Meggering with a GFCI in the circuit

[kwired]

I have no issues with a meg test from hot to ground or neutral to ground when a GFCI is in the circuit being tested.

Testing from hot to neutral however subjects not only the MOV installed between those two points but also the other electronics connected between those points to the test voltage. None of those elecronics or the MOV are connected to the EGC of the device so you have no path there unless you already have a failed component to start with, which is what the test is intended to find in the first place.

I might suggest only applying a 250 volt test signal to a piece of equipment with solid state components in it that is only designed to operate at a nominal 120 volts though. If you wish to test other portions of the circuit with a higher test voltage then maybe is a good idea to disconnect from the GFCI, before testing.

 

[Smart $]

Since when has max ratings become a failure point?

Exactly what do you think is the purpose for a maximum rating? Considering your response to other posts, this question is meant to be purely rhetorical.

 

[gar]

141202-1201 EST

Many plastic materials have a breakdown voltage of 1000 V DC per 0.001" short time, and sustained in the 200 ro 400 V DC per 0.001".

I believe that 0.25" over the surface has been a design critera for 120 V circuits. But this probably assumes a several thousand volt real breakdown. This will be sensitive to elevation.

Looking at a Hubbell duplex I see possibly 0.15" over the surface between a current conductor and EGC. I can probably apply several thousand volts before over the surface breakdown. Air is somewhere in the range of 75 V per 0.001". So 150 * 75 = 11,250 V. Sharp points and surface contamination will reduce the 75 V/mil.

Testing a 510 V (720 V breakdown) MOV I measure 0.015 microamperes with 60 V DC applied. Approximate resistance 4000 megohms. This will drop as you approach the limiting point.

.

 

[wptski]

Exactly what do you think is the purpose for a maximum rating? Considering your response to other posts, this question is meant to be purely rhetorical.

They are application specs.

 

[gar]

141202-1951 EST

What is the or a megger's maximum power output on each voltage range? May or may not be the same on different models, and different voltage ranges.

Maximum power in a resistive load occurs when the load resistance equals the internal resistance of the voltage source. For those that are not familar with this concept it is easy to verify with a few sample calculations.

.

 

[wptski]

141202-1951 EST

What is the or a megger's maximum power output on each voltage range? May or may not be the same on different models, and different voltage ranges.

Maximum power in a resistive load occurs when the load resistance equals the internal resistance of the voltage source. For those that are not familar with this concept it is easy to verify with a few sample calculations.

.

The AEMC 1060 is labeled as nominal 1mA so that's 1W at 1000V, .05W at 500V and so on.

 

[gar]

141202-2355 EST

wptski:

Suppose the MOV in a GFCI clamps at 250 V and we assume the 0.001 A, then power dissipated in the MOV at maximum would be 1/4 W. The MOV will easily dissipate that with no damage continuously. If the MOV starts to warm, then the resistance will decrease and power would be less.

Can the other GFCI components tolerate 250 V continuous? Possibly, but I don't know for sure.

.

 

[wptski]

141202-2355 EST

wptski:

Suppose the MOV in a GFCI clamps at 250 V and we assume the 0.001 A, then power dissipated in the MOV at maximum would be 1/4 W. The MOV will easily dissipate that with no damage continuously. If the MOV starts to warm, then the resistance will decrease and power would be less.

Can the other GFCI components tolerate 250 V continuous? Possibly, but I don't know for sure.

.

There are two types of megger tests, one is DAR(dielectric absorption ratio) which is 60 second test and the PI(polarization index) which is a 10 minute test. I think most meggers have a ON/OFF or Start/Stop button. I have a Fluke 1507 as well and it has PI/DAR hard wired functions. The AEMC 1060 has a programmed timer function, graphing with a PC interface.

A test of 250V or even 500V at the most but I pushed the extreme here.

 

[gar]

141203-0833 EST

With the megger current limit of about 1 mA and the MOV in the GFCI clipping around 210 V the voltage across the GFCI won't get higher than about 210 V.

With the megger set at 500 V and a single GFCI the resistance reading would be about 210 k. If the megger output is constant current, then raising the voltage to 1000 V would still result in a readung of 210 k, and power dissopation in the MOV would still be about 0.21 W.

