Hi,

EMI or EMC work like this is usually performed in an RF tight Chamber or certified open-field test sites, with calibrated antennas, and receivers or spectrum analyzers.

You are in the mode of identifying what the emission source(s) are first. It is either coming from the controller electronics, edge rate of the MOSFETs in the motor driver or from the motor.

The antenna for the emissions is the cable between the motor and the controller board including the traces on the controller board if they are surface routes. The USB cable could also be an antenna if it is not grounded well on both sides. Both sides need to grounded if the frequency exceeds 1 MHz or edge rates are faster than 1 us.

1.0 Servo Noise sources
-----------------------
You need to see if the noise is related to the controller electronics, Motor Driver, or the Motor.

1.1 Controller Noise Sources.
.............................
The controller will have a crystal controlled clock source on it near the processor chip. It divides this clock down and uses it for logic timing for the controller. The clock is used to create synchronous and asynchronous strobes, USB interface timing, and time databuss transaction. The noise will be heard as a whine for steady state clocks, clicks and chirps for strobes and buss transactions.

Note the oscillator frequency that is marked on it (like 4 MHz, 16 MHz, etc, it may be 3.579, 4.000 or 4000k). Multiply that frequency by 2, 3, 4, 5, 6, 7, 8, 9, and write the numbers down. Then divide the clock by 2 and multiply that number by 3, 4, 5, 6, 7, 8, 9, and write those down too. They will be suspect frequencies to look for as controller noise sources.

1.2 Motor Driver Noise Sources.
...............................
The Motor Driver usually makes noise based on the modulation frequency (5 khz to 20 kHz) or the edge rate of the MOSFETs (MHz). The modulation frequency will often be heard as a whine or gated buzz. The MOSFETs will generally be heard on the FM Radio and are usually pulsating white noise.

1.3 Common Motor Noise Sources.
...............................
Generally, the commutation of a common DC motor will generate predominantly broadband low frequency noise (hence, the AM radio for the receiver system) and also recessive high frequency noise from the commutation arcing.

1.4 Brushless DC Motor Noise Sources
.....................................
If it is a Brushless DC Motor, BLDC motors have noise in the higher frequency bands as the motor package re-radiate the conducted noise from the edge rates and modulation frequency of the Motor controller. On a spectrum analyzer it looks like 20 kHz (or the modulation frequency) narrowbands (spikes) spaced at the modulation frequency intervals on a carrier frequency based on the MOSFET Edge Rate. There can be groups of these throughout the spectrum. The edge rate of the MOSFETs determine how high in frequency the narrowbands clusters will go (how many 100s of MHz).


2.0 Sniffing to find the emission source
----------------------------------------
One inexpensive and easy way to see if you are experiencing EMI conducted or radiated emissions is to find an small AM/FM transistor radio (for portability) - digital dial may help too, and move it to locate the source and frequency of the offender. You can tune it from 530 kHz to 1600 kHz and listen for the noise. It is very likely that you will hear a buzzing or whining sound when it is idle or during operation of the motor. If not, check FM band and tune across the band from 88 MHz to 108 MHz.

2.1 Without The Motor Running
.............................
You might want to disconnect the motor for this test if it will not harm the Motor Driver chip or any feedback electronics (like position sense, tachometer, or Hall Effect sensors for BLDC commutation, etc).

2.1.1 Move the radio tuned to the AM band (tuned to 530 MHz) closer to the motor, then to the cable, then to the controller, and see which one is louder. Repeat at mid frequency band and 1600kHz. You might want to scan other frequencies too. If that doesn't show much, do the same thing with the FM band at 88 MHz, midband and 108 MHz. See if any of the frequencies match (are close to) the ones you wrote down.

Any frequencies shown here will likely be controller electronics generated noise.

2.2 With The Motor Running
..........................
You need to see if the noise is related to edge rate of the MOSFETs in the motor driver or from the commutation noise of the motor. These will be heard as a buzz or whine for the 20kHz (or whatever modulation frequency) of the controller, or a frequency buzz for motor noise.

Use a similar method to sniff as above in 2.1.1.


3.0 Source Suppression Fixes
----------------------------
If the noise is a constant whine in the FM band without the motor running, it is noise from a clock, strobe, or bus transaction. This usually has to be solved via shielding or source suppression at the circuit level, with proper signal line termination, local shielding of the controller or offending section. You need to see if it is radiated from the board or conducted through cabling.

3.1 Controller EMI Containment & Source Suppression
...................................................
The first easy thing to do is insulate the board and wrap it completely in aluminum foil, you may have to ground the foil to the connectors. This is called EMI containment. Radiated emissions can cover a large area. If the noise goes away or is drastically reduced, good. If not, do the same thing with the cable that goes to the motor and ground it on the motor and the foil wrapped around the controller board. If this solves the problem, you have a some options.

1) Keep it shielded (wrapped) with small 1/4 inch holes (paper punch) offset spaced 1/2 inch apart will allow cooling and containment.

2) If the noise is coming from the output cable, a ferrite core on the cable of Fair-Rite Corp #43 or Steward Corp. #28 will help. Both motor wires are twisted at 4 twists per inch and the wtisted pair is looped through the core 2 to 4 turns.

3) You may have to create a common mode choke (another email).

4) If the noise is on the cable, it may also help to twist the wires and wrap a shield (copper with tin lead braid - you need the ferrous characteristics of the braid for low frequencies). ground the cable on both sides. Shield termination of 360 degrees is preferred to a wire or pigtail termination.


3.2 Motor EMI Source Suppression
................................
If you do have buzzing noise in the AM band as the motor runs and it is quiet when the motor is running, solutions can be as easy as paralleling a 0.01uF to 0.1 uF ceramic disk across the motor pins (at the motor). This has to be approached with some caution as MOSFETs can dump a lot of current into the capacitor and may increase dissipation in the motor driver chip (there is a finite amount of Power Capacitance MOSFET Drivers can handle).

If it is higher frequency noise you may have to install small ferrite bead-on-lead components on the Motor and/or Motor Driver leads. Same materials and manufactures as above. The worst you may have to do is add "Y capacitors" from each output to ground. This would be another email. The Y Capacitors provide a preferred return path local to the Motor Driver instead of allowing the noise to travel across the cabling (read as - hey, I found an antenna to re-radiate the conducted noise on me) to the motor and return back to the Motor driver.

Sorry again for the long response, but, if you really think it might be EMI, I thought you would like to understand what it, how it is generated, how to find the source, and ideas on how to fix it. This only scratches the surface and there is more that can be done to fix it if it doesn't work. At least this can eliminate EMI as a potential problem and you can focus on code which I try to stay away from <grin>.

I know, I've heard it before - GEEEEK,

Ross