Calibration of Rogowski Coil at ~400 A AC RMS 50 Hz

[CosteC]

Hello fellow engineers!

Any ideas how to calibrate Rogowski coil with better than 0.5% accuracy at current levels of approximately 400 A?
I can easily generate 400 A using transformer, yet waveform is distorted as mains voltage is distorted and rogowski coil amplifies this.
As reference I can use 0.2% shunt or 0.2% rogowski coil - I haven't found any readily available precise AC current source
Voltage from both I plan to measure with 6 1/2 digit DMM, with around 60 mV it is not bad.

What do you think about powering current transformer from AC PSU - shall be cleaner/less harmonics than UPS and allows control of output current.

I am not able to use multiple turns with lower current - Rogowski coil to be calibrated is rigid one.

[jbb]

Just wanted to check: are you measuring a Rogowski + integrator setup, or just the Rogowski coil solo.

Do you need a particularly clean sinewave for the measurement? (I guess that a stand-alone Rogowski coil will emphasise the high frequency harmonics of the primary current.)

I expect that there are a number of calibration laboratories about that can do a 10 A high precision calibration; could you wind 40 turns of magnet wire through the fixed hole?

[CosteC]

I want to calibrate Rogowski coil with amplifier. No integrator, it will be in other place and will be digital.

As I mentioned I cannot use "apparent current" method with say 40 turns - Rogowski coil is rigid one (cannot be opened) - only one turn is what I have.

I need to do it in-house as part of product calibration process. Going to lab with every unit would be logistical and financial issue. Single piece would not bother me

[alm]

As I mentioned I cannot use "apparent current" method with say 40 turns - Rogowski coil is rigid one (cannot be opened) - only one turn is what I have.

What's the problem of taking a cable, connecting one end to your current source, winding it around the Rogowski coil a number of times, and connect the other end to the source?

I need to do it in-house as part of product calibration process. Going to lab with every unit would be logistical and financial issue. Single piece would not bother me

The off-the-shelf solution would be a clean sine voltage source and a transconductance amplifier. You can get them to 100A: Valhalla 2555A. Not sure about 400A.

[CosteC]

As I mentioned I cannot use "apparent current" method with say 40 turns - Rogowski coil is rigid one (cannot be opened) - only one turn is what I have.

What's the problem of taking a cable, connecting one end to your current source, winding it around the Rogowski coil a number of times, and connect the other end to the source?

Impossible.
1) it is part of larger assembly.
2) I cannot imagine production process for many pieces a day with technician putting many turns via coil in repeatable manner.
3) Hole is dimensioned for ONE turn of copper slab rated for 400 A.

I need to do it in-house as part of product calibration process. Going to lab with every unit would be logistical and financial issue. Single piece would not bother me

The off-the-shelf solution would be a clean sine voltage source and a transconductance amplifier. You can get them to 100A: Valhalla 2555A. Not sure about 400A.

Nice piece of kit, yet bit too low current. 0.4% accuracy may force another reference instrument if I need 0.5% final accuracy FS.

[alm]

Nice piece of kit, yet bit too low current. 0.4% accuracy may force another reference instrument if I need 0.5% final accuracy FS.

If you need more current and better accuracy, you can connect up to 10 of these in parallel: Fluke 52120A. You may want to be sitting down when requesting a quote, though. These would typically be driven from a voltage / multi-function calibrator like the Fluke 55xx and 57xx series, but could be driven from any sufficiently-accurate voltage source (DC or AC). This is how a cal lab would probably do it if they can't clamp it around a coil with multiple turns, which is obviously a much cheaper solution.

[CosteC]

Anything below price of house?
I am considering stabilised current source in form of AC PSU? Any serious issues with it? It is easy to verify stability over time...
Maybe there is clever method of getting rid of instabilities of mains driven current source or measuring in a way it does not matter?

[HighVoltage]

Why not using a larger diameter rogowski coil and use 40 turns from a 10A calibrator to calibrate it.

Then use this calibrated rogowski coil and compare it with first rogowski coil in comparison at a 400A transformer signal, both hooked up around the same wire.

[CosteC]

Why not using a larger diameter rogowski coil and use 40 turns from a 10A calibrator to calibrate it.

Product specific. As it goes into retro-fit product, you ask to rectify early XX century decisions on electric infrastructure

Then use this calibrated rogowski coil and compare it with first rogowski coil in comparison at a 400A transformer signal, both hooked up around the same wire.

