Megger and GFCI

[Saywatt]

Using a megger @500v, what would be the threshold # between either n-g or l-g before it tripped a gfci?

 

[ptonsparky]

You need to know the resistance at the applied voltage to allow .05 amp to flow.

E/I=R

Calculating the resistance for 500V will not give you the correct answer for a device that is designed for 120V.

Conductors that test out at either level are "bad" and should be replaced

 

[gar]

110519-1024 EDT

Saywatt:

Your question has insufficient details.

You can not use the megger on the GFCI with power applied to the megger, or at least should not.

Suppose input power was applied to the GFCI and the GFCI was in the tripped state, then you could apply the megger to any output terminal. But what would be the reason to do this with input power applied.

Note: There is no electrical connection except a very small amount of capacitance, a few pfds, between any neutral or hot terminal of the GFCI and the grounding terminal. Over the surface breakdown is probably in excess of several thousand volts.

If you want to megger the wiring on the output side of the GFCI, then trip the GFCI with the test button while input power is applied. Then remove input power and only then megger the output wires. 500 V for a test voltage should cause no problem for the GFCI if it is good. A good GFCI out of the box and connected to nothing will show very high resistance from any terminal to the ground terminal.

If in the latched state and no power applied, then the megger should not trip the GFCI.

.

 

[Saywatt]

Thanks for the replies.

Gar, I probably should have worded the question better.

What it is, is I got a call for a fairly new gfci breaker tripping periodically that feeds a few fish pond pumps. 120v.

When I got there everything was working fine.

Took an amp reading on L (only) and was reading 7 amps. Ok.

I unhooked the uf from the panel and at the receptacles that the pumps are plugged into. I megged L-N, L-G, and N-G.
L-N and L-G were fine, but N-G was reading 150k ohms.

Told the customer that the line is bad and needs to be replaced. I hooked everything back up and the GFCI breaker held. I was surprised it held.

Wouldn't you think that with only 150k R between L-N the gfci would trip?

I then took a amp reading and was getting 7a on L and nearly 9a on N.

WTF?

I'm guessing the breaker is bad too.

 

[wptski]

110519-1024 EDT

Saywatt:

Your question has insufficient details.

You can not use the megger on the GFCI with power applied to the megger, or at least should not.

Suppose input power was applied to the GFCI and the GFCI was in the tripped state, then you could apply the megger to any output terminal. But what would be the reason to do this with input power applied.

Note: There is no electrical connection except a very small amount of capacitance, a few pfds, between any neutral or hot terminal of the GFCI and the grounding terminal. Over the surface breakdown is probably in excess of several thousand volts.

If you want to megger the wiring on the output side of the GFCI, then trip the GFCI with the test button while input power is applied. Then remove input power and only then megger the output wires. 500 V for a test voltage should cause no problem for the GFCI if it is good. A good GFCI out of the box and connected to nothing will show very high resistance from any terminal to the ground terminal.

If in the latched state and no power applied, then the megger should not trip the GFCI.

.

gar:

Pressing the test button to trip the GFCI doesn't work, you'll still read a very low resistance and with a megger, it'll look like it's shorted. You have to disconnect the device. I think that has been mentioned here before.

 

[gar]

110519-2340 EDT

Bill:

When I took apart a Leviton 7899 and provided pictures of the internal construction it was clear that in the tripped state the outputs were connected to nothing.

If you measured across the input terminals with a low voltage ohmmeter there would be a multimegohms resistance from at least the MOV and some of the electronic circuit. With a megger it will be lower, but not real low. With the GFCI tripped the megger will see a very high resistance across the output terminals.

If mechanically pressing the GFCI does not trip the mechanical mechanism when there is not input power, then one simply applies power, pushes the test button, and then removes power. Now the output terminals are completely disconnected from the input and the electronics except thru the one 15 k ohm test resistor if the test button is pushed.

The Leviton 7899 test button mechanically trips the latch even with no applied power. The actual electronic testing of said Leviton occurs upon Reset. It won't close or reset if there is is a failure of the electronic test.

I do not know the specific construction of a GFCI breaker, but it should not be much different.

