Micsig Current Probe CP2100A/B Tests and Comparing

[Algasman]

What I was getting at is this thread is about comparing the performance of various probes against each other, including A vs  B Micsigs.  It doesn't make any sense to compare A with 5 turns vs B with 0.  Seems that the best is to compare no turns on every probe.

Right, comparison makes only sense if the conditions are equal. That’s what we did.

It is perfectly fine to use only one turn (just the wire) if you have a decent signal. A signal generator producing a current of only 50 mA RMS on a probe in the 10 Amps range did not give me a good enough S/N.

That’s why I multiplied the input signal buy using turns and adjusting the input signal to read the 2.5 volts RMS at the load resistor, compensating for the insertion impedance.

[Algasman]

The official manual for the CP2100 does not give any noise values. Has anyone measured values for the peak to peak and RMS noise values for the probe?

I finally had the chance to measure noise (empty clamp).
With almost 5mA RMS of noise, measuring a 50mA RMS Signal (as in my test case) with just one turn is more guessing than measuring.

[Martin72]

I´ve used the averaging function of my scope in that range (or below, 20mA/div), then it works pretty well.
Still my probe is at work and at last, I´ll do some real measurements next week, working on a power inverter.
Last friday done a quick test, using a fluke87V and an bnc adaptor - Under 10A it got a remarkable accuracy in comparison to a ZES LMG670 poweranalyzer.
Will check this further, posting the results here.
Although I´m very confident with it, this one still makes me curious:

https://de.aliexpress.com/item/32827427938.html

Good, it will cost nearly double but still miles away from the stuff of hioki and it´s "brands"....

[Algasman]

I´ve used the averaging function of my scope in that range (or below, 20mA/div), then it works pretty well.
Still my probe is at work and at last, I´ll do some real measurements next week, working on a power inverter.
Last friday done a quick test, using a fluke87V and an bnc adaptor - Under 10A it got a remarkable accuracy in comparison to a ZES LMG670 poweranalyzer.
Will check this further, posting the results here.
Although I´m very confident with it, this one still makes me curious:

https://de.aliexpress.com/item/32827427938.html

Good, it will cost nearly double but still miles away from the stuff of hioki and it´s "brands"....

Looks nice, but if the stated sensitivity is true, then you get only 20mV at full scale with ~1mV RMS noise on top. I hope that’s not true.


But anyway, the CP2100A is good value vor money.
What makes me curious is that this that this probe and all the fancy ones have a degaussing function whereas our Micsig and the other economy probes have not.
If it is so important to get rid of the residual magnetization shouldn’t the economy probes also be degaussed from time to time.
If yes, how?

[Martin72]

Hi,

What makes me curious is that this that this probe and all the fancy ones have a degaussing function whereas our Micsig and the other economy probes have not.

I think it´s because of the (DC- current)measuring method, the ICP and also Tek/Hioki got the degaussing function on board.
The current compensating method "eliminates" the magnetic field caused by the current flow by using a compansation winding on the core, so they can also use these windings to demagnetize the coil.

If it is so important to get rid of the residual magnetization shouldn’t the economy probes also be degaussed from time to time.
If yes, how?

See the uni-t ut210e clampmeter thread, there are several possibilities described or look for a watch-demagnetizer in ebay or amazon ( should also work).

This is an still open question in this thread I try to answer it in the next days, by testing the micsig with maximum dc current, if the core gets magnetized after this, you´ll get a "huge" offset until it will be demagnetized.

[Algasman]

See the uni-t ut210e clampmeter thread, there are several possibilities described or look for a watch-demagnetizer in ebay or amazon ( should also work).

There are some treads regarding ut210e. You mean the tread with the solder gun?

I have a long forgotten pre millennium audio head demagnetizer. Would that work as well?

[Martin72]

Hi,

Have a look on this :

https://www.eevblog.com/forum/testgear/a-look-at-the-uni-t-ut210e/msg3120776/#msg3120776

And the following posts after.

[Martin72]

Hi,

At last I could doing the test on "real current" at work today...

I´ve measured several AC and DC currents on a power inverter (dc to ac).

Will put the values in a table, post them later in the weekend.

Values are compared to the ones of a ZES LMG670 high precision power analyzer at the same time.
Values of the micsig are read out via bnc-adaptor to a calibrated fluke 89 multimeter with mV range.

Nice:

Pressing the zero button will bring the output level nearly to zero ( some µV reading from the fluke).

Interesting:

Range is given by a maximum of 70Arms and 100A dc - When exceeding this, a sound and blink alert appears.
Starting at appx 71Arms..good - But when measuring DC, alert will appear not over 100Adc, it appears at appx 82Adc.

The good news:

I´ll let a current of appx 80Adc flowing through the clamp for say 5 minutes - After that, the offset of the clamp was only 260µV and could be bring back to zero again with no problems.

