Current Probe Calibration, DIY 1:100 Amplifier [SOLVED]

[HighVoltage]

I was looking for a way to calibrate / check the calibration of my current probes and thought that this will be very easy. Well, it seems I was wrong.

As a current source I am using my Clark-Hess Model 828 calibrator with a 100mA or a 1A output at 50, 60 or 400 Hz. The AC output is verified by a Keysight 34470A.

As an amplifier I am using a 50 turns or a 100 turns loop of 0.5 mm insulated copper wire.
In theory I should get 10.00A or 100.0A output at the scope from the current probe with the 100 loop system.

But the problem is:

At an input of 100mA, the output is 10.4A with 100 turns, so about 4% high.
No matter what I do, I am getting an error of 4 to 5%

As you can see in the attached pictures, I tried a few different arrangements of the windings but it makes no difference. I tried different current probes and scopes and it also does not make any difference. The reading is always significantly 4% to 5% high.

The big question I have is why do I have a 5% error?
What is it that I am overlooking?

Thanks for any advise.

[Andreas]

Hmm,

the last time I had this the battery from the current clamp was near empty.

with best regards

Andreas

[HighVoltage]

Hello Andreas,

Thanks, but I tried three different current probes, all with new batteries.
Always the same result, about 5% high and repeatable by all 3 probes.

[Damianos]

Just some thoughts:
Did you tried to measure the output of the probe with a voltmeter?
Did you tried to measure the signal with the probe without "amplification" or with a few turns, like 5 for example?
Has the signal distortion, that is not easy to see on the screen? An FFT may show harmonics...
What measures the scope if the probe is replaced with a shunt resistor?

[capt bullshot]

Hmm, it's not clearly to see, but I think you've got some DC offset on the scope. This will result in a higher RMS value calculated by the scope than expected.

[khs]

Maybe there are different ranges / gains of the internal amplifier of the probe.

I would vary the current from 1 to 100 mA.

Maybe you can see some nonlinearities.

[fcb]

Current clamp probes are sensitive to the absolute position of the wire through the centre of the probe (video below) - that might account for a few %. Another possibility (unlikely!) is that you have too many coils - probably as likely as your 34470A being off..

[Andy Watson]

Have you taken steps to limit the bandwidth of the signal from your probe? The meter probably gives up at 300kHz but the scope goes up to 350Mhz - that's a lot of extra bandwidth for noise to be added to your signal.

[HighVoltage]

Thank you for all the replies.

I tried the following so far:

- Slower scope (100 MHz and 200 MHz): Same results
- DMM measuring AC voltage output from current probe instead of scope: same results
- Used completely different brand current probe: same result
- The input signal is confirmed with another 34410A in series with the 34470A
- Moving the probe around all the copper wires does not make a difference.

Hmm, DC on the output of the calibrator, overlaying the AC is a possibility, I will check that tomorrow.
I will also try only 10 windings

So far, only if I reduce the windings from 100 to 96, will I get an ok reading, confirming again about 4 to 5 % error.

It seems I am overlooking something obvious, really weird.

[dl1640]

try another gauge of wire?

[VintageNut]

Core saturation?

[BNElecEng]

I'd second the DC offset suggestion. The pictures of the sine waves show approx. 3.875 Divisions to the top and 3.625 Divisions to the bottom, so there's a difference of about 0.25 Divisions. Assuming it is offset, then taking half that value multiplied by the actual Amps per division gives the DC offset value. In the case of the 5.3A rms measurement, it's 2A per Div which gives an offset of 0.25 A rms. That works out to 5% error right there.

I hope this helps.

[HighVoltage]

Thanks for all the great feedback.

OK, problem solved !

I tried 1mm copper wire instead of 0.5mm and no difference
DC offset, measured with a 34470A was about 3 mV DC, so not really a problem.
The scope was AC coupled anyways and the bad readings were confirmed by a DMM

All three of these current probes that I had tested are calibrated by an official lab, therefore I did not expect a problem with the probes, especially since they are very close to each other on the readout.

