Electrical Interference from Roboteq

<P style="MARGIN: 0px">We are using two Roboteqs to control steering, brakes, tranmission and 2WD/4WD on our 2005 DARPA Grand Challenge vehicle. During integration of our sensors in the vehicle we began to notice garbling our our serial port communications.
<P style="MARGIN: 0px">
<P style="MARGIN: 0px">In particular, signals from our GPS and other sensors had serial port problems. Sometimes the data would come through garbled, then would suddenly come out normal. At all times the roboteqs were on but not moving the motors. However, the serial cables run about 8 feet, and come within 2 feet of both Roboteqs. We noted our computer monitors were showing flutter when they were this near to the Roboteqs.
<P style="MARGIN: 0px">
<P style="MARGIN: 0px">At the same time we noted that the Roboteqs had fewer problems connecting to the same computer via serial ports. This was true with direct connections and USB-serial. Could this be due to the less error-prone comm protocol of the roboteqs?
<P style="MARGIN: 0px">
<P style="MARGIN: 0px">So my question is: does the Roboteq put out RF interference when it is not actually moving motors - in particular something that would affect serial ports? What would be the best way to shield our serial and other network cables?

<P style="MARGIN: 0px">I assume that you have determined that switching off the controller causes your other peripherals to communicate OK.
<P style="MARGIN: 0px">
<P style="MARGIN: 0px">I suggest that you use an oscilloscope to ckeck the shape of the RS232 signals near your computer and near the GPS. RS232 signals swing from +12V to -12V so they are very much noise immune. If you get garbled data, you should see it in the waveform.
<P style="MARGIN: 0px">
<P style="MARGIN: 0px">The controller switches the motor with a 16kHz PWM cycle and extremely sharp rise and fall times. While this improves efficiency by minimizing switching losses, it may induce high frequency noise throught the wiring's inductance.
<P style="MARGIN: 0px">
<P style="MARGIN: 0px">However, when the motors are 100% stopped, there is no PWM switching going on and there should be no noise generated from that source.
<P style="MARGIN: 0px">
<P style="MARGIN: 0px">Noise reduction is always tricky.
<P style="MARGIN: 0px">
<P style="MARGIN: 0px">I suggest you try to add an RC snubber between the green and white wires, as close to the motors as possible. This is a simple circuit made of a resitor and capacitor in series. Try 20 Ohm and 0.1uF.
<P style="MARGIN: 0px">
<P style="MARGIN: 0px">Then add ferrite cores at each end of the RS232 cables. You can have the cable go through the core, or make one loop through the core. You may also try putting a core around each of the thick cables. See <A href="dkc3.digikey.com/PDF/T052/0892.pdf" target=_blank>dkc3.digikey.com/PDF/T052/0892.pdf for cores examples. The snap on are the easiest to use.
<P style="MARGIN: 0px">
<P style="MARGIN: 0px">Beware also about the possible problems from ground loops. Best would be to have the black ground wires go directly to the battery terminal rather than to the vehicle's chassis.
<P style="MARGIN: 0px">
<P style="MARGIN: 0px">
<P style="MARGIN: 0px">