.

 

[wptski]

141203-0833 EST

With the megger current limit of about 1 mA and the MOV in the GFCI clipping around 210 V the voltage across the GFCI won't get higher than about 210 V.

With the megger set at 500 V and a single GFCI the resistance reading would be about 210 k. If the megger output is constant current, then raising the voltage to 1000 V would still result in a readung of 210 k, and power dissopation in the MOV would still be about 0.21 W.

.

gar:

The reading in every range went to max resistance, at that point, I'd stop the test which took maybe 10 seconds. The max resistance at 50V is 200G ohms and at 1000V is 4000G ohms or 4 teraohms.

 

[gar]

141203-2242 EST

wptski:

You did not have the megger between "line hot" and "line neutral" terminals.

I did two experiments on a Leviton GFCI.

Experiment 1:

Full wave bridge rectifier from a 0 to 240 V 60 Hz sine wave source (utility power, sine wave is fairly good) connected thru a 1000 ohm current sense resistor to "line hot" and "line neutral" terminals. No filter capacitor. Peak voltage of the rectified signal = Ave DC reading/0.636. DC voltage measurement across the 1 k resistor needs the same 0.636 ratioing.

Following are the average measurements with full wave rectified:
15 V ... 0.5 mA .... R = 30,000 ohms
60 V ... 4.4 mA .... R = 14,000
118 V . 10.2 mA .. R = 12,000
200 V . 18.0 mA .. R = 11,000


Experiment 2:

DC Measurement, filtered DC supply.

+ to hot input terminal
10 V ... 0.28 mA ... R = 35,000
20 V ... 0.66 mA ... R = 30,000
23 V ... 1.00 mA ... R = 23,000

- to hot terminal
25 V ... 0.50 mA ... R = 50,000

I do not care to determine why the differences exist in the apparent resistance values. The differences are not important relative to meggering. What is important is that 1 mA won't hurt the GFCI. With 1 mA current limiting in the megger the voltage across the GFCI won't get very high. When the MOV is in good condition it has no bearing on the above measurements.

I am suprised that at the 200 V input, peak about 314 V, I did not see more change in the average DC current. To see how or where the MOV is clipping I need a different experiment.

.

 

[wptski]

141203-2242 EST

wptski:

You did not have the megger between "line hot" and "line neutral" terminals.

I sure did.

 

[wptski]

141203-2242 EST

wptski:

You did not have the megger between "line hot" and "line neutral" terminals.

I think I have found the reason for some of the differences. I was always testing the GFCI in the "tripped" condition. I tried it after a RESET from L-N at 50V which gave the same results but at 100V after a few seconds the voltage drops to 79.5V and the resistance fluctuates all over the place even worse at 250V at that same 79.5V again.

 

[gar]

141204-1032 EST

wptski:

Your last post. The jumping in resistance measurement is going to result from whatever output circuit design exists for current limiting, and possibly some other factors.


141204-0916 EST

If you have a circuit that includes GFCIs, and a megger with 1 mA current limiting, then I do not believe you will damage the GFCIs with 1000 to 2000 V. This assumes that internal voltage breakdown does not occur to EGC. I don't believe 2000 V will cause such breakdown.

If you connect hot to neutral the resistance should be low. With a Simpson 260 on R*10k with + to hot the reading is 70,000 ohms, and - to hot 200,000 ohms. Meter open circuit source voltage is 9 V, and older units are 6 V. Fluke 27 reads 18 k either polarity on ohms. Fluke open circuit is 0.75 V. No explanation for the differences other than the GFCI is not a linear circuit as viewed from the terminals under different measurement conditions.

Under whatever test conditions within reason the resistance of the GFCI from hot to neutral is low compared to the expected resistance of an insulator.

If GFCIs are left in a circuit and the circuit is meggered from hot to neutral the resistance should be low, and with 1 mA current limiting there should be no damage to the GFCI.

Measurement of either hot or neutral to EGC should be high if insulation is good. Measuring either hot or neutral to EGC is effectively measuring both hot and neutral to EGC simultaneously because of the relatively low resistance of the GFCI compared to a good insulator.

If the hot or neutral to ECG insulation test fails, then it is necessary to remove the GFCIs to isolate the insulation problem area. That problem could be in a GFCI.

