equation for E field due to current in loop (near-field)

[DHambley]

I need to model our present system to show a predicted before-and-after E-field noise in our attempt to meet FCC part 47 class-A (industrial location). We're not spending the money for EMI lab time until we have a prototype EMI fix.

I'm looking for a generalized equation to predict the E-field from a loop consisting of a cable bundle and chassis. The motor currents are differential so that the 3 phase currents within the bundle sum to zero but, there is a high current spike in the cable-chassis loop each time a voltage square wave changes state.

A simplified descrition of our system: Cables ~1 meter length and 20cm average off the chassis (so area A = 0.02m). Voltage Wavefrom = 300V pk-pk square waves at 20KHz. The motor stray capacitance to chassis is 5nF on each phase and so, this capacitance has to charge and dischage at every square wave transition.

I'm thinking that I can model the loop as an inductor (which is in series with the stray capacitance being charged) and calculate the resulting votage waveform as its inductance slows the current rise into the stray capacitance, but I don't know what to do with this number to find a resulting E-field. How would you predict the resulting E-field as it would be detected by an antenna 10 meters away?

thanks in advance for any help with this.

 

[hacksaw]

try motor controls forum

 

[VEBill]

To control EM emissions, in general, one does the opposite of what one would do to make a good antenna. You mentioned the capacitance from the cable to the chassis, and if I'm interpreting your words correctly, that the capacitance is part of the problem. In general one would install the cable as close to the chassis as possible to reduce emissions. Such closer possitioning increase the capacitance to ground, but is normally a good thing.

But beware of where the ground currents run if you're dealing with common mode noise currents.

Notice I wrote currents. I think it's safe to say that most EM issues originate from currents as opposed to voltages. In my view of the world, this is because long thin conductors are so much more common than large flat plates.

Ideally one might make a NEC (Numerical Electromagnetic Code) model. But that can obviously be a time- and money-sink.

Or just make all the most-obvious corrective actions including shielding, ferrite beads, bypass filtering, etc.