Saywatt:

I assume that if you read 7 and 9 A for L and N, that this was done with the GFCI breaker bypassed. If it wasn't it should have tripped. A GFCI device does not care what the ground current is. What a GFCI device does is measure the current difference between line and neutral and a difference greater than a threshold trips the device. For short durations this trip threshold has an inverse time characteristic.

.

 

[ptonsparky]

You need to know the resistance at the applied voltage to allow .05 amp to flow.

E/I=R

Calculating the resistance for 500V will not give you the correct answer for a device that is designed for 120V.

Conductors that test out at either level are "bad" and should be replaced

Oops...make that .005

I would fail the uf cable with a 150k reading between N-G conductors even though that should not technicaly allow the required +- 5 ma imbalance. Have you megged these conductors to earth? Clamp around the neutral and hot at the same time what is the reading then? Should be the 2 amp difference you indicated and if that is the case how in the world do you gain 2 amps on a neutral unless you have an additional fault on a different circuit.

Any way you look at it, the UF is bad, most likely the GFCI breaker, and you may very well have an additional problem.

 

[wptski]

110519-2340 EDT

Bill:

When I took apart a Leviton 7899 and provided pictures of the internal construction it was clear that in the tripped state the outputs were connected to nothing.

If you measured across the input terminals with a low voltage ohmmeter there would be a multimegohms resistance from at least the MOV and some of the electronic circuit. With a megger it will be lower, but not real low. With the GFCI tripped the megger will see a very high resistance across the output terminals.

If mechanically pressing the GFCI does not trip the mechanical mechanism when there is not input power, then one simply applies power, pushes the test button, and then removes power. Now the output terminals are completely disconnected from the input and the electronics except thru the one 15 k ohm test resistor if the test button is pushed.

The Leviton 7899 test button mechanically trips the latch even with no applied power. The actual electronic testing of said Leviton occurs upon Reset. It won't close or reset if there is is a failure of the electronic test.

I do not know the specific construction of a GFCI breaker, but it should not be much different.



.

gar:

That would be true on a GFCI breaker but not a branch circuit with a GFCI receptacle.

I've meggered a branch on which was a GFCI but I wasn't on the GFCI receptacle. Past practice I start on a low 50V incase I've missed something left plugged in. IIRC, this was way back, it read zero ohms. I remembed that I had replaced a outside receptacle with a GFCI. I pressed the TEST button and retested with no change.

 

[gar]

110520-2031 EDT

Bill:

The GFCI I referenced is a receptacle. The measurements I made on it are in free space, nothing connected.

When tripped there is absolutely nothing conductive connected to the output points. The only material in contact is the molded plastic enclosure. Thus, you will get only a very large resistance measurement across the output terminals when tripped. Because there is an MOV across the input terminals the high voltage (500 v) from the megger will produce some moderate current, limiter by the megger, thru the MOV because the breakover voltage of the MOV will be probably about 250 V.

My photographs at http://beta-a2.com/GFCI.html show the internal construction of the Leviton 7899

In going back thru some of the old threads I found these references:
http://cache.national.com/ds/LM/LM1851.pdf
http://forums.mikeholt.com/showthre...ght=GFCI"+"Leviton"+"7899

.

 

[wptski]

110520-2031 EDT

Bill:

The GFCI I referenced is a receptacle. The measurements I made on it are in free space, nothing connected.

When tripped there is absolutely nothing conductive connected to the output points. The only material in contact is the molded plastic enclosure. Thus, you will get only a very large resistance measurement across the output terminals when tripped. Because there is an MOV across the input terminals the high voltage (500 v) from the megger will produce some moderate current, limiter by the megger, thru the MOV because the breakover voltage of the MOV will be probably about 250 V.

My photographs at http://beta-a2.com/GFCI.html show the internal construction of the Leviton 7899

In going back thru some of the old threads I found these references:
http://cache.national.com/ds/LM/LM1851.pdf
http://forums.mikeholt.com/showthre...ght=GFCI"+"Leviton"+"7899

.

gar:

Unless I'm mistaken which is always possible. It was long time ago when I did that and maybe my memory has failed me? I'll have to test that circuit again.