I can get 0.2% accurate reference Rogowski coil, but what next? Two synchronised 6 1/2 meters? Current transformer powered from AC PSU? Something more subtle? At given accuracy those are not trivial questions.

[alm]

The problem with using current signals other than clean sines (like just a transformer) is that the value now depends on the frequency response of the coil. The coil around the wire probably won't be very resistive, so a sinusoidal voltage does not guarantee a sinusoidal current. The effect of this will depend on where in the spectrum the components of the signal are, the difference in frequency response of the coils, and the desired accuracy. AC is quite a bit more complicated than DC in this regard.

[CosteC]

The problem with using current signals other than clean sines (like just a transformer) is that the value now depends on the frequency response of the coil. The coil around the wire probably won't be very resistive, so a sinusoidal voltage does not guarantee a sinusoidal current. The effect of this will depend on where in the spectrum the components of the signal are, the difference in frequency response of the coils, and the desired accuracy. AC is quite a bit more complicated than DC in this regard.

I am pretty aware of difficulties. So I am looking for fresh ideas, from those who may know more in the topic.

P.S.
Actually precise measurement of large DC currents is more complex IMHO.

[alm]

Could you fit a multi-conductor cable with connector through it, for example 4x100A or 40x10A, and terminate that in a connector wired in a way to form a coil? That should make connection easier and more repeatable. Make the cable as short as possible for minimal variation in geometry.

[jbb]

OK, I now understand why multi-turn primary is a no-go.

Given you’re making a trade off between test procedure (want single primary turn) and test equipment cost (I’m not made of money either) it may be time to think about compromises.

In general, a Rogowski coil should be nice and linear because it has no magnetic core (but watch out for people swapping out any stainless / brass screws etc!). How about:

- using something like the Valhalla 2555A someone mentioned to do your per-unit calibration (and perhaps adjustment) at 100 A
- sampling a small number of parts off the line (eg first x units from the batch and then y randomly sampled units) for the time-consuming multi-turn primary test to go all the way up to 400 A

Additionally - sorry if you’re already on top of it! -  the current return paths can be quite important to a Rogowski. So your product design and per-product calibration should have a defined and mechanically stable current return path. When doing a multi-turn primary, the path the windings take around the Rogowski should mimic this.

https://valhallascientific.com/shop/repair-calibration/2555a/

[johansen]

You can use an intermediate transformer to deliver 400 amps in series with a resistor shunt through the rogouski coil from a higher voltage lower current source. Make the secondary big enough to remove the return current effects near field to the coil.

You can then compare the resistor shunt with the rogowski coil.

You can then use your fancy 16 bit ADC to compensate for the harmonics.you can also compensate for 90% of the VARs needed to do it this way  with a bank of capacitors. (Maybe 50uf at 240vac).

If your mains ac is so bad, do this test from a sine wave output UPS..

You should not need more than a few hundred watts real power. (Use a variac from the sine wave inverter to gradually load the step down transformer. No switching capacitors on line needed.)

[mzzj]

I'd use large toroid transformer and AC power source.
Budget option for AC power source would be large PA amplifier. 0.01% THD/harmonics is way better than most AC sources.
1kW 4ohm capable PA amplifier goes for 300 euros new, add hefty 4 ohm series resistor to keep it happy and you have about 16A RMS available.
IIRC 2kva toroid transformer is about 1 volt/turn so 230  primary has about 230 turns and can handle 10A. Single turn secondary gives 2300A

[gf]

I can easily generate 400 A using transformer, yet waveform is distorted as mains voltage is distorted
...
I want to calibrate Rogowski coil with amplifier. No integrator, it will be in other place and will be digital.

If you digitize anyway, can't you measure only the amplidude of the fundamental 50Hz component and ignore the harmonics? (for both, the DUT and the reference coil) A frequency-selective detector is easy in the digital domain.

[CosteC]

Digitalisation is possible on far later assembly process stage. Calibration of transducer itself also allows easy replacement. If I will calibrate with digitiser (and ADC, and amplifiers and firmware etc) it will became large bulky and far harder for service.

I am checking options regarding AC PSU..

UPDATE:
Both AC PSU and Power amplifier with transformer does not cover distortion created by transformer. As power is rather large, it is to be expected.