BUT ... I was so wrong with this assumption.

In the pictures below, I am testing with 10 loops of copper wire (1 mm diameter) and all 4 probes are hooked up on the same loop:

- Channel 1: Tektronix A622, 1.047A (4.7% Error)
- Channel 2: Keysight N2893A, 1.006A (0.6% Error)
- Channel 3: Tektronix A622, 1.043A (4.3% Error)
- Channel 1: FLUKE 80i-110s, 1.029A (2.9% Error)

So, obviously it has nothing to do with the wires and setup, just the probes are out of calibration.

Now I can continue this project and make a few nice loop arrangements.

I am planning to make a 5, 10, 50 and 100 loop gadget and this way I would be able to test up to 1000A with a 10A input.

[SZA263]

There is something weird in your measurements. The crest factor for a sinewave is 2 * sqrt(2) ~= 2.82843. In your 100 turn photo the peak/rms values are 29,21/10,39A, Crest Factor is ~2,821; In the 50 turn photo they are 14,96/5.299A, Crest Factor is ~2,823. So I think of an high distortion of the calibrator on this high inductive load. Can you make an FFT on the output signal, to see the harmonics? I dont believe on the DC offset (how can a DC cross a transformer?), but an high value of the even harmonics can be seen as an offset.
I hope this helps.

Regards,
Adriano

[RobK_NL]

There is something weird in your measurements.

No, there isn't. A scope is not a precision instrument and the differences you point out are in the order of 0.25%. That's about as close as you can get!

[HighVoltage]

Ok, it is confirmed that the problem is with the probes.

Two original Agilent / Keysight probes are almost spot on with 0.1% and 0.8% error.
And the tektronix and Fluke probes are 4.9 and 3.4 % off

I guess, this means: "You get what you paid for" although the Tektronix and Fluke probes were not cheap.
Somehow, I just did not expect such a large error.

[capt bullshot]

Two original Agilent / Keysight probes are almost spot on with 0.1% and 0.8% error.
And the tektronix and Fluke probes are 4.9 and 3.4 % off

They are all rebadged:
The larger ones are made by Chauvin Arnoux, rather lower end stuff,
The small ones are Hioki AFAIK, high end stuff. You can get similar ones from Yokogawa, also rebadged.

[HighVoltage]

Two original Agilent / Keysight probes are almost spot on with 0.1% and 0.8% error.
And the tektronix and Fluke probes are 4.9 and 3.4 % off

They are all rebadged:
The larger ones are made by Chauvin Arnoux, rather lower end stuff,
The small ones are Hioki AFAIK, high end stuff. You can get similar ones from Yokogawa, also rebadged.

Interesting, thanks
Are you sure the larger one is originally from Chauvin Arnoux?
I always thought that they originated by AEMC, the Model SL261
http://aemc.com/products/html/moreinfo.asp?id=20201&dbname=products

Depending, who is offering the re-badging, they cost  between 400 and 800 Euro
Well, I should also have read the specs first, they are listed with a +/- 3% accuracy only!
And perform worst than specs!


The Hioki, Yokogawa/ Keysight high end stuff are almost 10 fold in price!
And the Keysight is listed with a +/- 1% accuracy.
And perform better than specs!

So, what do you do, if you need better than 1% accuracy in a clamp on current probe?
May be it has not been invented?

[capt bullshot]

Two original Agilent / Keysight probes are almost spot on with 0.1% and 0.8% error.
And the tektronix and Fluke probes are 4.9 and 3.4 % off

They are all rebadged:
The larger ones are made by Chauvin Arnoux, rather lower end stuff,
The small ones are Hioki AFAIK, high end stuff. You can get similar ones from Yokogawa, also rebadged.

Interesting, thanks
Are you sure the larger one is originally from Chauvin Arnoux?
I always thought that they originated by AEMC, the Model SL261
http://aemc.com/products/html/moreinfo.asp?id=20201&dbname=products

Depending, who is offering the re-badging, they cost  between 400 and 800 Euro
Well, I should also have read the specs first, they are listed with a +/- 3% accuracy only!
And perform worst than specs!


