GFCI's do not need an EGC (moved from another thread)
- Thread starter quogueelectric
- Start date
- Status
gar
Senior Member
- Location
- Ann Arbor, Michigan
- Occupation
- EE
080739-2251 EST
I agree with mivey that the EGC is not needed to function or test a GFCI.
In fact the Leviton 7899-1 GFCI that I recently purchased to test, see post #126, does not even need any power to the GFCI in order to have the so called TEST button trip the device.
Enough of the respondents to this thread have said the same thing. Namely that you only need a small unbalanced current between the hot output and neutral output to trip the GFCI. There is no use of the EGC terminal in the device except to ground the metal mounting bracket and in turn the metal box.
From a mechanical design perspective this is a very nice device. Electronically it may work well. On the negative side there did not appear to be any flux removal from hand soldered high current areas, and there is very little over the surface path length between the input hot and neutral.
There appears to be a 15 K resistor, only about 1/8 W, connected thru a mechanical contact that is normally open. Pushing the TEST button may close this contact. Closing the contact would place the resistor from one input line to the opposite output line. This should electrically trip the electronics if the TEST button causes closure of the contact. But from the test with no power applied to the GFCI and the operation of the TEST button causing mechanical tripping I do not see this as an effective test of the electronics in combination with the coil as an unlatch function. Maybe there is something I do not understand of the mechanics at this time.
The device is built such that when put together I can not see the operation of the latch operation.
At this time I can not have confidence that the TEST button really tests what I want tested.
Although this is a 20 A device there is only a small silver contact on each side of the unit. This is about 1/8" in diameter. There are actually two contacts on each side but only one is used.
I am interested in knowing if Hubbell units are built this same way for test.
Clearly if you want to test GFCIs and they are all built this way, then you need something like a 12,000 ohm or some what larger resistor to do your own test. If you do not know how to do such a test, then ask and I or someone else will explain.
Has anyone else torn one apart and analyzed its operation?
.
I agree with mivey that the EGC is not needed to function or test a GFCI.
In fact the Leviton 7899-1 GFCI that I recently purchased to test, see post #126, does not even need any power to the GFCI in order to have the so called TEST button trip the device.
Enough of the respondents to this thread have said the same thing. Namely that you only need a small unbalanced current between the hot output and neutral output to trip the GFCI. There is no use of the EGC terminal in the device except to ground the metal mounting bracket and in turn the metal box.
From a mechanical design perspective this is a very nice device. Electronically it may work well. On the negative side there did not appear to be any flux removal from hand soldered high current areas, and there is very little over the surface path length between the input hot and neutral.
There appears to be a 15 K resistor, only about 1/8 W, connected thru a mechanical contact that is normally open. Pushing the TEST button may close this contact. Closing the contact would place the resistor from one input line to the opposite output line. This should electrically trip the electronics if the TEST button causes closure of the contact. But from the test with no power applied to the GFCI and the operation of the TEST button causing mechanical tripping I do not see this as an effective test of the electronics in combination with the coil as an unlatch function. Maybe there is something I do not understand of the mechanics at this time.
The device is built such that when put together I can not see the operation of the latch operation.
At this time I can not have confidence that the TEST button really tests what I want tested.
Although this is a 20 A device there is only a small silver contact on each side of the unit. This is about 1/8" in diameter. There are actually two contacts on each side but only one is used.
I am interested in knowing if Hubbell units are built this same way for test.
Clearly if you want to test GFCIs and they are all built this way, then you need something like a 12,000 ohm or some what larger resistor to do your own test. If you do not know how to do such a test, then ask and I or someone else will explain.
Has anyone else torn one apart and analyzed its operation?
.
gar
Senior Member
- Location
- Ann Arbor, Michigan
- Occupation
- EE
080730-2303 EST
I just put some of the pieces back together and I have found that the extra power switch contacts contact the back output terminals for additional outlets. The other contacts mentioned in my previous post only supply the slots in this GFCI receptacle.
.
I just put some of the pieces back together and I have found that the extra power switch contacts contact the back output terminals for additional outlets. The other contacts mentioned in my previous post only supply the slots in this GFCI receptacle.
