Cooper GFCI

[gar]

080911-0942 EST

ELA:

That is a neat discovery. It is also another means to manually trip the latch. Basically it is the end of the solenoid plunger and covered over with a white label.

How did you discover this spot?

.

 

[ELA]

Gar,
When I saw the four screws on the rear of the unit I could not resist opening one up. It was so easy to do I could not refuse

I did find the electromechanical latch mechanism to be interesting.

As far as quality goes I found the same remants of solder flux on the board.
I also identified a cold solder joint! It was on the trace that leads from the incoming neutral terminal to the input MOV. So I guess that would go unnoticed in actual operation.

 

[gar]

080911-1400 EST

ELA:

If that MOV failed to do its intended task, then it is likely that diodes or the SCR might fail and cause the coil burn out problem previously mentioned. Even if those components were not damaged a failure of the MOV to clamp transients might allow transients to cause false trips.

For those unfamiliar with the characteristic curve of an MOV it provides a very soft clamping action in comparison with a Zener diode type device.

We need many people to comment on field failures of GFCIs. This should include information on brand, model number, possible age, and what is found from measurements on the device and disassembly. That kind of information could help others to determine what devices to use. Field failure information is one of the best ways to evaluate a product.

.

 

[ELA]

Gar,
Did you notice the second MOV directly across the diode bridge input?
Looks like even more protection for the diodes and SCR (possibly from its own coil transients.)

 

[wptski]

gar:

What's the diagnostic line of attack on a intermittent tripping GFCI? I know somebody that has one that trips maybe once a week, sometimes goes longer. You mentioned that transients may cause false trips if there is a damaged component. I'm curious.

 

[peter d]

We need many people to comment on field failures of GFCIs. This should include information on brand, model number, possible age, and what is found from measurements on the device and disassembly. That kind of information could help others to determine what devices to use. Field failure information is one of the best ways to evaluate a product.

I have found that receptacle GFCI's typically die when installed in outdoor applications, regardless of brand. However, Leviton is the dominant brand in my neck of the woods so I run into those the most. The old style just become regular receptacles, while the newer "lockout" one won't reset.

Apparently this was a big enough problem for the NEC to address this issue with a "weather resistant" requirement for receptacles installed in wet locations in the 2008 edition.

 

[gar]

080911-1954 EST

ELA:

The small MOV is across the AC input points of the bridge rectifier, and also in parallel with the anode and cathode of the SCR. The resistance and inductance of the trip coil is in series with the bridge. And a capacitor is across the small MOV. This should provide fairly good transient filtering.

This is a rather neat design. First, is the MOV directly across the line, then a dropping impedance to the input of the bridge, another MOV, and a shunt capacitor. This should do a good job of protecting all the electronic components, and holding the dv/dt down. Obviously assuming the components are good and no excessively large input energy pulse occurs.


wptski:

A false trigger with long random times between occurances is tough to troubleshoot. Also is it really a false trip.

Does this tripping occur if there are no loads on the GFCI? If so, then are there any large transient generators in the home? 8 ft Slimlines produce large transients. Motors with mechanical relay contacts do also. Switched inductive loads. Is there anything noticeable occuring at the time of trip?

If there is a load on the GFCI, then possibly start looking for problems in the load.

The brand and year of manufacture and others experience with the same device might help provide background information.

Are there other GFCIs in the house that do not randomly trip?

Why does a machine repairman have a better knowledge of troubleshooting electrical problems and knowledge of instruments than many electricians?


peter d:

With the construction that I have seen I would expect wet environments are likely to cause problems in various areas of the device. This might be the reason Cooper has a conformal coating on part of the circuit.

.

 

[wptski]

080911-1954 EST

wptski:

A false trigger with long random times between occurances is tough to troubleshoot. Also is it really a false trip.

Does this tripping occur if there are no loads on the GFCI? If so, then are there any large transient generators in the home? 8 ft Slimlines produce large transients. Motors with mechanical relay contacts do also. Switched inductive loads. Is there anything noticeable occuring at the time of trip?

If there is a load on the GFCI, then possibly start looking for problems in the load.

The brand and year of manufacture and others experience with the same device might help provide background information.

Are there other GFCIs in the house that do not randomly trip?

Why does a machine repairman have a better knowledge of troubleshooting electrical problems and knowledge of instruments than many electricians?
.

gar:

It's my sister's home, rebuilt in 1991 after a electical house fire. The GFCI's have no name on the faces. It's in a attached garage, not sure about the loaction in the circuit or loads but there are several FL's in there. There are several power tools there. Sometime after the rebuild, a sub-panel was added to carry extra loads in the garage, so some branch circuits are still powered from the original service panel. As far as I know, it's the only one that trips. I wasn't sure if my brother-in-law repalced it or not but on my last visit, it was still tripping.

I've been following your GFCI autopsy with interest. Even if I remember all my military electronics training, it's all obsolete! When you mentioned possible false tripping, my eyes really opened!!