 

[wptski]

110520-2031 EDT

Bill:

The GFCI I referenced is a receptacle. The measurements I made on it are in free space, nothing connected.

When tripped there is absolutely nothing conductive connected to the output points. The only material in contact is the molded plastic enclosure. Thus, you will get only a very large resistance measurement across the output terminals when tripped. Because there is an MOV across the input terminals the high voltage (500 v) from the megger will produce some moderate current, limiter by the megger, thru the MOV because the breakover voltage of the MOV will be probably about 250 V.

My photographs at http://beta-a2.com/GFCI.html show the internal construction of the Leviton 7899

In going back thru some of the old threads I found these references:
http://cache.national.com/ds/LM/LM1851.pdf
http://forums.mikeholt.com/showthre...ght=GFCI"+"Leviton"+"7899

.

gar:

IIRC, a NIB GFCI receptacle is tripped. A Pass&Seymour measured 3.3M across the L and N terminals with my Fluke 189.

 

[kwired]

Thanks for the replies.

Gar, I probably should have worded the question better.

What it is, is I got a call for a fairly new gfci breaker tripping periodically that feeds a few fish pond pumps. 120v.

When I got there everything was working fine.

Took an amp reading on L (only) and was reading 7 amps. Ok.

I unhooked the uf from the panel and at the receptacles that the pumps are plugged into. I megged L-N, L-G, and N-G.
L-N and L-G were fine, but N-G was reading 150k ohms.

Told the customer that the line is bad and needs to be replaced. I hooked everything back up and the GFCI breaker held. I was surprised it held.

Wouldn't you think that with only 150k R between L-N the gfci would trip?

I then took a amp reading and was getting 7a on L and nearly 9a on N.

WTF?

I'm guessing the breaker is bad too.

Did you meg the pond pump? When I get these types of calls the equipment that gets submerged in water is the problem probably 90% or more of the time.

 

[brian john]

While standards exist (DUH) in the field my experience meggering cables at 1000 VDC is.

New conductors when properly installed, typically will read full scale of the meter up to about 11,000 megohms. (different meters have different full scale readings. Now there are variables in this, but if you have 4-condiuctors (3-phase and a grounded conductor/neutral) the readings of all conductors to ground should be very similar.

Existing cables, obviously the higher the reading the better and hopefully conductors of the same circuit should (hopefully) be similar.

I like to see a minimum of 50 megohms.

5 ohms works on 120 VAC circuits but really should be checked closely to determine why the readings are so low.

The often touted, sometime specified 1 megohm, in my experience is waiting for failure and/or disaster.

 

[gar]

110521-0742 EDT

Bill:

If your L to N terminals are the input, then 3.3 megohms with a low voltage ohmmeter is quite possible. Should not be anything less than very high on the output terminals. If there was another MOV on the output side. then the output reading could be lower than very high.

I can't find a new MOV to test at the moment, but a 50 year old GE 5J 20S1 reads about 5 megohms with a Fluke 27. This is probably a low voltage unit rather than 120 V.

An MOV being a very nonlinear device will make the apparent resistance reading very different with different applied source voltages. Further the current limiting means in combination with the source voltage in the ohmmeter will influence the reading.

.

 

[Saywatt]

I did not meg the pumps. I will when I go back to replace the uf early next week.

I honestly thought that 150k l-n would have been low enough to trip a gfci, but simple ohms law using 120v/5ma says otherwise?

This is growing. The customer now wants me to install a whole house generator and an auto dialing device inline with the pond pump circuits (4 of them). I'm thinking of using a few nc cube relays wired in with the gfci breakers and the the voltage monitoring auto dialer. Any other suggestions? Small plc?

She's eccentric, but she pays good.

 

[Saywatt]

While standards exist (DUH) in the field my experience meggering cables at 1000 VDC is.

New conductors when properly installed, typically will read full scale of the meter up to about 11,000 megohms. (different meters have different full scale readings. Now there are variables in this, but if you have 4-condiuctors (3-phase and a grounded conductor/neutral) the readings of all conductors to ground should be very similar.

Existing cables, obviously the higher the reading the better and hopefully conductors of the same circuit should (hopefully) be similar.