.
quogueelectric
Senior Member
- Location
- new york
CLUCK CLUCK CLUCK CLUCK CLUCK!!!!!!!!gar said:
Come on for goodness sake grab it to ground without the ground hooked up you are already set up you are sooooooooooooo close. Dont go across the chest like I did just go across the hand . And just for the record once I hooked up the ground it tripped without shock or incident as soon as the trigger was pulled with no sensation of shock whatsoever.
gar
Senior Member
- Location
- Ann Arbor, Michigan
- Occupation
- EE
080731-0648 EST
quogueelectric:
I believe in your example you had a drill motor with a metalic housing connected to the EGC (ground wire) in a three wire cord with a three prong plug.
When the ground terminal of the GFCI was floating the fault current from the drill contacting a hot wire flowed thru you to the wet floor. There was nowhere for it to go at the GFCI and it flowed thru you.
After you connected an EGC conductor from the main panel to the ground terminal on the GFCI, then you provided a path for the fault current to the drill to its case to the drill's EGC, to the drill plug ground pin, to the GFCI ground pin, and to the EGC back to the main panel. I would not expect you to get a shock in this instance because the housing of the drill probably only rises a few volts above the earth potential.
Whether the GFCI tripped and/or the main breaker tripped is not clear when you had the EGC ground terminal correctly connected (meaning connected).
What I have just said has been mentioned in previous posts.
With a non-GFCI receptacle and the ground pin correctly connected to an EGC to the main panel I would not expect you to get a shock. Thus, the GFCI probably had nothing to do with whether or not you got a shock.
.
quogueelectric:
I believe in your example you had a drill motor with a metalic housing connected to the EGC (ground wire) in a three wire cord with a three prong plug.
When the ground terminal of the GFCI was floating the fault current from the drill contacting a hot wire flowed thru you to the wet floor. There was nowhere for it to go at the GFCI and it flowed thru you.
After you connected an EGC conductor from the main panel to the ground terminal on the GFCI, then you provided a path for the fault current to the drill to its case to the drill's EGC, to the drill plug ground pin, to the GFCI ground pin, and to the EGC back to the main panel. I would not expect you to get a shock in this instance because the housing of the drill probably only rises a few volts above the earth potential.
Whether the GFCI tripped and/or the main breaker tripped is not clear when you had the EGC ground terminal correctly connected (meaning connected).
What I have just said has been mentioned in previous posts.
With a non-GFCI receptacle and the ground pin correctly connected to an EGC to the main panel I would not expect you to get a shock. Thus, the GFCI probably had nothing to do with whether or not you got a shock.
.
Last edited:
LarryFine
Master Electrician Electric Contractor Richmond VA
- Location
- Henrico County, VA
- Occupation
- Electrical Contractor
First sentence, incorrect. However, a plug-in tester does need the EGC.jinglis said:
Second sentence, correct.
ELA
Senior Member
- Occupation
- Electrical Test Engineer
jinglis said;
"A GFCI does need a ground in order for the test button on the unit to work."
What, I thought this tread had died a rightful death. I guess some people just like to stir things up again - and without any value added.
You see this more and more now with no-clean soldering processes.
They are probably built to be ROHS compliant as well.
Did you look real close and see any whiskers growing :grin:
"A GFCI does need a ground in order for the test button on the unit to work."
What, I thought this tread had died a rightful death. I guess some people just like to stir things up again - and without any value added.
gar said:
You see this more and more now with no-clean soldering processes.
They are probably built to be ROHS compliant as well.
Did you look real close and see any whiskers growing :grin:
gar
Senior Member
- Location
- Ann Arbor, Michigan
- Occupation
- EE
080731-0926 EST
ELA:
When incorrect information is presented it should be corrected.
My long term experience with no-clean has not been satisfactory, but newer fluxes might be better. On a brand new device I do not expect whiskers, but in time under moist conditions I think there may be whiskers from RHOS.
It is true that the flux was in high voltage and high current areas and the low voltage sensitive circuit areas were clean.