Martin

[Martin72]

Will put the values in a table, post them later in the weekend.

And here they are, "real" current measures...
As already written, I´ve used a ZES LMG670 Poweranalyzer for comparing and a Fluke 89IV for readout the micsig probe.
Don´t wonder about the random-like values, it´s a "living" test enviroment and I must get what I could get at the moment, so there´s is no 1,2,3,4 A possible - Real life work.

So we´ve done bandwith-measures, we´ve done rising time measures via the deskew fixture and finally "real" current measures instead of turning some wires through the clamp.
So my final verdict is already known.

Best from what you can get for the money - "More" will cost much more in this case, because in my opinion, only the hioki things will "beat" the micsig clearly and in every aspect.
So don´t waste your money for clamps which couldn´t do it better, but cost much more (see links in the very first post).
Personally, I need a clamp which can measure milli-amps to several amps and so I´ll still looking for an tekprobe with amplifier or giving the china-clone a try.
Until that, I could live easily with this gem here.

Noise in the mA range is no problem when you use the averaging function of your scope and normally, the micsig would already fits for everything I do.
But sometimes I´ll need a smaller clamphead.
Nevertheless, best bang for money, best bang under 1000bucks or more, depending on new or used probes.
That´s it.

Martin

[Martin72]

Looks nice, but if the stated sensitivity is true, then you get only 20mV at full scale with ~1mV RMS noise on top. I hope that’s not true.

It´s not true...

On their homepage, there are more detailed specs for it avaible.

On this, they claimed a conversation ratio of 100mV/A, with max 1mA noise....Not bad...

http://www.instrance.com.cn/En/ProductView.asp?ID=63

[EEVblog]

Micsig just said half of my new stock is 2MHz, and half is 2.5MHz. So pot luck 
They must have made the change/improvement half way through manufacturing my batch.
So I'm guessing all new units manufactured from now on will be 2.5MHz

[jragee]

Could this be the difference?

[Noy]

Yes same thing i spotted. And its something in the signalpath.. So its possible. Maybe  its same resistance like the one beside?
Didn't opend it up yet so not a safe guess..

[RobbiTobi]

Hi folks,

I would like to share some comments on the MicSig CP2100A&B current probes compared and discussed here.

First of all the CP2100A probe is a real bargain and good quality probe - I agree completely.
You all talk mainly about the bandwidth and FRF of the probes which are very similar for both (CP2100A&B) going up to approx. 3.5MHz.
That´s ok, but I am missing the focus on one important parameter of the probes here.
Just have a look at the phase-shift of these probes - this is more important than the max. bandwidth, i.m.h.o.
For what purpose you are using the current probe mainly? Probably, power/motor/switching testing applications.
So, measuring currents with a phase error of more than 10° will become critical for reliable power measurement and signal analysis.
In few words, these probes are only good for up to ~300kHz respectively - which is anyway quite impressive!
We have seen that both models (A&B) show a phase shift of 45° @ ~800kHz - upps, interesting due to the fact that the model A is marked 800kHz bandwidth.
My probe (A model) has a rise-time of about 100ns --> theoretical bandwidth of ~3.5MHz, however with phase-shift of ~5° @ 200kHz.

My personal assumption is that MicSig started this product by designing the A model first with "realistic specs".
And for marketing/sales reasons with the help of binning on cascaded filters "created" an "upgrade" for making money, but not really useful as the phase error is completely "out of order" above 800kHz.
Concluding, i.m.h.o the B model is not worth its money - just stay on the A model.

Cheers!

P.S.:
By the way, I suppose you all have checked the UT210E modifications made by Joe Smith years ago - great job Joe!
However, by changing few resistors only you can achieve a low-cost probe with a bandwidth of ~20kHz@2A and ~100kHz@20A - impressive current probe for some basic tasks!
Find a simple testing comparison with the CP2100A attached.

[Algasman]

I think the probe and it’s electronics introduce a more or less fixed delay which can be compensated by most digital oscilloscopes. If this delay is compensated to nearly zero, so is your phase shift at a given frequency. I compensated mine measuring the current and voltage of an ohmic resistor by aligning the two ac traces using the scopes delay function.
Give it a try!

[Algasman]

One the voltage and current traces are properly aligned you can use the oscilloscopes math function to calculate power or energy

[RobbiTobi]

You have to distinguish two facts:
1. the current probe has a physical signal delay, which is a constant delay as specified in the specs (here <150ns)
2. the phase-shift or phase-error is a function of frequency due to the signal processing chain (filters and more...)

