Question about magnetic shielding (Solved)

For weight/balance reasons, the best option for mounting the Xavier on my rover is about 1-2 inches away from a ~250W brushless motor.

In this picture, the red can is the motor, and the Xavier is right there.

Do I need to worry about magnetic shielding? Is it likely that the motor will generate sufficient EMI to interfere with the I/O coming out of the Xavier, or even the Xavier board itself?

For various reasons, I’d prefer not to rotate it, and weight distribution would be worse if I moved it further forward.

Hi, did you meet EMI problem? The motor power looks high and distance looks close, better to add shielding.

I haven’t tried it with this position yet. I had the computer higher up (another 2 inches) and further to the front (another 2 inches there, too) before, and that worked OK.

My question is whether there are particular EMI design targets for the Xavier. Current limits for close wiring? Field strengths? I couldn’t find anything in the specification.

No such info can share by now, you can find Environmental & Mechanical Screening in module datasheet, it contains a series of industry standard tests.

Yes, just not the tests I want :-)

Thanks for the answer. I will chicken out and move the module further forward, and live with the bad balance.

You could just put a wire mesh cage around it.

That shields against electric fields, but doesn’t really help me with the wires coming in and out of the cage. Also, such shields are less useful in low frequency ranges – the motor runs in kilohertz ranges, not megahertz, so a cage might give me 10 dB or so of shielding. Not insignificant, but one-over-r-squared might do better …

Not to mention, highly impractical in practice :-)

I think the motor can is aluminum, unfortunately. Wrapping it with mumetal, or very pure iron, might do it?


Sorry for off-topic, but is it a FOCBOX (VESC-X) ESC from Enertion you are using in your car? If so, how do you like it? Is it able to work reliably with your sensored brushless motor at slow speeds and which motor model is it? Are you running it in FOC or BLDC mode? Sensored or sensorless? Did you try Ackmaniac’s firmware?


Yes, that’s the one. I like it. I’m using sensored mode. I forget whether that maps to FOC or BLDC?

The motor is this hobby motor:
They claim it’s 1800 W, but that’s just a marketing claim :-) It would obviously melt within seconds if you pushed that power into it other than just during a quick start-up transient.

I have not tried other firmware. I initially got the focbox because the sensored mode on the original ESC wasn’t very good, and the FOCBOX is better. It runs smoothly at reaosonably-slow speeds after calibration. Not as slow as a proper PID controller on a brushed motor with encoders, but slow enough for me to walk faster than it.

I’ve been meaning to update the firmware to be able to read out a cumulative tick count of the internal sensor phase count, but I haven’t had the time to do that. I control it and read back status using a serial port UART from a Teensy microcontroller that in turn I talk to from the Jetson. I generally prefer to keep a $20 interface between the wild motor currents and the $1200 computer, just to be safe :-) Also, the Teensy decodes my radio transmitter signals, and can apply an emergency cut-off more robustly and reliably than the Xavier if needed.

3M makes a shielding tape and sheet.

I use it on r/c helicopters that I am turning into autonomous UAVs. There is not a whole lot of room on the birds as you can probably imagine.

Right now, we are assessing its effectiveness with shielding effects from the motors and servos with the magnetometers, compasses and various receivers as well as evaluating the sensors. So far, the stuff is working well.

Additionally, where required you can also use shielded cable.

Hope this helps.

I am curious about the weight of that motor…it simply lists as “445” with no units. What would you guess it weighs?

Thanks! My son is working on his rover project using 1/7 scale car (cen reeper) and he wants to replace its stock hobbywing ESC with a VESC as well. The car came with a hobbywing sensorless motor from the factory, not sure if we will end up replacing it with a sensored motor. AFAIK, in VESC BLDC mode is always sensorless, but you can configure FOC mode to be either sensored or sensorless. I guess you must be running in sensored FOC mode. I’m curious if you had a chance to run your vehicle sensorless with VESC and with what results if so… My son is using teensy too! But Xavier is mounted upside down bolted to the car chassis by its massive heatsink.



@linuxdev Motor stats page I linked to: Weight: 329g
I think that’s very close to correct. Less than a pound, but more than half a pound, when holding it in my hand.
Note that the terminals are solder-only – it comes without wiring.

@lkealey64: the shielding tape is typically intended for RF intereference rather than low-frequency magnetic coupling. Sometimes it’s made of copper, which isn’t very good for low-freq. However, there’s steel (iron) mesh available from McMaster which could work okay. There will still be leaks through holes where things enter/exit – if I wrap the motor, the drive shaft, and the wiring.

@albertr I haven not tried it in sensorless mode. If you’re doing an autonomous rover, you really do want sensored, and ideally reading counts back somehow. Some people glue black/white striped paper to the inside of the tires and read with a phototransistor; that’ll work too.

Yes, of course we want to read/calculate distance based or motor or wheel rotations. I think it’s pretty easy to read motor RPMs from VESC, but I’m not sure how accurate this info is. Are you calculating your driving distance based on the data read from the VESC? If so, can you share your experience on its accuracy? Installing hall or opto sensors on wheels should IMHO give accurate data, but it’s cumbersome. Right now we are more concerned with slow speed accurate and stable driving using brushless sensorless motor we have. Gotta fix driving first, then will take a look distance calculations.