The GFCI is a safety product and thus the expectation is design and construction should be directed toward reliability. Superficially the Leviton unit was nicely designed. But I am concerned about what the test button really does.
I have looked at the mechanics more and I think the TEST button only trips the mechanical latch. It appears the 15,000 ohm resistor test may be done at the time the RESET button is pushed. This would mean the TEST button is only a mechanical latch test, and that after you trip the latch it is necessary to RESET and at this time a current unbalance is detected. I am not sure on this yet.
I will have to get a new unit and not take it apart as completely and still try to hold the pieces together to try to see how it operates. There may be patent reasons for what is being done.
.
ELA:
When incorrect information is presented it should be corrected.
My long term experience with no-clean has not been satisfactory, but newer fluxes might be better. On a brand new device I do not expect whiskers, but in time under moist conditions I think there may be whiskers from RHOS.
It is true that the flux was in high voltage and high current areas and the low voltage sensitive circuit areas were clean.
The GFCI is a safety product and thus the expectation is design and construction should be directed toward reliability. Superficially the Leviton unit was nicely designed. But I am concerned about what the test button really does.
I have looked at the mechanics more and I think the TEST button only trips the mechanical latch. It appears the 15,000 ohm resistor test may be done at the time the RESET button is pushed. This would mean the TEST button is only a mechanical latch test, and that after you trip the latch it is necessary to RESET and at this time a current unbalance is detected. I am not sure on this yet.
I will have to get a new unit and not take it apart as completely and still try to hold the pieces together to try to see how it operates. There may be patent reasons for what is being done.
.
quogueelectric
Senior Member
- Location
- new york
It was a hilti te 12 I believe. Plastic 2w cord in a pouring rain condition. Gun was soaked wooden ladder was soaked I was soaked head to toe. No metalic case.gar said:
gar
Senior Member
- Location
- Ann Arbor, Michigan
- Occupation
- EE
080731-2111 EST
quogueelectric:
That would seem to imply my previous theory was incorrect.
Now the problem is to figure out an explanation of how an EGC connection to the ground terminal on the GFCI could change the conditions that caused your shock.
This connection would ground the box and anything conductively connected to the box.
If the line your drill contacted had something to do with the circuit associated with the GFCI, then this might be a starting point.
You may not remember all the exact conditions associated with the problem so it may be difficult to figure out why the results were as you described.
.
quogueelectric:
That would seem to imply my previous theory was incorrect.
Now the problem is to figure out an explanation of how an EGC connection to the ground terminal on the GFCI could change the conditions that caused your shock.
This connection would ground the box and anything conductively connected to the box.
If the line your drill contacted had something to do with the circuit associated with the GFCI, then this might be a starting point.
You may not remember all the exact conditions associated with the problem so it may be difficult to figure out why the results were as you described.
.
gar
Senior Member
- Location
- Ann Arbor, Michigan
- Occupation
- EE
080801-2033 EST
I purchased a new Leviton 7899-W GFCI and machined away a portion of the back molded plastic to expose the circuitry where the switch that connects the previously mentioned 15 k resistor is located.
As I had previously concluded the TEST button does nothing except trip the mechanically latched contacts. So this only tests the tripping of the latch. This does not close the contact to apply the 15 k ohm test signal.
But they have a very ingenious electro-mechanical design that does in fact test the electronics thru the trip solenoid. The actual testing is done when the contacts are reset. Unless the electronics and the solenoid work you can not reset or close the latchable contacts.
You can not mechanically latch the contacts by pushing the RESET button unless there is adequate power to the input terminals, and the electronics and solenoid operate correctly. Just mechanically pushing the RESET button can not close or latch the contacts. While the RESET button is pressed in the contacts in series with the 15 k resistor close and the circuitry must detect the unbalanced current of about 120/15,000 = 8 MA and operate the trip solenoid. The momentary operation of the trip solenoid allows the mechanical components to latch and close the power contacts.
Thus, electronic and trip solenoid testing is done on the leading edge of pushing in the RESET button. This requires input power. The contacts do not close unless this test is successful.