The Hioki, Yokogawa/ Keysight high end stuff are almost 10 fold in price!
And the Keysight is listed with a +/- 1% accuracy.
And perform better than specs!

So, what do you do, if you need better than 1% accuracy in a clamp on current probe?
May be it has not been invented?

I believe AEMC is Chauvin Arnoux or whatever, see the bottom of the page:
© 1999 - 2016, Chauvin Arnoux®, Inc. d.b.a. AEMC® Instruments

The Hioki ones are quite good, better than the famous Tektronix AM503 / A6302 and others. Those were the best you could get back then.
Be careful not to break the ferrite core, this hurts the low frequency accuracy. It's quite interesting, I've got one of these from work that was very abused, the clamping mechanism doesn't lock anymore and the core is visibly damaged - It's still performing well for higher (> a few kHz) frequencies but has some percent error (reading is too low) at 50Hz. A square wave looks peaking and then settles, but the settled line has too low amplitude, the height of the initial peak is the correct amplitude. At work nobody noticed until scheduled calibration failed ...
Seems for higher frequencies less core is required for these things to work. Internally they have a high frequency and a low frequency path, the low frequency path using a Hall element for zero flux compensation through the same winding on the core that is used for the high frequency path. See the Tek manuals for all the interesting details, the Hioki work exactly the same way.

I don't know if the CA ones have a winding on their core or use the Hall element as the only sensor. Otherwise, when having a compensation winding on the core, it's quite difficult to achieve such a large error. So I guess, they are rather open loop, no wonder there's such a large error.

[dl1640]

i believe the Agilent one is just a Hioki made, hall element current probe

[tszaboo]

Fluke 80i-110s AC/DC Current Clamp
40 A to 80 A ±12% of reading +50 mA
80 to 100 A ±15% of reading
DC - 100 kHz

Tektronix A622:
No real spec. But it is going to be awful guess-o-meter type probe.

Keysight N2893A:
Accuracy (Probe Only)b ±1% of reading ±1mV (DC or 45 Hz to 66 Hz)

I dont see problems. They are all within spec. I think you are "just" expecting them to be better, while they are obviously the guess-o-meter category. Like 99% of current probes.

Try finding a probe which not only has good DC and AC specification, but good transient response. I placed a square wave on few of these probes. Some's output (fluke i400 or somethingsimilar) even started to go the opposite direction before changing it's mind and going the right way. Try investigating the transient response of your high current device with that.

[HighVoltage]

Hello NANDBlog

You are so right, I had much better expectations in the accuracy of these probes.
I had done some DC comparison before and they lined up right on spot to the Agilent probes.
Although 50 Hz or 400 Hz is not fast, it makes a huge difference already.

Good idea with to test the response on a square wave.
I will just make a setup where I can turn ON and OFF a 10A current through a precision resistor and see what the different clamp on probes will show.


 

[tszaboo]

Well, I was talking about more about the experience I had with them.

Try investigating the transient response of your high current device with that.

Good luck with that NANDBlog... Talking about the current probe that I had and the device I had to test.

I ended up building a test probe with a big shunt and a x100 or x1000 amplifier (Just an instrumentation amp).
But yes. If you want to see how bad they are, even a low frequency square wave will show it.

[Kleinstein]

If the probes use a different path for AC and DC, the cross over is likely rather low, as the DC path with a Hall sensor is not that accurate and prone to drift. So 50 Hz might very already be the AC part only.

[PTR_1275]

Has anyone made a current coil like the transmille or fluke ones for calibrating current clamps?

I only need DC, but am thinking of making something like the fluke one as it's design seems a lot simpler. Get some magnet wire, wrap it around some pvc piping, bend it around and epoxy in place with some terminals.

Or am I overlooking things?

Does anyone have a fluke cool that they can measure the copper diameter and coil diameter?