You can compensate the constant signal delay i.e. by the CHskew setting of your scope - this is correct.
However, the phase-shift you can hardly compensate - only in the case you work with simple CW signals like a pure sinusoid.
As soon as you are dealing with complex signals like i.e. a square-wave you get a signal composed of harmonically related sinusoids with different frequencies (see Fourier series).
Your measured signal of the probe will get shifted and distorted due to the phase-shift error vs. frequency.
In that case you hardly can compensate the phase-shift of each sinusoid component and your procedure as described will not work anymore.

[Noy]

You are right if you want to use it for power measurments. But even then you don't know is the Phase shift from your load or current clamp..
But if you use the clamp to verify current curves for example Motor/power supply switching current through inductor you don't have to care about Phase shift inside the clamp...

Would be interesting if Martins New Toy (other thread) has a phase shift / correction..

[Algasman]

I tried to measure delay over frequency - unfortunately with the following restrictions/constraints:

Sinusodial Current of only 140mA Peak-Peak measured in 10A Range of current probe, therefore very noisy.
Proper measurement should be done with 1 to 5 Ampere signals, I don’t have a source for that.
Signals are 32x averaged
50 Ohm load resistor
Phase & Delay measurement with oscilloscope with high standard deviation on low frequencies due to noise.
Temperature also seems to have an effect on delay

I’d say this kind of test setup is only suitable for frequencies > 100kHz
The almost constant delay from 300kHz - 2MHz is impressive (bang for the buck)
The test equipment is the same as in my other posts

[RobbiTobi]

Would be interesting to see how your probe works in reverse mode.
Just repeat your FRF and delay measurement with reverse current and see if the response is still the same... if the probe is well balanced.

[Algasman]

The signal was unbiased, so it should include both. I put my gear away, but if I give it another try, I could measure the rising and the falling delay.
The 10mA offset was an oscilloscope issue, both channels were AC coupled. The offset did not affect the measurement, I checked that.

[rhb]

You have to distinguish two facts:
1. the current probe has a physical signal delay, which is a constant delay as specified in the specs (here <150ns)
2. the phase-shift or phase-error is a function of frequency due to the signal processing chain (filters and more...)

You can compensate the constant signal delay i.e. by the CHskew setting of your scope - this is correct.
However, the phase-shift you can hardly compensate - only in the case you work with simple CW signals like a pure sinusoid.
As soon as you are dealing with complex signals like i.e. a square-wave you get a signal composed of harmonically related sinusoids with different frequencies (see Fourier series).
Your measured signal of the probe will get shifted and distorted due to the phase-shift error vs. frequency.
In that case you hardly can compensate the phase-shift of each sinusoid component and your procedure as described will not work anymore.

Once it has been digitized you can correct phase and amplitude errors with DSP to a high degree of accuracy.  This is essentially no different from calibrating an analog  SA/TG or  VNA.  For an SA/TG, you store a thru cal and apply that.  Then any difference is device error.  The cost is 2 FFTs and  a multiply.  Something an FPGA can do *very* fast at typical sweep lengths.  This yields sampling as the limiting variables, sample rate and # of bits.

Take a look at this:

https://www.eevblog.com/forum/testgear/a-high-performance-open-source-oscilloscope-development-log-future-ideas/msg3326904/#msg3326904

We've had hours of discussion about what can be done via DSP on an FPGA.  I have extensive seismic DSP experience.  There is very little cost to doing two FFTs per trace window and a multiply per frequency.  And once you've got a signal in both time and frequency you can do anything quite cheaply.

Ideally you would load FPGA images from flash memory which could be loaded to the fabric in any order and with any connections so that one can form arbitrary pipelines.  This allows writing highly specialized fabrics for triggers and other cool stuff.

Have Fun!
Reg

[Peter_O]

Just got a CP2100A. For my very basic purposes this 800kHz version is fine.



The last posts about phase shifts are many months old, but nevertheless:
Here's a quick bode plot of my CP2100A, showing the phase shift.


100kHz square wave, magnitude of 200ns rise time well within spec.
(F1 blue: average n=128)

 

A very basic test with a 60W light bulb:



DMM shows 255.6 mA
Micsig shows RMS 258.6 mA




And a shot of a PC power supply starting into standby with about 20A within the first half wave. Micsig in 100A mode.

[Curious]

Apologies, waking up an old post...

Can anybody help translate MicSig's specified Accuracy, identical language for CP2100A & B.  It's entitled "DC accuracy" but then uses "Apk" which I assume means "Amperes Peak", which wouldn't be DC.

DC accuracy (typical)     
3%±50mA (10A)                          {I'm guessing this line is DC, on 10A range, but no equivalent for 100A range}

4%±50mA(100A,500mA~40Apk)  {maybe AC accuracy if Amps peak is <40Amps, up to bandwidth limit  .8 or 2Mhz}

15%(100A,40Apk~100Apk)     {maybe AC accuracy if Amps peak is >40Amps, up to BW  .8 or 2Mhz.  saturation of steel core??  so reads lower than reality}