The reason for the double set of contacts is that the output terminals are connected by one pair and nothing else connects to the output terminals. Thus, if the GFCI is wired backwards there will be no power applied to the electronics and thus the contacts can not be closed and nothing happens.
.
I purchased a new Leviton 7899-W GFCI and machined away a portion of the back molded plastic to expose the circuitry where the switch that connects the previously mentioned 15 k resistor is located.
As I had previously concluded the TEST button does nothing except trip the mechanically latched contacts. So this only tests the tripping of the latch. This does not close the contact to apply the 15 k ohm test signal.
But they have a very ingenious electro-mechanical design that does in fact test the electronics thru the trip solenoid. The actual testing is done when the contacts are reset. Unless the electronics and the solenoid work you can not reset or close the latchable contacts.
You can not mechanically latch the contacts by pushing the RESET button unless there is adequate power to the input terminals, and the electronics and solenoid operate correctly. Just mechanically pushing the RESET button can not close or latch the contacts. While the RESET button is pressed in the contacts in series with the 15 k resistor close and the circuitry must detect the unbalanced current of about 120/15,000 = 8 MA and operate the trip solenoid. The momentary operation of the trip solenoid allows the mechanical components to latch and close the power contacts.
Thus, electronic and trip solenoid testing is done on the leading edge of pushing in the RESET button. This requires input power. The contacts do not close unless this test is successful.
The reason for the double set of contacts is that the output terminals are connected by one pair and nothing else connects to the output terminals. Thus, if the GFCI is wired backwards there will be no power applied to the electronics and thus the contacts can not be closed and nothing happens.
.
- Location
- Illinois
- Occupation
- retired electrician
Gar,
That is interesting information and it tends to support the idea that these GFCIs do not actually trip when the electronics fail, but only prevent you from resetting the device after a test. The marketing information has suggested that these are self testing and will automatically trip when the electronics have failed. Based on a very close reading of the technical details that I have been able to find and your testing, I believe that these newer "lockout" GFCI devices are correctly named, that is they "lockout" or prevent a reset when the electronics can no longer detect the current unbalance, but they do not self trip on the failure of the electronics.
That is interesting information and it tends to support the idea that these GFCIs do not actually trip when the electronics fail, but only prevent you from resetting the device after a test. The marketing information has suggested that these are self testing and will automatically trip when the electronics have failed. Based on a very close reading of the technical details that I have been able to find and your testing, I believe that these newer "lockout" GFCI devices are correctly named, that is they "lockout" or prevent a reset when the electronics can no longer detect the current unbalance, but they do not self trip on the failure of the electronics.
crossman
Senior Member
- Location
- Southeast Texas
Thank you GAR for the work and your findings. Very interesting indeed.
cschmid
Senior Member
- Location
- Northern cold country
I also thank you for the extra effort..nice.. :smile:
JohnJ0906
Senior Member
- Location
- Baltimore, MD
crossman said:
Thank you!
Maybe we can have the "Mike Holt Forum testing division" :grin:
gar
Senior Member
- Location
- Ann Arbor, Michigan
- Occupation
- EE
080814-1104 EST
More information on the Leviton 7899-W GFCI. I am only using this particular device in tests because it available, not too expensive, and may be representative of other units. It would be useful if someone else would analyze something like a Hubbell.
Another thread titled Overheating of GFCI Breaker had some comments that the breaker being hot might be due to the electronics. This seemed unlikely. Thus, a couple experiments on the 7899 since I do not have a GFCI breaker and do not know what brand was in the panel.
First, what is the power dissipated in the electronics. With the GFCI set by applying 120 V to the input and pushing the RESET button the electronics are energized. With no other load I measured the input current with a Fluke 27 and 87 and the results were different but generally in the ballpark of each other. The higher reading was about 8 MA and this means about 1 W at most. One watt internal will not produce much temperature rise on the outside surface of the GFCI.
