Why are current probe readings so inaccurate?

[treez]

We are using the TCP303 current probe with TCPA300 current probe amplifier to measure DC average current. We are connecting the TCPA300 into a Lecroy Wavesurfer oscilloscope.

The readings are coming out much more inaccurate than the “+/-3% of reading” suggested in the current probe datasheet.

The current waveform that we are measuring the average of is as in the attached waveform diagram.
Can you confirm that our accuracy should be within 3%, even though we are well below the 150A maximum limit of the TCP303 current probe?

TCP303 current probe and TCPA300 Curent probe amplifier:-
**broken link removed**

Lecroy Wavesurfer 44MXs scope:
https://teledynelecroy.com/oscilloscope/oscilloscopemodel.aspx?modelid=4681&capid=102&mid=504

 

[KlausST]

Hi,

You complain about measured value accuracy....but you don't give any measured value (besides scooe picture), no average value and you don't give any expected value.

So we are not able to compare measured with expected values.

Klaus

 

[FvM]

Presumed you are operating current probe correctly:
- 50 ohm termination for probe amplifier
- probe degaussed and properly locked

You can expect better accuracy than worst case +/- 3%. At low current levels, offset and noise are the only expectable additional errors. Your measurement suggests that offset and noise aren't a problem.

-> Why are you thinking the measurement is wrong? Couldn't it be wrong expectation about the measured quantity?
-> It's very easy to check the probe calibration with a power supply and an amperemeter.

 

[treez]

We think our reading is wrong because our measurement, in conjunction with a voltage measurement across the capacitor, suggested that an smps that charged the 1.5mF capacitor from 60v to 300v delivered 80w over the charge interval (1 second), whereas 0.5*c*v^2 calculation says it should have been around 64W over the charge interval.

These values are too different.

 

[FvM]

Do you remember the previous discussion about electrolytic capacitor energy content and the 1.1 to 1.5 factor stated in applications notes? https://www.edaboard.com/threads/362265/#post1551889

Besides this simple explanation, there can be other reasons for getting different results. Probe error would be my last assumption.

 

[treez]

Do you remember the previous discussion about electrolytic capacitor energy content and the 1.1 to 1.5 factor stated in applications notes?

Yes i remember that, thankyou for giving it.
In the end in that case i got a more realistic answer using a sense resistor instead of the current probe.........i then declared that the current probe may be faulty, but the boss just said that it was because it was a 150A current probe and i was using it to measure under one amp...and therefore he declared, the current probe was inaccurate and not faulty.

It seems that the dc capacitance is only about 10% more than the AC capacitance, and our current probe was making it seem way more that that.

 

[FvM]

then declared that the current probe may be faulty, but the boss just said that it was because it was a 150A current probe and i was using it to measure under one amp...and therefore he declared, the current probe was inaccurate and not faulty.

This might happen if you have mistreated the instrument, e.g. run higher currents through it without power supply. The whole thing sounds unlikely to me though, and I don't believe a word unless you have verified the scale fault in a DC measurement.

 

[schmitt trigger]

Using a previous generaion of the same Tektronix probe, we once experienced a very similar problem.

Long story short.....
Turns out that someone had dropped the probe into the hard tile floor and told no one.
What happened was the the ferrite flux concentrator on the probe's tip had cracked. The probe had to be fully disassembled to notice this.

Unfortunately we never knew who had damaged the probe.

 

[mtwieg]

I don't see enough evidence to put the blame on the probe. I would do a simple test running DC current through the probe and good ammeter to compare their measurements.

If you are measuring currents far below the probe's rated maximum, then offset errors may become significant. +/-3% refers to the gain accuracy I think, which is separate from offset. But offset can be calibrated or corrected.

 

[dick_freebird]

I once had to build a semi-automated bench test
for a chip of mine that had a bunch of trimmed
pulse current outputs. Better specs than available
current probes and better than 'scope channel
specs as well. So I had to make a calibration too,
run before each channel test. I used a long roll of
50 ohm coax (TLP style) with a good tolerance
termination resistor. Measured line charge voltage
and resistor value with a 6-digit DMM, pop the load
with a mercury relay, measure the CT output and
figure the scale factor. Then on to the real testing.

This approach might let you pass judgment on the
current probe accuracy independent of whatever you
think the SMPS "right answer" might be - seems like
you need a "tie breaker" that strips away other
assumptions / beliefs.

 

[treez]

If you are measuring currents far below the probe's rated maximum, then offset errors may become significant. +/-3% refers to the gain accuracy I think, which is separate from offset. But offset can be calibrated or corrected.

Thanks, we did de-gauss the probe before each measurment and checked that the output was zero amps when there was no curent being measured