Cheers
-Chris

[HighVoltage]

I am working on a design with a few different dimensions.

- One with smaller wires to accommodate the smaller opening of the Keysight probes (1147B) and
- One with heavier wires for the larger probes.

[PTR_1275]

I'd be interested to see your design. I really like the transmille one with its 3 different coils.

I don't have a current reference so my accuracy is not going to be anything great, but having a better way of checking clamps at higher currents would be nice.

I was honking of putting a DPDT switch in mine so that I can make the current flow one way or the other for checking the difference between positive and negative currents easily.

Once again, I'm only thinking about DC at the moment, but would be interested to see how well it goes on AC and how to test for any problems.

[HighVoltage]

Here was my first attempt for a simple design for 20 loops of 2.0mm diameter copper wire.
Just two of the same parts, mounted together
- The inner opening height will be 24 mm.
- The tube diameter will be 10 mm with a 8mm hole in it
- Material: Black PE
I think I will just place an order in the machine shop and see how good it works and then improve on it.

This one is for the larger current probes and will not fit the Keysight 1147B

[bopcph]

Its funny, I've been working with the Tektronix AM503 / A6302 setup some years ago, and my experience is that they are quite accurate.

A well maintained and calibrated AM503/A6302 will easily give you precision within the limits of your scopes precision and 1-2% on your HP34401 - within the ACV limits of the 34401, crest and frequency.

We used it for work on SMPS, invaluable !, from a few kHz/100mA to 20Amps at 500kHz with risetimes that indicated well over 15MHz (fundamental) - all within 1-2%.
Remember to terminate it with 50ohms, remember to degauss, don't through it on the floor, don't use it as a hammer, don't chew on the cable, don't use it as a wrench - do I need to say more?.

Seeing how many questionable (spare)parts available on eBay inspired me to the list above 

Just handle it like you would your crystal champagne glass, those you only use for special occasions - you don't through your HP3458A on the floor a few times a day either do you - and would you trust it with more than a handfull of digit if you did ?   

[HighVoltage]

Its funny, I've been working with the Tektronix AM503 / A6302 setup some years ago, and my experience is that they are quite accurate.

I think I have a Tektronix A6302 or similar (by the looks) in a drawer without the amplifier, directly to BNC
This one might has the smallest opening of all, so I need to make my design for the amplifier to fit that probe.
Thanks for that reminder!

I have never used the AM503 amplifier, may be something to look in to.

[capt bullshot]

Its funny, I've been working with the Tektronix AM503 / A6302 setup some years ago, and my experience is that they are quite accurate.

I think I have a Tektronix A6302 or similar (by the looks) in a drawer without the amplifier, directly to BNC
This one might has the smallest opening of all, so I need to make my design for the amplifier to fit that probe.
Thanks for that reminder!

I have never used the AM503 amplifier, may be something to look in to.

If it has a direct BNC connector, it's not the P6302/A6302. It's an AC probe, they look quite the same as the P6302 but don't have the hall element for the DC path. Maybe it's an http://w140.com/tekwiki/wiki/P6016 or the like, they can be used with a passive terminator.

[macboy]

Its funny, I've been working with the Tektronix AM503 / A6302 setup some years ago, and my experience is that they are quite accurate.

I think I have a Tektronix A6302 or similar (by the looks) in a drawer without the amplifier, directly to BNC
This one might has the smallest opening of all, so I need to make my design for the amplifier to fit that probe.
Thanks for that reminder!

I have never used the AM503 amplifier, may be something to look in to.

The P6019, P6020, P6021, P6022 probes have a BNC connector. Note however, that these all require one of the following: Its own special passive terminator/compensation box, or an external amplifier (e.g. Tek model 134), or a dedicated scope plugin (e.g. Tek 3A9 or 7A14). The probe must never be operated without one of those terminations. Plugging it directly into a 1 MOhm scope input will not terminate the current transformer secondary properly and high voltages can result, damaging the transformer and/or scope.