Second test. Measurement of the power dissipation from 18.5 A flowing thru the GFCI. This was done with the GFCI preset, a 5" #12 wire shorting from the hot to neutral output screw terminals, and applying an adjustable input voltage to hot and neutral input to obtain 18.5 A. The voltage was measured between the screw heads of the two input terminals. Results I = 18.5 A, V = 0.129, P = 18.5*0.129 = 2.38 W, and R internal = 0.129/18.5 = 0.007 ohms. Obviously this produces more power dissipation than the electronics, and there is no power dissipated in the electronics at this very low input voltage.
I am working on photos to show the construction of the 7899.
.
More information on the Leviton 7899-W GFCI. I am only using this particular device in tests because it available, not too expensive, and may be representative of other units. It would be useful if someone else would analyze something like a Hubbell.
Another thread titled Overheating of GFCI Breaker had some comments that the breaker being hot might be due to the electronics. This seemed unlikely. Thus, a couple experiments on the 7899 since I do not have a GFCI breaker and do not know what brand was in the panel.
First, what is the power dissipated in the electronics. With the GFCI set by applying 120 V to the input and pushing the RESET button the electronics are energized. With no other load I measured the input current with a Fluke 27 and 87 and the results were different but generally in the ballpark of each other. The higher reading was about 8 MA and this means about 1 W at most. One watt internal will not produce much temperature rise on the outside surface of the GFCI.
Second test. Measurement of the power dissipation from 18.5 A flowing thru the GFCI. This was done with the GFCI preset, a 5" #12 wire shorting from the hot to neutral output screw terminals, and applying an adjustable input voltage to hot and neutral input to obtain 18.5 A. The voltage was measured between the screw heads of the two input terminals. Results I = 18.5 A, V = 0.129, P = 18.5*0.129 = 2.38 W, and R internal = 0.129/18.5 = 0.007 ohms. Obviously this produces more power dissipation than the electronics, and there is no power dissipated in the electronics at this very low input voltage.
I am working on photos to show the construction of the 7899.
.
Last edited:
gar
Senior Member
- Location
- Ann Arbor, Michigan
- Occupation
- EE
080815-1328 EST
In another thread
http://forums.mikeholt.com/showthread.php?t=102851&page=2
M. D.
provided this reference at post #65 which has a very good discussion
http://ecmweb.com/mag/electric_leaving_mark/
Looks like Leviton has a unique competitive advantage unless others have licensed or been able to get around Leviton patents.
However, of the two Leviton's I analyzed I think there is the possibility of a humidity problem.
.
In another thread
http://forums.mikeholt.com/showthread.php?t=102851&page=2
M. D.
provided this reference at post #65 which has a very good discussion
http://ecmweb.com/mag/electric_leaving_mark/
Looks like Leviton has a unique competitive advantage unless others have licensed or been able to get around Leviton patents.
However, of the two Leviton's I analyzed I think there is the possibility of a humidity problem.
.
peter d
Senior Member
- Location
- New England
gar said:
I'd be curious to see how the P&S and Cooper devices compare to the Leviton design. Perhaps you can cut some open for your next project?
As an aside, P&S has filed lawsuits against some Chinese importers for patent infringement of their GFCI's, presumably for copying their lockout device.
gar
Senior Member
- Location
- Ann Arbor, Michigan
- Occupation
- EE
080815-2047 EST
A more accurate power measurement of the electronics on the Leviton 7899-W, but not with a wattmeter. Still only by current and voltage.
The Fluke 87 read 8.9 MA and the 27 read 8.73 MA. Thus, the maximum possible power is 1.09 and 1.07 W respectively. This was at 123 V.
Suppose we have 50 of these devices in a home, then 50 W of residual load and at my electric rate this is 50*24*365*0.13/1000 = $57/year.
.
A more accurate power measurement of the electronics on the Leviton 7899-W, but not with a wattmeter. Still only by current and voltage.
The Fluke 87 read 8.9 MA and the 27 read 8.73 MA. Thus, the maximum possible power is 1.09 and 1.07 W respectively. This was at 123 V.
Suppose we have 50 of these devices in a home, then 50 W of residual load and at my electric rate this is 50*24*365*0.13/1000 = $57/year.
.
gar
Senior Member
- Location
- Ann Arbor, Michigan
- Occupation
- EE
- Status