PWM Signal Noise?

[Danny Diaz]

I have been thinking about the wiring design for Victor speed controllers lately (and sensors as well), and I was wondering about the effects of DC EMI on the PWM signals and on any analog sensors we might have on the robot. It appears to be quite common to run the PWM wires intended for driving the Victors in parallel with the DC power lines to Victors in a "wiring harness", especially when attempting to make the wiring on the robot as clean as possible. I have seen a couple threads on this board about having interference with camera servos when the PWM leads get too close to FP motors, but couldn't you have similar problems with the PWM signals going to the Victors when the motors are pulling quite a bit of current? I understand that the effect is greater over greater distances, but I would have figured this would be a bigger problem/concern than it appears to be. Is it, and why/why not?

Thanks!
-Danny

[Racer26]

Well, 1075 has had PWM and camera wiring run with power wiring for years and never seen any noticeable crosstalk there... we have on occasion noted some with the radio being too close to the CIMs... Although this year, I think its just the crap radios that got handed out.

[Mike Betts]

Originally Posted by Danny Diaz

I have been thinking about the wiring design for Victor speed controllers lately (and sensors as well), and I was wondering about the effects of DC EMI on the PWM signals and on any analog sensors we might have on the robot. It appears to be quite common to run the PWM wires intended for driving the Victors in parallel with the DC power lines to Victors in a "wiring harness", especially when attempting to make the wiring on the robot as clean as possible. I have seen a couple threads on this board about having interference with camera servos when the PWM leads get too close to FP motors, but couldn't you have similar problems with the PWM signals going to the Victors when the motors are pulling quite a bit of current? I understand that the effect is greater over greater distances, but I would have figured this would be a bigger problem/concern than it appears to be. Is it, and why/why not?

Thanks!
-Danny

Danny,

Coupled noise on the PWM wires (voltage spikes) would be from high di/dt, a fast rise or fall in the current going to or from the Victors. If the load on the victors was resistive, the current draw would be a square wave at the switching frequency of the Victors and you would expect crosstalk. The fact that the motors are a large inductive load smooths the current into more of a DC current which lessens the cross coupling to the PWM lines.

Also, the PWM signal is digital in nature. It takes quite a bit of noise to interfere with a digital signal.

Analog signals are more susceptible to crosstalk but most ADC inputs have an anti-aliasing filter built into the front end. This filter would also mitigate the effects of noise.

Note that high di/dt can also be caused by the make/break of the brush contacts of the motors. More on this in a second...

The one component which has demonstrated a noise coupling susceptibility is the voltage feed to the CMU. A particularly bad noise generator is the Fisher-Price motor. In Atlanta, it was demonstrated to FIRST that FP/CMU interactions exist and that adding a capacitor to the FP motors improved the situation.

I was hoping that this year's rules would allow teams to install a noise filter to the CMU power input and to the motors but, alas, I do not see it.

The problem is that there exists no formal way to feed engineering issues back into the FIRST system.

Sorry for the digression. I hope that the beginning of this post helped to address you question.

Regards,

Mike

[Al Skierkiewicz]

Danny,
I agree with Mike on this. The FP motor (as most brush type motos) generates considerable RF energy due to brush noise. Just looking into the motor will show why. There is a lot of lightning inside when motor is running. I suspect that the CMU cam has a relatively high input impedance on PWM loop (and no RFI filter) while the robot components have lower impedance inputs and filtering. The Victor and Spike also use an optocoupler on the input. By adding capacitors to the motors, the RF output is significantly reduced. It is a method used with motors of these types all the time. The Globe is also a great little generator. I have to add capacitors to this motor when installed in a variety of applications in which they are used. One of which is a rotation control for boom cameras. The noise would get into the video for any camera even close and often generated havoc with wireless mics and guitars amps which have a very high input impedance.
If asked, I would recommend that First not only allow, but recommend, the addition of 0.01 mfd, ceramic capacitors to at least the FP and Globe motors, at the motor leads.