It should be obvious that both hot and neutral are must be isolated for the test, that all GFCIs are in the non-tripped state, and there are no loads plugged into the circuit..

.

 

[wptski]

gar:

From the AEMC 1060 using their Dataview software which generates a report in PDF format of the GFCI connected L-N at 500V for 10 minutes: https://dl.dropboxusercontent.com/u/21386595/GFCI%201%20500V%2010M.pdf

 

[meternerd]

That's Hi-Pot testing in a lab. You have never used a Hi-Pot tester for anything other than to destroy what's under test? We used a Hi-Pot tester on special motors that turned 38,000 RPM.

In the utility industry, a Hi-Pot test is considered a destructive test. It tends to grow "trees" in the insulation, which will lead to a premature failure. Only performed on a cable that has faulted. New cable is never field tested above rated peak voltage. Same deal with a "thumper" which applies short term, high voltage to a faulted underground cable so that an audible arc can be heard with a microphone. You can discover the fault location and repair it, but you can also expect another failure in the near future. I would think MOV's would slowly lose rating if they carry "fault voltage" very often. I've seen the blackened remains of a lot of 'em.

 

[wptski]

In the utility industry, a Hi-Pot test is considered a destructive test. It tends to grow "trees" in the insulation, which will lead to a premature failure. Only performed on a cable that has faulted. New cable is never field tested above rated peak voltage. Same deal with a "thumper" which applies short term, high voltage to a faulted underground cable so that an audible arc can be heard with a microphone. You can discover the fault location and repair it, but you can also expect another failure in the near future. I would think MOV's would slowly lose rating if they carry "fault voltage" very often. I've seen the blackened remains of a lot of 'em.

You've seen the blackened remains of a lot of 'em in what device?

 

[gar]

141204-2120 EST

wptski:

Relative to your test data:

What was being measured? Do the two different plots relate to the same part?

What does the 500 V setting really mean?
If it is the voltage applied to the sample under test, then why does the plot clip at 200 V?

What does the tabulated value 78.14 V mean?

Are the plots an actual instantaneous waveform, or a connected point plot with straight lines between sample points?

If you measured a 50 ft extension cord what would the plot look like?

.

 

[wptski]

141204-2120 EST

wptski:

Relative to your test data:

What was being measured? Do the two different plots relate to the same part?

What does the 500 V setting really mean?
If it is the voltage applied to the sample under test, then why does the plot clip at 200 V?

What does the tabulated value 78.14 V mean?

Are the plots an actual instantaneous waveform, or a connected point plot with straight lines between sample points?

If you measured a 50 ft extension cord what would the plot look like?

.

It's reading from the hot side to the neutral. It set for 500v, normally the output will be no less than the setting but sometimes it output a bit over. In this case because of GFCI's electronics it was 78.14V. Both graphs are the for the same test only second one is graduated in 10G instead of 20G.

Where do you get 200V from? Do you mean 200G ohms? It's a connected point plot, I believe as it looks that way. A 50 ft extension cord would be mostly a flat line except at the start. I have one at 250V where it started low for a short period of time then near flat lined at the max or 1000G for that range and the voltage stayed at 250V. The max for the 500V range is 2000G so that plot shows that it never got near max.

I have a circuit in my home that has a GFCI on it which I can test. I have tested other circuits with no GFCI on which I saved tests from that I can post.

 

[gar]

141205-1939 EST

wptski:

I had not looked at the Y axis labeling.

Now looking at it as resistance plot it does not make sense.

In the GFCI one can ignore the MOV for voltages less than the 1 mA conduction point of the MOV. Below the 1 mA voltage point (Vp) the MOV resistance is greater than Vp/0.001 . Assuming Vp is 220 V for a GFCI MOV, then R = 220/0.001 = 220,000 ohms.

The various resistive components of the GFCI at a voltage below Vp for the unit I tested were way below 220 k.

What does your megger do if you test a 100 k or 10 k resistor?

What resistance does a VOM read on your GFCI?

A GFCI is not a simple resistance, but it would seem for a megger test it would somewhat simulate a resistance.

We need to know more about how your megger works to try to understand the plots.

.