To answer your last question. Like I mention, I have a background in electronics, most of which I forgot now. I maintained machines that nobody wanted to work on because they were a mechanical nightmare. I love to troubleshoot. Some electricians resented my electrical knowledge becasue I challenged them at times but others liked it!

I'm waiting on a Klixon overload for my dehumidifier right now.

 

[ELA]

Older Style GFCIs were notorious for nuisance tripping. One very common occurance was when a bathroom fan turned off. Transients would take them out all the time. This is why the manufacturers started addressing this in the newer devices.
Now a days you can read in some manufacturers specifications notes about enhanced noise immunity.

If an older GFCI is nuisance tripping the first thing I would do is replace it with a more recently manufactured unit.

 

[wptski]

Older Style GFCIs were notorious for nuisance tripping. One very common occurance was when a bathroom fan turned off. Transients would take them out all the time. This is why the manufacturers started addressing this in the newer devices.
Now a days you can read in some manufacturers specifications notes about enhanced noise immunity.

If an older GFCI is nuisance tripping the first thing I would do is replace it with a more recently manufactured unit.

ELA:

I'd rather replace a defective device after finding the cause than to just guess at it. Sure, they are easy to replace and you have a 50/50 chance that your guess is correct too.

I'll just have to find out what's on its circuit and other circuits in their garage. I have a couple of different options of measuring PQ.

 

[gar]

080912-1151 EST

wptski:

I just ran my 8' Slimline transient test on the Leviton GFCI. The GFCI was connected to my bench outlet with about 8' of #16 and the GFCI EGC terminal was left unconnected. The bench outlet is not a real low impedance.

Two tests were run.

In the first test I randomly cycled the fluorescent on and off with the light plug at the bench socket. Generally this will randomly produce 4000 V or so transients. No trip of the GFCI.

Second test. Same as first except the GFCI was the receptacle into which the light was plugged. No trip of the GFCI.

.

 

[wptski]

080912-1151 EST

wptski:

I just ran my 8' Slimline transient test on the Leviton GFCI. The GFCI was connected to my bench outlet with about 8' of #16 and the GFCI EGC terminal was left unconnected. The bench outlet is not a real low impedance.

Two tests were run.

In the first test I randomly cycled the fluorescent on and off with the light plug at the bench socket. Generally this will randomly produce 4000 V or so transients. No trip of the GFCI.

Second test. Same as first except the GFCI was the receptacle into which the light was plugged. No trip of the GFCI.

.

gar:

How did you measure or know that your 8' Slimline produces a 4000V transient?

 

[gar]

080912-1440 EST

Bill:

I did it with my scope. A little difficult, crude, and not accurate. But any great accuracy I did not need, just an idea.

A better way would be with a peak hold circuit. Needed is a compensated high voltage probe. I have one good for over 30000 V that I made in the 50s for ignition system development work. I have not tested it in years. Basically to make a high frequency probe the divider network needs a compensating capacitor to provide a high frequency capacitive divider because your low voltage end of the probe looks like a shunt resistor and shunt capacitor. A typical scope input is 1 meg shunted by 15 to 57 pfd (these were measured value on an old Tektronix scope).

See the following site for a schematic of a probe and scope input. Note there is considerable probe cable capacitance that is also added to the scope input capacitance.
http://coefs2.njit.edu/ECE291/ECE291VII.htm
The following for different compensation waveforms:
http://www.future-mag.com/0705/070518.asp

If I had a correctly working MOV across the line it would greatly reduce the peak voltage seen at the receptacle from the switching of the fluorescent, and this is the case with the Leviton (current model).

A crude peak hold circuit can be built with a diode and capacitor.

.

 

[wptski]

gar:

Thought maybe you had some sort-of PQA? I can measure transients with my Fluke 43B. I have a couple small fluorescents that I'll test tomorrow.

 

[wptski]

gar:

The most my two 4' fluorescents could generate was .24 kV transients.

 

[gar]

080916-0925 EST

bill:

The peak transient voltage will greatly depend upon the source impedance of the line where the lamp is connected, and all the various loads that may be on that line, and the maximum energy stored in the inductor of the lamp.

My previous concern was with noise generated by relays in a control panel and the effect on electronic circuits in the same machine. Today with most machines using solid-state output relays there is little noise generated during the machine cycle.

When the Van Dyke axle operation started in 1975-76 all the control logic was with relays, and these were a noise problem. I kept Bob Farmer, the machine serviceman, of Trio at the plant for almost 24 hours one day debugging the pinion shim machine logic. We had haywire all over that panel. Tim Hay, the process engineer, went and found some sort of "director's chair" so he could sit and watch us. He was the Ford person to make sure we had what we needed. You well know the kind of pressure there is getting new machines working in the plant to meet some critical deadline for production. The Numalogic used on the 77-78 machines was a disaster for debugging.

Some programmable controllers make logic debugging easier than relay logic. What is not easier is when a sequencer mode is used to control a machine when combinatorial logic should have been used.

.