Accuracy of fluke meter to measure inrush current

[eeprom]

Hello,
I just did some tests on a 75 Hp wound rotor motor. I wanted to see what the starting currents looked like in both the rotor and stator. I build a data collection circuit with a couple CTs and a few op amps. I measured the gain directly for each of the circuits. My sampling rate was 1000 Hz. I ran each test for 1 second. I would start the sampling, then immediately hit the start button on the motor. The graphs looked very good. I could see the entire current waveform all the way from inrush to steady state.

So, during this test, my electrician friend put two clamp-on fluke meters onto each of the CT circuits which I had been using for my data collection. His measured inrush maximum was always less than what I got in my data collection. The difference varied quite a bit. The fluke measured inrush currents were 50% to 90% of the currents measured by my circuit.

Here's the question(s)... does anyone know how the Fluke measures inrush? And how accurate is it? How likely is it to get or miss the maximum value?

thanks

 

[waross]

The early electronic meters measured the peak current and scaled it to guess what the RMS may be.
Now when a motor starts there is a high inrush in the first cycle or so that varies greatly depending on the point on the wave that the switch was closed.
The old analog meters could not respond to this short peak and showed the normal starting current that follows the first cycle inrush. This is what an electrician or engineer wants to know for trouble shooting. When you are working on motors you quickly gain a feeling for acceptable starting currents and often more important, the duration of the starting current.
First cycle peak is usually worthless for trouble shooting.
Look at your current graphs after the first or second cycle and see if that correlates with the Fluke. I will be interested to know at what point on your graph you get correlation with the Fluke reading.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[catserveng]

Try this link for some info,

http://www.aikencolon.com/assets/images/Fluke/meters/clampmeters/33Xpdfs/2562862_6115_eng_a_w.pdf

In general most current clamp-ons don't process fast enough to catch the peaks during inrush. Newer clamp-on designs address this, but "standard" clamp meters usually won't give you all the info you need for a fast transition.

The Fluke forums have some very good discussions on this type topics.

Hope that helps.

 

[eeprom]

Mr. WAROSS,
That was a very intuitive suggestion. I looked at the curves and found that the fluke was measuring the current waveform approximately 0.12 seconds later than the highest peak actually occurred. I would like to add that these fluke meters are intended to read inrush. It would seem that they would react faster than 0.1 seconds. I will look on the fluke forums. That's an excellent place to see if someone knows how fast the sampling is.

thanks

 

[eeprom]

Well, what do you know... I just looked up the fluke specs and it is stated by fluke that some of their clamp on meters intended for inrush have a 100 millisecond sample time. So, they check the current every 100 milliseconds after the trigger. I will have to check the clamp meters when I get to work to see if the model number matches up with the article I just read.

thanks again to both of you for your help

 

[eeprom]

I just wanted to follow up with this. I work with a lot of people who think that Fluke is synonymous with perfections. But it seems that the Fluke does not measure as accurately as an oscilloscope or a fast data acquisition circuit. This may be obvious to many, but it is new to me. The fluke starts sampling based on a trigger, and then samples "many" data over the following 100mS. I couldn't find an exact number, but I did find a statement in the fluke forums that there were 40 samples after the trigger. 40 samples would put the sampling rate at 2.50ms, or 400 Hz.

The maximum peak of inrush is going to occur in the first cycle. That settles out very quickly, within a couple cycles. But depending on when in the inrush cycle the fluke is triggered, it might just miss all the high spots in the first few cycles. That first current surge is the magnetization current, and based on my tests, I would conclude that the Fluke is not reliable for finding the maximum value for inrush current. However it works well for seeing the locked rotor current.

 

[controlsdude]

Most flukes I used had a high peak detect where u put this setting on and measures the highest current regardless of the normal current. Did the electrician use this setting?

 

[eeprom]

Yes. He did use this setting. The problem with measuring the "highest" current with a digital sampling meter is that the meter can only detect the highest sampled current. If the sample time doesn't coincide with the highest current, then the highest current is invisible to the meter. One other benefit of using a scope and/or data acquisition circuit is that you can begin sampling before the motor is started. A meter has the disadvantage of a small memory and having to depend on a trigger.

Ideally, the sampling should be fast enough to see 10 samples per 1/4 cycle. This is just my opinion, not related to nyquist frequency. But if I do this again, I will run my sampling at 2400 Hz.

 

[electricpete]

You certainly cannot get any meaningful data on a 60hz waveform by sampling once per 0.1 sec. I am sure the fluke is capaable of sampling much faster (like 1 khz) for it's routine features like true rms (does it display true rms?).

The characterization of an inrush is somewhat ambiguous.... you have to say whether you're looking at true peak, rms over some period (what period), etc. By the way true peak is a number tha surprises people, true peak can approach 2.8 times the numerical value of the rms locked rotor current. There can also be questions about how accurately is that decaying dc comopnent transmitted thru various types of current proble. Programming a delay may just be a way to skirt these issues and give a more well defined number corresponding roughly to the locked rotor current.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[PQ101]

Which Fluke meter are you utilizing or speaking of (i.e. Fluje 43B, 1700 series, etc.). Monitoring cycle-by-cycle is the preferred method to measure the inrush, so you can see the voltage-current waveforms, timeplots, etc. Typically 1 second is a lifetime in power quality world, and most events last 1-20 cycles, so 1sec sampling rate will not be as accurate indication because the electronic circuit are not as sensitive enough to display the voltage-current waveforms. Unfortunately what you pay is what you get.

 

[eeprom]

PQ,
I think you misunderstand my test. I did not sample at 1 second intervals. My circuit sampled at 1000 samples per second, for a period of one second. I think I got plenty of resolution, although when I do this again, I'll probably bump that up to 2400 samples/sec. I used a CT with a 1 ohm resistor load. Then I took the voltage across that resistor and scaled it to -9V to +9V to maximize the range of my data acquisition card. I don't remember which Fluke I used; it wasn't mine. For this type of in-shop testing, I prefer to use my own circuits as I can build them to suit the application. In the field, it is usually more practical to just use a meter.
EE

 

[PQ101]

I'm sorry eeprom. I assume both of you measured for the same type reading: either RMS (effective) or the peak. The op-amp circuit on each Fluke type meter may be different on internal design as you know.