Measuring VFD (PWM) V & I

[buzzp]

Let me start by saying most of my field experiences have been with PAM drives. As a new engineer in the oil field several years ago, we used standard fluke rms meters to measure the voltage on the output of the drive (including digital and some old analog Simpson meters). We used a RMS clamp on meter for the current measurements.
Now I realize the complexities of measuring a PWM signal that is driving a resistive load (no smoothing of the PWM signal so all content of the signal is important). Obviously, these signals have high crest factors potentially causing errors in readings with some meters. I would like to discuss some REAL world experiences with meter errors as they pertain to drives. I guess I am of the opinion that no most meters can not measure the voltage of a 'raw' PWM signal. However, given the load (motor load), the motor acts to smooth out all of these fast rising edges of the PWM signal. I mean this is why we can use such signals for motor loads because the waveform is integrated enough to look like a sine wave.
Lets assume you have a fluke 87 (common field meter), I would like to discuss the errors of this meter on a PWM VFD with a max carrier frequency of 20kHz (ones I have seen are often ran at much lower carrier freq).
Also, with the meter technology in mind, what methods do the drive manufacturers implement for voltage and current measurements that is different than a fluke 87?

 

[Skogsgurra]

Thanks for the invitation buzzp!

There is a lot of confusion going on regarding PWM motor voltage. There are several reasons:

1 Many people still struggle with the meaning of rectified average, RMS and True RMS and think that TRMS tells the truth in all circumstances.

2 There are several different inverter technologies and some of them (current source inverters, CSI) have a motor voltage that can be read with any voltmeter - even the cheapest rectified average type - and still give a reasonable correct answer, while a PWM inverter invariably shows too high a voltage. Regardless of AVG/RMS/TRMS.

3 Scalar (V/Hz) drives usually have a rather constant volt/hertz ratio (except for the low frequency "boost" region) while vector controlled drives have a varying V/Hz ratio, which is highly dependent on the load.

4 Salestalk. Which can make any engineer confused.

The best approach to the problem is to make sure that you understand what voltage you want to measure. In this case, it is the fundamental voltage. The sine wave voltage that the drive is set to produce at the frequency that can be read off the display.
The fundamental voltage (usually in the 0 - 100 Hz region) is only part of the total motor voltage. As can be seen on many home-pages from drive manufacturers, the PWM shape consists of a series of square waves with varying widths, hence Pulse Width Modulation, PWM.
The RMS value of a symmetrical square wave is the same as the peak value of the square wave (consider rising the signal to 2, which makes the negative values positive, and then extracting the square root, which brings both positive and negative values back to the positive and constant "DC" peak value).
So, the PWM will mostly show a high and almost speed independent value. But we want to read the RMS value of the fundamental frequency.

A simple way of doing it is to filter all high-frequency components away. That can be done with a simple low-pass filter and there are instruments with such filters built-in. You can also do a Fourier analysis of the motor voltage to find out the RMS value of the fundamental component.

A "1 Hz filter" is described in thread237-114050 and

http://assets.fluke.com/appNotes/1989104_w.pdf

describes how to do a simple Fourier analysis.

http://www.yokogawa.com/tm/wtpz/wt3000/tm-wt3000_10.htm

goes a little deeper and discusses power measurements on VFDs.

 

[buzzp]

I guess that is my argument, the motor is a filter so we dont really care about the high frequency stuff since the motor never really sees it as high frequency signals after it is smoothed by the inductive properties of the motor. Does that make sense?
I have looked at the output of a single phase PWM drive on a 500MHz scope before using a shunt (so technically looking at the current). There was no visible PWM signals only the carrier frequency showed up as little spikes. So what happened to the pulses? They must be smoothed out by the motor since current can not change as quickly as it does on a fast rising edge (such as a square wave). So how important is it to measure the frequencies above the fundamental? I say not real important since the energy at these higher frequencies is negligible compared to the energy of the fundamental. Yes these meters would miss the higher frequency stuff but I don't see the high freq stuff being very important in the overall scheme of things.

On a side note, I want to generalize your statements to make sure we are on the same page: your argument is the meters respond to well to these high freq components vs one with a smaller bandwidth so this gives erroneous results. I could see this with some meters made exceptional well. When I made an RMS measuring device, I would check the AD results to see if it was too 'high' in relation to the previous results or based on previous readings. These results would be throwed out. Also, the bandwidth was only about 1kHz.
Sorry if I am all over the place in this discussion but would you then say that an rms meter with a much lower bandwidth would be a better measuring device (for drives) than say a fluke 87 with a bandwidth up to 20kHz?

Measuring drive parameters has been an issue with me for several years and I have heard many opinions concerning measuring these parameters. I believe I have read flukes app note but I will check out the other links you gave me to see what they offer.
I am also curious to know what the mfgs do to get these measurements. I mean, some of the best rms chips only handle crest factors up to five some a bit higher. So are they incorporating a DSP (and developing their own algorithm) for these measurements or what? If they are not, then they're filtering it and that tells me they are ignoring the high freq stuff (as you are suggesting). It would be nice to look at the output of some of these drives with a spectrum analyser.
So if the high freq component is ignorable, then we should be using a cheaper rms meter with a much smaller bandwidth (or filter).

Okay Ill go away now for a bit and look at your links.

 

[buzzp]

Okay, I reiterated your statement. Obviously you agree that the high freq stuff is ignorable. But would you agree that the motor itself acts as a filter to high freqs associated with VFD's (maybe not the greatest dB attentuation but still)?

 

[Skogsgurra]

Yes, you are "all over the place".

Current measurement is one thing. The motor is a low pass filter due to the inductance. The measured current is almost pure fundamental.

Voltage measurement is another bottle of beer. You will find out from the links why that is so.

 

[buzzp]

I guess the only thing I was missing was the bandwidth of the newer meters. I can see why you may get erroneous readings with the digital volt meters with extended bandwidths.

Of course when you filter out say above 100Hz as you suggest, you are missing some of the signal. Your assuming it is negligible and it most cases it probably is. I would favor an old analog meter over a digital (with filter) so you really get the RMS value and not part of it. I guess it probably dont matter much.

Anyone know how these newer meters work, generally? I mean do they have a fast processor to calculate the rms values from the sampled data (software algortihm-square reading, add to other readings, then square root of the average)? Or are they using a temperature circuit(rms~heat)? off the shelf IC? I suppose it depends on the mfg.

Its not the drive that presents the problem with measuring the voltage its the fast meters we have nowadays with bandwidths that most people will never use.

Meter guy: put in a selectable filter

 

[cbarn24050]

Hi a simple low pass filter and a TRMS meter gives a pretty accurate reading. Try something like 3 CR networks (1K, 100nF) in series. That will kill off all but the lowest carrier but still allow all the important harmonics through.

 

[DickDV]

Am I missing something or why isn't the drive keypad being used to read output voltage frequency and current?

 

[buzzp]

This discussion is not really addressing the drive capabilities. My interest is with one drive running multiple motors. Of course, the voltage will still be the same on each motor as the drive puts out but the current will not be same. Some say a standard CT is enough to measure the current accurately, others say no.

I wanted to know more about scogs concern over using digital meters to measure the voltage. Never heard of that concern but seems valid. Sometimes it is nice to verify the drives readings while troubleshooting.

So to all, thanks for the input.