I have two P6021 probes, one with the passive terminator and one with a model 134 amplifier. The 134 amplifier gives better sensitivity and extends the low frequency response significantly (from 450 Hz to 12 Hz) but reduces the high end a little (from 60 MHz to ~50 MHz). One of my P6021 probes has the extra long cable option which reduces the bandwidth quite a bit.

The 134 amplifiers are plentiful on ebay.

[HighVoltage]

Thanks for the feedback on the Tek probe.
Yes, it is a P6022 it came with a proper terminator.
It is a really need little probe, just too bad it is only for AC

I think I should look for the 134 Amplifier but only see 110V versions on ebay.
Can the 134 amplifier be rewired inside for 230V?


 

[dl1640]

Here is 9100 calibrator coil, this paper is from Fluke UK

I have both this coil set and 5500 calibrator coil at work,, but both do not fit some probe clamp
Also inductance will hard to drive for higher current, on 5500 series calibrator we need to enable LCOMP
which is degrading the accuracy according to the document

I need a 100 turn coil to simulate 500A at DC and 50Hz but home made one is difficult to determine the uncertanty, but worth playing..

[HighVoltage]

Thank you dl1649, that is a very interesting paper.

Building a current amplifier like this is not as easy as I first thought.

[dl1640]

Anyway, use the x turn coil is easy to check full range of clamp meter and usually they are not so accurate device.

So maybe we can use a thru-hole CT with very good linearity and wind your coil onto it to make a fixture and calibrte them as one,
then we can calibrate clamp meter more accurately.

[tautech]

Thanks for the feedback on the Tek probe.
Yes, it is a P6022 it came with a proper terminator.
It is a really need little probe, just too bad it is only for AC

I think I should look for the 134 Amplifier but only see 110V versions on ebay.
Can the 134 amplifier be rewired inside for 230V?


 

These P6021 and 22 probes are gold, please treat them as such.
The 134 amp has an external supply.....wallwart if you like. Two versions, the 110VAC is easy to find, sadly the 234VAC not so much. The Tek part # on one I have is 015-028 and I luckily got it from member whitevamp who found a box of old Tek stuff.
Not sure if he's got more.
https://www.eevblog.com/forum/testgear/unknown-tektronix-part/
https://www.eevblog.com/forum/buysellwanted/fs-us-misc-tek-parts-and-fairchild-voltage-regulator/msg1113899/#msg1113899

[HighVoltage]

My first attempt to build a 1:20 Amplifier with this small 80 mm diameter holder just failed miserably and looked horrible.

I had no idea, how difficult it can be to get 2mm copper wires through such an opening, although in theory it should fit perfectly.

So, I started over with 1:10 and that works well to get 10 loops through the 8 mm hole.

Next is to scale it it up for 1:20 
And may be eventually 1:100, will see...

[dl1640]

So cute, I like it

[HighVoltage]

Continuing this project, I got some coils professionally built with 50 turns of copper wire as close together as possible. These insulated copper windings are forced glued during the winding process and stick together, as if they are one piece.

This is a rectangular shaped coil that I potted with epoxy in to a little holder and built an adapter box for the connection around it.

It works surprisingly well and could hold 4 of these Keysight/Yokogawa current probes together.

Here are the first test results.
 

[HighVoltage]

The 50 windings are small enough to accommodate the very small opening of the Tektronix P6022 probe.

[HighVoltage]

If anyone is interested, I have some spare parts of the copper windings, already epoxy potted in to the little holder with some short cables attached. All tested and working perfectly.

2 x 50 turns
1 x 59 turns

Euro 10 each plus shipping, send me a PM.

=====================
Edit:
Both 50 turns coils are gone.

[HighVoltage]

And here is a test with a pulse current to take the FLUKE 80i-110s to its limit of 100A,
within a pulse length of 90 μs and a primary current of 2A

[HighVoltage]

With a sine wave input of 1V RMS , the amplitude stays stable until about 1 kHz and the -3 dB point is at 26 kHz.
Here is a picture of a sweep from 1 kHz to 26 kHz.