What is the best method for calculating a drives PWM frequency for a given distance?

Status
Not open for further replies.
V

VinceS

Senior Member
Most drives allow the user adjustment of the PWM switching frequency and can be adjusted over a range as broad as 500 Hz to 20 kHz. What method is best for calculating the best choice of switching frequency?

Second question:
Since the distance and PWM frequency affect the peak voltage is anyone considering this when selecting load side conductors voltage rating?
 
modification point

modification point

Assume that no line reactors are being used for the second question. Using reactors makes the question, more or less, moot.
 
Assuming by "switching frequency" you mean the carrier frequency.

Increasing the carrier frequency has little or no value other than it's effect on the audible noise. The lower the carrier frequency, the better it is for the motor and drive because the switching losses are less. So that means that essentially, the best practice is to always use the lowest carrier frequency you can live with.

High carrier frequencies can potentially increase the "standing wave" effect based on the cable capacitance in long lead lengths, so I suppose it follows that lowering the CF could diminish the effect. But if you have an issue with long leads and cable capacitance creating standing waves that cause motor winding damage, I'm not convinced that lowering the CF alone is a potential solution.

I have never heard of an issue with changing the carrier frequency to allow the use of different motor lead conductor voltage ratings however. I'm not sure why that would matter. You would never run conductors rated for less than the RMS of the line, and I have never heard of using conductors rated for more than that either. So I'm not sure where you were going with that second question.
 
Something else to keep in mind is that changing the carrier frequency probably doesn't change the speed that the transistors actually switch. It will change the number of switching events per second, but the switching events will have the same duration, the same dV/dt, etc, and thus the response of the motor leads and conductors will be pretty similar.

-Jon
 
I agree that the RMS voltage output will be pretty much the same no matter what the carrier frequency is. Either you have a bunch of short duration "on/off" events or a small number of longer duration "on/off" events.
 
I think the RMS voltage will be as expected, but not to just be stirring the pot, but when a given wire is rated for a given voltage, is the voltage a RMS measurement or a peak?

I have always understood it to be specified for a maximum, thus a peak reading would be a maximum (I need to verify this thought with a reference).

I understand high voltage transits occur, and normally wire doesn't have to be rated to cover an unknown unpredictable event. When using a drive with a PWM output its peak waveform is predictable and based on distance, switching frequency, wire size, and possible load, should be calculable.

It truly may not be worth the effort, because most of the drives I have installed line reactors are also installed. If you follow the MFR's install directions the same thing is always repeated... Lower the PWM carrier frequency for longer distance.

I under stand why, was just looking for a way to predict a carrier frequency for a given install to better provide a more imperial installation. "Yes Mr X these are the drive setup parameters and this is why there set..."
 
winnie said:
Something else to keep in mind is that changing the carrier frequency probably doesn't change the speed that the transistors actually switch. It will change the number of switching events per second, but the switching events will have the same duration, the same dV/dt, etc, and thus the response of the motor leads and conductors will be pretty similar.

-Jon
Click to expand...
I agree.
I have seen a number of motor failures resulting from excessive dv/dt.
In one case, a number of 780kW motors failed within a few weeks of installation. Of course, it bcame a bit of a witch hunt. And a steep learning curve for all involved.
The main contractor bought the variable speed drives from us and the motors from another supplier. Somebody was going to take major pain.

As one wag, and a good friend of mine, put it:
"800 kW motors should not be considered as disposable items."
 
In phase failure - first to second turn – think distributed capacitance, pits insulation until causing in phase motor failures a MEGGER cannot know

Hold your knuckles in a fist – and think of a wave coming (or rising) at your index finger followed by another so fast the next wave couples across to your middle finger before the first one is finished and doubles the magnitude - between
 
Last edited:
"Standing wave" ??? is that when the power goes out and the battery powered radio is playing Brian Eno?
 
Last edited:
VinceS said:
I think the RMS voltage will be as expected, but not to just be stirring the pot, but when a given wire is rated for a given voltage, is the voltage a RMS measurement or a peak?

I have always understood it to be specified for a maximum, thus a peak reading would be a maximum (I need to verify this thought with a reference).

I understand high voltage transits occur, and normally wire doesn't have to be rated to cover an unknown unpredictable event. When using a drive with a PWM output its peak waveform is predictable and based on distance, switching frequency, wire size, and possible load, should be calculable.

It truly may not be worth the effort, because most of the drives I have installed line reactors are also installed. If you follow the MFR's install directions the same thing is always repeated... Lower the PWM carrier frequency for longer distance.

I under stand why, was just looking for a way to predict a carrier frequency for a given install to better provide a more imperial installation. "Yes Mr X these are the drive setup parameters and this is why there set..."
Click to expand...

Vince, I don't have a reference per se, but using logic and experience, you can see that the rating of cables is an RMS rating. Take a 480 V installation. These installations use 600 V rated cable. The peak on a waveform is around 680 V for a 480 V rms. If the cable rating was a peak rating, this would not work.

I do not know how to predict what the ideal carrier frequency would be either and would also like to find out if anyone knows.:)
 
VinceS said:
I understand high voltage transits occur, and normally wire doesn't have to be rated to cover an unknown unpredictable event. When using a drive with a PWM output its peak waveform is predictable and based on distance, switching frequency, wire size, and possible load, should be calculable.
Click to expand...
Yes, should. But I don't think it is by any means simple. The best approximation I have got is by application of (short) transmission line theory. Even then, it can be difficult to get a good handle on relevant cable and motor data. Results can be no better than the data. Then there is the interpretation of the results. How bad is bad? Is 1200V/us ok? How often can the motor take that? Etc.
It's at that point you need to move from predictive to evidential. Most of us in the commercial world can't afford the luxury of destroying systems to test the limits. So we then should look at recommendations from institutions who can collect and analyse information on the subject.
 
Food for thought

Food for thought

I've spent way to much time on this subject. I miss the tech support folks at Reliance. I could ask a question and get a simple answer, normally with ref's. This web link to a Reliance .pdf actually sources a IEEE study on the subject. I've found this area the subject of at least 4 EE Thesis's (found freely on the web) and numerous IEEE articles (Pay per view), I had to get out the old calculus book, Ahhh....

To cut it short it appears through 1994 IEEE study, 6Khz is the optimal PWM switching frequency. I still don't know how this value was obtained, but I'm sure they were getting funded to research the subject.

Of specific interest was the table found at the end of the .pdf describing the relationship of drive to motor distance and dv/dt and its corresponding peak voltage.

http://www.reliance.com/pdf/drives/data_sheets/D7171D.pdf
 
Status
Not open for further replies.
Top