ability of clamp-on probe to see exponentially-decaying dc current

[electricpete]

We have recently been using AEMC current probe model 651 together with a digital recorder to measure motor starting current waveforms. The current probe is identified as an ac current probe - more details here:

http://www.aemc.com/products/pdf/2113.45.pdf

When we look at the current waveforms, they look the way I would expect them to look if our measurement was properly seeing the true current. Specifically, I see an exponentially decaying dc component on each of the three phases.

But it raises a question: do we expect this type of probe to correctly indicate a motor starting current waveform which includes a decaying dc component?

I have thought through the same question before for a CT and the answer was yes. I know how to analyse a CT based on textbook equivalent circuit (but I don't know if the same applies to a clamp-on). Here is the analysis for a CT:

electricpete's analysis of CT response to signal containing dc:
Assuming a linear magnetizing branch (L), purely resistive burdren (R), with all quantities referenced to the secondary:
I1 = Im + I2 = V2/(L*s) + I2 = (R*I2)/(L*s) + I2 = I2* [1 + R/(Ls)]
I2/I1 = 1 / [1 + R/(Ls)] = s / [ s + R/L]

This transfer function I2/I1 has a zero at the origin and a pole at s = -R/L. For typical installation L > > > R, and the pole -R/L is very close to the origin.... perhaps at 0.001 sec^-1. The pole and zero being so close together means that their effects will cancel for all (complex) frequencies except those very close to the pole zero pair.

H(s) = s / [ s + R/L] transforms into a quirky "impulse response" h(t) which itself includes an impulse (delta). It's a little easier to look at the step response which is (1/s) * H(s) = 1 / [ s + R/L] or step_response(t) = exp(-Rt/L) So the system responds perfectly to the high-frequency components in the rising edge of the step, but then decays away as the steady-state (dc) response starts to kick in.

As long as the rate of decay of the motor dc component is much faster than the R/L of the CT circuit (which appears to be the case R/L of CT ~ 0.001/sec, R/L of motor decay ~ 1/sec or greater), the dc component should pass through the CT intact

So... the clamp-on probe is a little different device than the CT. The output of the clamp-on probe is voltage, not current. And what is inside of the clamp-on probe I don't know how to model. Is it reasonable to expect that the waveform we see on our scope fed from this clamp-on is a true representation of our actual current, including exponentially decaying dc offset?


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

 

[electricpete]

By the way, I emailed the current probe manufacturer tech support. Here is their response:

Hi Peter,

Please see specifications on the SR651 attached.
This is an AC probe and without the Hall effect circuitry, I can't see how it will measure the DC.
I am sending this email to our factory for further comment.
Thanks for asking.
[name deleted]

----- Original Message -----
From: electricpete
To: aemc.com
Sent: Thursday, May 15, 2008 1:23 PM
Subject: SR 651 ac current probe response to non-sinusoidal inputs during motor start

We are using the SR651 with a digital storage oscilloscope to record motor starting waveforms.

Those waveforms contain an ac component and a decaying dc component.

Is this probe expected to accurately measure a decaying dc component to give the correct waveform?

An equivalent question which will allow me to analyse for myself: If the probe were exposed to a step-increase in dc current, the probe output voltage I assume would be a step followed by an exponential decrease. What would be the time constant of the exponential decrease of probe output voltage following a step-increase in probe input dc current?

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

 

[Skogsgurra]

Hello Pete!

The spec says 30 Hz - 5 kHz. So, there's very little chance to measure a DC component longer than half a period of 30 Hz with stated accuracy. The DC decay normally takes several periods, I would use a DC clamp instead.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[electricpete]

Thanks Gunnar.

We have been looking for example at a 460 vac 60hp KVC code G motor with no permanently installed CR, we put the clamp-on around the power cable. We expect locked rotor current is in the range 527 - 593 A RMS. We expect the true peak can be up to 2*sqrt(2) higher for worst-case dc offset, i.e. 1179A - 1326A true peak.

The DC hall-effect probes that we have on-site go up to 10A. As far as I can tell, Fluke doesn't offer these above 100Adc, so dc probe doesn't seem to be an option.

http://www.myflukestore.com/c537/current_probes.php

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

 

[electricpete]

Also, I have a vague suspicion that the ac clamp-on is providing an accurate readout of both ac and rapidly-decaying dc components (a few cycles) based on the results that we got which pass a reality check (meaning they don't look impossible), and from comparison to CT, which I know provides accurate readout of quickly-decaying dc components as long as it remains below saturation (although I'm not quite positive that the response of a clamp-on voltage output will mimick the response of a CT output current). Has anyone ever measured the response of an ac probe to a step increase in dc? I would suspect that you get a step response similar to that described above for ct (step increase and then slow exponential decay). If that is the case, I believe it would accurately capture the rapidly-decaying dc components. Has anyone ever investigated this aspect of performance of ac current probes?

