Am i right in saying - EMC / PWM content

[max_torque]

Imagine i want to drive a DC motor with a typical H bridge using the normal PWM techniques. The motor is mounted a short distance away from the controller, and linked by wires.  With no  additional inductance (other than the parasitic bits) between the bridge outputs and the motor, those wires see a dV/dT equivalent to the switching rise times multiplied by the supply voltage. So, if we switch in say 1us, and we switch 10v, then that's 10MV/sec, and because the waveform is likely to be very "square" we will get a large harmonic content potentially radiating out up to a significantly higher frequency?

So, in order to minimise radiated EMC, installing some small amount of additional inductance on the controller, will tend to round off the voltage waveform and push it towards a more sinusoidal shape, with slower edges, and hence cut the radated energy?

If i followed that inductance by a small amount of capacitance to the H bridge Ground (or supply, or both) then i would provide a local HF return path for current, ie a snubber, say something like a 10Nf cap with a 50 ohm resistor (to prevent high freq oscillation)


Am i wide of the mark here?  I'm not worried about cost or efficiency etc.  Some basic simulation suggests matching the motors inductance gives the best result?

Any traps "for young players" lurking??   
 

[T3sl4co1l]

Is it really 1us?  That seems awfully pokey for an H-bridge.  Plus there's a lot of feed-forward from stuff behind the transistors: gate drives, logic, supplies; you don't them see on the scope because they're swamped by the huge edge, but nonetheless radiate well over limits if not controlled.

Assuming this is a brushed PMDC motor, you should worry also about brush noise.  I've seen that kill product development on things that didn't even have a switcher inside 'em!

If you need the certainty, you can do very well by using shielded cable from H-bridge to motor, making sure to ground (RF ground -- use lots of bypass caps around its perimeter if, for whatever reason, you can't have galvanic grounding between the two) the shield, thoroughly, to the motor case and circuit ground plane.

Tim

[max_torque]

Voltage rise time (10-90) measures at around 800nS, using a TLE5206 integrated H bridge. Clearly switches fast so it can carry large currents with a small thermal overhead!  Motor is Brushed, but is 'static' in operation (ie it just holds a fixed position rather than actually rotates) for most of the time (99.99999% of the time in fact)

[T3sl4co1l]

Ah, that is unusual.  Some light filtering of the outputs, to take the edge off the edge, might be desirable.  Otherwise, that shouldn't be too bad then.

FYI, don't use ferrite beads, they saturate at low currents (~200mA even for large ones).  Use real inductors, not much needed, 0.3 to 1uH is fine.  Use a small cap (~1nF?) in parallel with an R+C (C = 2.2nF, R = sqrt(L/C)) to dampen ringing.

(Also, pedantic note: nS = nanosiemens (conductivity), ns = nanoseconds.)

Tim

[max_torque]

Yes, i was a little surprised myself on how fast the switch times were! 

The application is to hold a (variable) static torque in a Defense environment, so the EMC requirements are quite draconian!

I think i'll knock up a test rig, play around with inductors and snubbers and see what's what (i have a TEM cell / SpecAna so i can at least make some comparative analysis).

I guess in extremis i could filter right down to a really low frequency, and effectively drive the static motor with DC