I like to see a minimum of 50 megohms.

5 ohms works on 120 VAC circuits but really should be checked closely to determine why the readings are so low.

The often touted, sometime specified 1 megohm, in my experience is waiting for failure and/or disaster.

Brian, isn't the 1meg rule mainly for motors? I have a neta chart that I go by for cables.

Also, If these pumps are so critical to her, I'm thinking about offering to trend these pumps for her.

The megger I have is no good for that. I was considering picking up a 1507 or an ideal 61-797. Any thoughts on those two?

 

[kwired]

I did not meg the pumps. I will when I go back to replace the uf early next week.

I honestly thought that 150k l-n would have been low enough to trip a gfci, but simple ohms law using 120v/5ma says otherwise?

This is growing. The customer now wants me to install a whole house generator and an auto dialing device inline with the pond pump circuits (4 of them). I'm thinking of using a few nc cube relays wired in with the gfci breakers and the the voltage monitoring auto dialer. Any other suggestions? Small plc?

She's eccentric, but she pays good.

Depending on how important it is to notify that pumps are not working, a pressure switch in the pump discharge will alert you of more than just power losses.

 

[wptski]

110521-0742 EDT

Bill:

If your L to N terminals are the input, then 3.3 megohms with a low voltage ohmmeter is quite possible. Should not be anything less than very high on the output terminals. If there was another MOV on the output side. then the output reading could be lower than very high.

I can't find a new MOV to test at the moment, but a 50 year old GE 5J 20S1 reads about 5 megohms with a Fluke 27. This is probably a low voltage unit rather than 120 V.

An MOV being a very nonlinear device will make the apparent resistance reading very different with different applied source voltages. Further the current limiting means in combination with the source voltage in the ohmmeter will influence the reading.

.

gar:

Yes, the on input terminals. If your meggering upstream from a GFCI receptacle that's what you'll read when in theory it should read the max possible on your megger.

 

[gar]

110521-0914 EDT

Saywatt:

You need to disconnect any output wires from the GFCI breaker.

Trip the breaker. Remove power to the breaker. Meg both L-G and N-G at the breaker output. I expect L-G to be very high, full scale of the megger, assuming there is no output side MOV. I have not seen the inside or tested a breaker, but my guess is that neutral is not switched, and thus, would read virtually 0 ohms to ground.

With both the line and neutral wires disconnected at the breaker then megger the circuit with and without the motors connected at the destination end of the circuit.

If you see no problem from the megger data, then reapply power to the circuit. This would be from a non-GFCI breaker. Measure the line and neutral currents again. Previously you had a 2 ampere difference. Also as ptonsparky said put the clamp-on ammeter around both the line and neutral. If the difference of the individual lines is still 2 A, then this reading of the pair of wires should be about 2 A. If you still have this disparity, or anything over about 5 mA, then disconnect the motors and repeat the measurements.

The results will determine the next step.

.

 

[Saywatt]

110521-0914 EDT

Saywatt:

You need to disconnect any output wires from the GFCI breaker.

Trip the breaker. Remove power to the breaker. Meg both L-G and N-G at the breaker output. I expect L-G to be very high, full scale of the megger, assuming there is no output side MOV. I have not seen the inside or tested a breaker, but my guess is that neutral is not switched, and thus, would read virtually 0 ohms to ground.



With both the line and neutral wires disconnected at the breaker then megger the circuit with and without the motors connected at the destination end of the circuit.

If you see no problem from the megger data, then reapply power to the circuit. This would be from a non-GFCI breaker. Measure the line and neutral currents again. Previously you had a 2 ampere difference. Also as ptonsparky said put the clamp-on ammeter around both the line and neutral. If the difference of the individual lines is still 2 A, then this reading of the pair of wires should be about 2 A. If you still have this disparity, or anything over about 5 mA, then disconnect the motors and repeat the measurements.

The results will determine the next step.

.

Thanks. I am anxious to get back over there to investigate the 2a disparity.

Depending on how important it is to notify that pumps are not working, a pressure switch in the pump discharge will alert you of more than just power losses.

That is a great idea! Thank you.