’t be an issue because they use a common power supply, but if the GPS is powered with some sort of wall plug adapter, then I would become much more suspicious of the wall plug adapter. I know the one NVIDIA provides for dev kits is quite good with respect to this, but you might mention what kind of power arrangement you have for the Jetson and the GPS (and if not a dev kit, the power supply used).
I think that the USB3 cable is acting as an antenna too. I see a similar problem with my laptop connected to the webcam (but not as bad as connecting to the Jetson Nano). The cable is shielded and expensive/good quality from Roline (“USB 3.1 high-speed cable in high quality with improved shielding”).
My setup is like this:
I’ll get to work with the copper tape and report back here once I’ve done that. I really appreciate the support!
This tends to point at a ground loop. Is your laptop running isolated from a power supply (on battery)? If so, ground loops are probably not an issue, though I suppose a wired connection related to some other device on the laptop (or Jetson) could still cause a ground loop.
In your above block diagram (btw, thanks, that is easy to read!) I see that normally there is nothing touching an external power supply. The battery is isolated from something like a house ground. However, a single converter is shared between the Nano and the Autopilot (which contains the GPS). So it is possible that there is a ground loop involved there. Even if it is temporary, would it be possible to power the Nano and the Autopilot through two independent power sources?
Related, but not as severe as a ground loop issue, imagine that the power supply is regulated, but not perfect at the frequencies involved in the interference. The Nano could in theory “modulate” the power going to the Autopilot. RF filters, e.g., just a ferrite bead, might solve the issue.
In the case of the USB wire itself being an issue, then you could ground yet another shield around the USB cable, but be certain it is grounded at only one side during testing. The goal would be to avoid a current flowing through that shield between the two devices by means of the shield. Grounds can be antennas too.
When I test with the laptop connected to the webcam, there is no mains connection, just battery. I bring the USB cable close to the GPS and see it lose signal.
From your informed post I am able to do 3 things:
- Power the autopilot separately from the Jetson Nano and see if the GPS still works
- Use a ferrite bead for the power going into the Nano. I just had a look at these and I think I would have trouble using these due to their voltage drop, especially if the Nano is taking several amps. Maybe I could use a capacitor instead?
- Apply shielding to the Jetson Nano (I’ll use copper tape).
What you need is a low pass filter. You are correct about some ferrite beads causing problems, but I suspect the frequency is high enough that one of the lower inductance (and lower loss from resistance) beads would do the job. Also, consider putting it on the GPS power instead of the Jetson power. I will emphasize though that you are quite correct that Jetsons are sensitive to power drops.
Something like a bead on the GPS power, plus perhaps a small capacitor (part of low pass) would be ideal, but at that frequency there is a significant chance that the bead by itself will work. A capacitor by itself could work, but I would not be confident with that.
As far as shielding goes, this would be your most reliable way of getting things done, but only if the power supply is not passing the signal. You kind of just have to test and find out. You copper could be a mesh or even have holes in it bigger than what might be considered a “mesh”.
I’d like to give an update having done loads of testing today. The short version is, the Nano itself is causing minor radiated interference at all times, but when using USB 3.0 it is significant enough to take out the nearby GPS. EMI is caused both by the Nano and the shielded USB cable.
The longer version with more details:
- My GPS is about 7cm from the Nano.
- I powered the Nano from a double-insulated mains DC 5v adapter
- Initially, there were no peripherals attached to the Nano (GPIO or USB) except for an unused CSI/MIPI camera.
- The Autopilot/GPS was powered from a battery via a UBEC as in the previous image. i.e. the Nano was totally electrically isolated from the Autopilot/GPS.
- I grounded the GPS cable shielding and twisted the wires.
- I installed a split ferrite core on the Nano power supply cable (right next to the Nano).
- With the above setup, I plugged in the USB 3 webcam and the GPS dropped out immediately. I connected it to the USB port furthest from the GPS and made sure that the USB lead itself was going away from the GPS to ensure reduced interference.
- Adding a ferrite core to the GPS wires did not help, nor did adding the core next to the existing core on the Nano power wires.
So next step will be to make a Faraday cage for my Nano. BRB!
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.