Accurate Measurment of Chopped AC waveforms

[mrickey]

A bit of background: We have a system consisting of several single-phase, phase angle fired SCRs that provide variable power to a resisitive load. We have installed devices in the SCR panels that measure the voltage, current, and power being delivered to the load (these are specifically designed to measure a chopped waveform such as that of an SCR).

What we need is a way to independently measure and verify that these are operating correctly. We have tried a couple of devices, but the voltage measurement is usually shown as several volts less than the transducers and diverges as the phase angle is increased. A good explanation to this is in this white paper.

Ideally, we would like something that can measure 3 channels at once (up to 600V, 1000A) and be able to be controlled via a network connection as the load is inaccesable when engergized for safety reasons.

Does anyone have a suggestion for such a device?

 

[gar]

101217-1531 EST

mrickey:

You have implied, but not explicitly stated, that you want to measure the RMS value of current and voltage.

If you want an RMS measurement, then use a measurement device that measures RMS.

If you strictly have resistive loads why measure voltage, current, and power?

Various instrument makers make true RMS devices. One item of importance is the crest factor capability. Another is bandwidth.

What accuracy do you require? What linearity? How monotonic? What is the purpose of your measurements?

You might find some battery powered devices with fiber-optic output to get the high voltage area data to the outside world.

Since you apparently only want to check whatever is already in the power equipment this would appear to be temporary setup. Thus, you might use fewer instruments and keep your capital costs lower.

.

 

[GeorgeB]

Ideally, we would like something that can measure 3 channels at once (up to 600V, 1000A) and be able to be controlled via a network connection as the load is inaccessible when energized for safety reasons.

Does anyone have a suggestion for such a device?

The power is a separate issue, but if the load is resistive, it can be easily calculated from V and I. I do data acquisition in electrohydraulic systems of motor amps, pressure and flow. Take a look at the products from Dataq (www.dataq.com), in particular the DI-718B-E(8 channel) of DI-718BX-E(16 channel). I believe the RMS module is already available despite their "available soon". If not, the larger and older 5B modules are definitely available in RMS for the higher $$ units.

 

[ELA]

How much time and effort do you want to expend?
Labview is a powerful tool that can be customized to your needs.

Here is a link;
http://zone.ni.com/reference/en-XX/help/371361G-01/lvwave/averaged_dc_rms/

You an hire people to develop a Labview based system for you if you like.

 

[dbuckley]

If you are using phase shift control, you can calculate the RMS voltage across the load if you know (or can measure) the time at which the SCR is turned on relative to the zero cross, and can measure the supply voltage accurately.

(No, I dont know how, not beng a mathematician, but a tall forehead explained it to me several decades ago...)

A 1000A three phase load is a mighty big load for a phase shift system - what is the load?

 

[mrickey]

Thanks for all of the replies. The application requires calibrated voltage, currnet, and power (this is a testing setup that supplies power to heating elements for thermal testing). The load is *mostly* resistive, but there is some reactive componenets to it, and it is non-linear. Most of the solutions that have been shown are average RMS which will not work per the white paper in my original post.

We currently have a Hioki 3196 that measures average RMS and the value obtained is significantly different from the values obtained by the "True RMS" devices.

We need to maintain an accuracy of +-1.8V rms with good repeatability and linearity. As this is for calibration, this is definitley a temporary setup.

Thanks again.

 

[Besoeker]

FWIW

For measurements of this kind I use a digital storage oscilloscope that I can download the data from usually in *.csv format and then drop it into a spreadsheet.
From there I can do the calculations for power, rms values, harmonic content etc.

 

[gar]

101220-1205 EST

mrickey:

To me your use of the two words "average RMS" does not convey your intent.

If I connect a true RMS meter to my main panel and sample this every 2 seconds, then by the nature of the system I will have an RMS reading that fluctuates up and down relative to some nominal value. I can average this reading over some time. This I would call an average RMS reading.

What you are probably describing by these two words is something like --- a measurement based on the average value of a full wave rectified signal calibrated to read the RMS value on a sine wave.

In simpler words maybe something like this --- average measurement calibrated RMS on a sine wave, or simpler for your purposes --- average reading.


What do you mean by

We need to maintain an accuracy of +-1.8V rms with good repeatability and linearity. As this is for calibration, this is definitley a temporary setup.

Apparently this is with a full scale of 600 V. That would be an accuracy of +/-0.3% of full scale. If full scale was 1000 V, then the accuracy needed is +/-0.18%. So fundamentally you need a +/-0.1% accuracy instrument.

Do you need repeatability and and linearity better than the accuracy.

For some reference: Fluke lists the 289 AC voltage accuracy (true RMS) at 0.4%. This might be as bad as +/-0.4% of full scale. The DC accuracy of this meter is much better. I suspect the true RMS sensor is basically an analog device and thus the reason for the lower accuracy.

Look at
http://cp.literature.agilent.com/litweb/pdf/5968-0162EN.pdf
At 60 Hz the accuracy is +/-0.04% of reading + +/-0.02% of full scale. So I believe at 600 V on the 750 V range the accuracy would be +/-0.39 V.

You have to study the vendors specs closely to really understand their meaning. Then you may want to use an independent means to verify the claims.

You are asking for relatively good accuracy for an RMS responding instrument.

Any instrument with a display will have some sort of averaging function performed within it.

.

 

[ELA]

Thanks for all of the replies. The application requires calibrated voltage, currnet, and power (this is a testing setup that supplies power to heating elements for thermal testing). The load is *mostly* resistive, but there is some reactive componenets to it, and it is non-linear. Most of the solutions that have been shown are average RMS which will not work per the white paper in my original post.

Thanks again.

What Gar has said.

Reread your white paper. Many of the solutions here will give you what you wanted (maybe not as accurate as you hope for). The solutions presented are calculation based "TRUE RMS" solutions - not by filtering the signal into an average value and applying a correction factor on the assumption of a sine wave.

 

[PetrosA]

I'm pretty sure I understand what you mean by True RMS vs. averaging RMS but what I'm seeing in the Hioki link you posted is that that model is listed as True RMS. Regardless, I think Besoeker is probably correct that you may ultimately need to use an oscilloscope for your readings.

I would recommend you contact people at Agilent, Fluke and/or Tektronics, describe your needs and see what each of them offers for advice. I've met some of the people from Agilent's handheld division and if they're any indication of other divisions of the company, you should find some very helpful people.

Good luck and let us know how things go.