Will PWM frequency play a significant role in copper losses such as skin effect, proximity effect?

[Clyde38]

I have read the threads pertaining to losses dealing with skin effect, proximity effect, etc. plus numerous articles about the subject. In an SR motor that has a commutation frequency of 300Hz to 400Hz, skin effect does not appear to be an issue and I have not had experience with these types of losses at low speeds (under 1000 rpm). If the drive operates at a PWM frequency of 20 kHz as a hysteretic current regulator with a hysteresis band of 5%, will this be a cause for concern as far as skin effect? I can't see that it is. Depending on the current, I can see an issue as far as proximity effect, not however due to the PWM frequency. Given the added expense of using Litz wire and the additional issues with its use, or winding methods such as Roebel windings, I'd like to be sure that it is either needed or not. I'm currently of the opinion that the PWM frequency in this type of drive is not an issue with respect to these types of losses. I believe that it is the commutation frequency that is relevant. As always, I welcome any comments or advice available.

 

[2dye4]

I would think a 20 khz pwm signal being switched in polarity 300-400 times a second would have a spectrum
appearing a spike at 20 khz smeared by sidelobes that match the spectrum of 300-400 hz squarewave.
The convolution of the spectrums in other words.

So much of the signal energy will be near the 20 khz mark.

If the wire radius is say 2 or more times the skin depth of the material in the windings (probably copper)
then skin depth and prox effect will be minimal.

Otherwise your milage may vary.

I found one site that gave skin depth of copper at 20 khz to be .46 mm

Search for skin depth of copper and check wire diameters.

 

[PeterKsiazek]

Clyde38,

The current in the machine will have two components, the low frequency average current at the machine speed (which your hysteretic regulator is setting) and the high frequency ripple current at the PWM frequency. With regards to skin effect losses it is only the high frequency component which must be considered. Because this ripple is generally small it is not usually an issue.

2dye4, you are correct about the spectrum of the Voltage, however the current is filtered by the inductance of the machine winding.