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

 

[Skogsgurra]

There are several that should work. I use APPA that do 600 A RMS, which is about 850 A DC and have a rather good overrange - although with reduced accuracy.

There's also the Fluke I-1010, which I haven't used, but has a 1000 A DC range. I guess there are parallel cables. So you could measure one cable and double or triple the result.



Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[ScottyUK]

LEM's PR-1030 is a reasonable probe at a reasonable price. It lacks the bandwidth of the Tektronix probes but it is rated up to 1000A AC or DC and will pass over a 30mm diameter conductor. Presumably the 1000A AC will allow it to reproduce 1400A peak. I've had mine for a few years and it has been a good tool.

I don't think the big AC probes are much more than a good CT with an accurate burden and a compensating network to stretch the frequency response out. Tek's CT-4 is a range extension CT for the smaller Tek probes. It manages a bandwidth of a few MHz and with the right probe can also go down to a few Hz. Quite a piece of design, and with a price tag to match.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[electricpete]

I provided some more discussion on this subject in
thread238-227073

One new piece of information: I did test a similar (but smaller) current probe to examine it's response to step dc increase. The results was a step increase followed by an exponential decay to zero with around 110 millisecond time constant.

I have provided an analysis of my "model" attached (same as posted on the other thread).

As I mentioned in the other thread, even if the model assumptions are not correct, the model will correctly predict the system response if it matches the step response of our system and if our system is in the range where it rmeains LTI (linear time invariant).

So I have two time constants:

Tau_motor is decay constant of the exponentially decaying component of the m otor starting waveform. I assumed Tau_motor ~ 0.020 throughout my attachment. That was estimated from a motor waveform. Corresponds to starting power factor ~ 0.13

Tau_meas is the is measurement time constant (time constant of decay after injecting step). I varied Tau_meas over a range from far above Tau_motor to far below Tau_motor

The results seem to confirm the conclusion:
1 - If Tau_meas far above T_motor, the waveform is accurate.
2 - If Tau_meas gets dowon close to T_motor, we start to see errors.

There are two noteable errors that become more pronounced as I decrease Tau_meas
Error Type 1 - The dc component of the indicated (output) waveform appears to oscillate even though the the actual (input) dc waveform has no oscillation... just decays to 0.
Error Type 2 - The peak of the indicated waveform falls short of the the true peak of the indicated waveform.

Do you agree with these general conclusions.

Another conclusion I am wandering towards: if I see these Type 1 errors in my waveofrm (oscillations in dc component of my indicated current waveform), then my tau_meas is relatively low and I should begin to suspect that I may also have type 2 errors (indicated peak less than actual)?

Note that Type 1 error depends on relationship between Tau_meas and T_motor while Type 2 error depends more on relationship between Tau_meas and line frequency I think. But for the particular parameters I have chosen (T_motor = 0.020), these errors begin to appear at similar values of Tau_meas.





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

 

[electricpete]

Forgot the attachment. Here it is.

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

 

[electricpete]

Scotty
LEM's PR-1030
Looks like it was around $250. For 3 phases that would be $750.
I guess we would break down and fork over the money to get better data, but it is no longer in production.

http://www.techedu.com/LEM_PR1030.asp

Gunnar – none of your mentioned probes would meet my application as I stated 1179A - 1326A true peak. We have measured 1100A true peak using the AEMC ac probe and I'm guessing it goes higher than that.

I googled but haven't stumbled on any more alternatives yet.

I would welcome any suggestions.

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

 

[electricpete]

By the way, if anyone is interested, I have attached a more detailed analysis of the ac-clamp-on response. Although we will look for a better probe, we have a lot of data collected with the ac probe (40+ starts) and would like to understand exactly what it means.

I do think the clamp-on is acting like a CT as was suggested by Scotty.

Step response of a similar (smaller) ac current probe is shown in figure 2, and to my way of thinking validates the model.

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

 

[ScottyUK]

Thanks for the feedback ePete. Looks like Fluke have bought LEM's probe designs and killed off the useful one!

Have you had a look at LEM's Hall Effect range? If you can maintain good electrostatic shielding (or hang the sensor on an earthed conductor so it does not see high dv/dt relative to ground) their products are pretty good for the cost.

http://www.lem.com/hq/en/content/view/269/206/

for the industrial range

and

http://web4.lem.com/hq/en/component/option,com_catalog/task,displayserie/serie,LT/output_type,/

for one I've used in the past with good results.






----------------------------------

If we learn from our mistakes I'm getting a great education!