700' GFCI Protected Circuit

Status
Not open for further replies.

infinity

Moderator
Staff member
Location
New Jersey
Occupation
Journeyman Electrician
Someone asked me today if they installed a 20 amp GFCI CB and had branch circuit conductors of 700' in length would there be a problem with the GFCI nuisance tripping? The conductors are #8 THHN and let's assume they're properly sized for voltage drop.
 
Someone asked me today if they installed a 20 amp GFCI CB and had branch circuit conductors of 700' in length would there be a problem with the GFCI nuisance tripping? The conductors are #8 THHN and let's assume they're properly sized for voltage drop.
My experience says yes, especially underground.
 
My answer was that it was probable but I couldn't give any technical substantiation.
 
I agree with Larry in a run that long there could be nuisance tripping. Intall a GFIC receptacle at the end of the run.
 
... 20 amp GFCI CB ... 700' ...

Infinity,

Sounds like you could get leakage that way,
or that (underground) leakage could occur in the future.
Try installing pristine, perfect THHN conductors,
with no scrapes or scratches.
Try a load of slickem,
in straight conduit PVC (RNC) with end-to-bell couplings going in the pull direction.

Better yet,
Try placing the GFCI at the Load end of the circuit.
:)
 
I'm with 'yes' too. And I actually have a reason :)

700 foot of insulated conductor will have a significant capacitance to ground.

In an AC circuit, that capacitance will appear as the equivalent of DC resistance, so lets use DC as it makes the terms more familiar.

Ohms Law tells us that current is voltage divided by resistance. Lets assume the two conductors have about the same resistance (capacitance) to ground, but one (the neutral) is near ground potential, and the other (the hot) at line potential. Thus for the same resistance, the two different voltages on the two condutors will mean two different currents will flow.

700 ft might be long enough that the current on the hot never drops below 5ma so you can never close the GFCI, let alone it nuisance trip...
 
091130-2157 EST

I doubt that 700 ft will produce sufficient capacitance to trip the GFCI. Assume 25 pfd per ft to ground. The result is 0.018 mfd. You guys practice your theory and calculate the Xc. I think my 25 pfd is an adequate ballpark estimate using #12 Romex hot to EGC for reference.

.
 
All conductor insulations have a leakage value.
Most GFCIs will open upon 5 miliamps of current that is not flowing within the proper circuit path. Leakage current is not the proper path.


Typically, 100ft of cable is really 200ft of conductors that contribute to the leakage current.
If the installed run is 700ft, the length of conductors that contributes to the leakage current is 1400ft.

I do not know the figures, but my experience tells me this length of conductor will undoubtily create enough leakage current to operate the GFCI function.
 
Square D lists a specific circuit length (I think it is 200') for their QOB_GFI breakers to prevent nuisance tripping. I think it is a footnote right in the Square D Digest.

Steve
 
091130-2157 EST

I doubt that 700 ft will produce sufficient capacitance to trip the GFCI. Assume 25 pfd per ft to ground. The result is 0.018 mfd. You guys practice your theory and calculate the Xc. I think my 25 pfd is an adequate ballpark estimate using #12 Romex hot to EGC for reference.

.

I get Xc = 147.365813 ∠ -90o

(2*3.14150*60) = 376.9908*.000018 = 0.0067858344

1/0.0067858344
 
091130-2157 EST

I doubt that 700 ft will produce sufficient capacitance to trip the GFCI. Assume 25 pfd per ft to ground. The result is 0.018 mfd. You guys practice your theory and calculate the Xc. I think my 25 pfd is an adequate ballpark estimate using #12 Romex hot to EGC for reference.

.


Is your "25 pfd per ft" for one conductor to ground? If so, we need to use a total length of 1400 feet. (Both the line and neutral will leak to ground.)

So I get 35 nF.

With 1/(2*pi * f* C) I get about 76 Kohms. With 120 volts, that's 1.5 mA.

It could be higher if your capacitance to ground doesn't include the capacitance between a wire and the metal conduit it is in.

Steve
 
Is your "25 pfd per ft" for one conductor to ground? If so, we need to use a total length of 1400 feet. (Both the line and neutral will leak to ground.)
So I get 35 nF.
With 1/(2*pi * f* C) I get about 76 Kohms. With 120 volts, that's 1.5 mA.
It could be higher if your capacitance to ground doesn't include the capacitance between a wire and the metal conduit it is in.

Steve

If both conductors leak, the leakage currents will cancel each other and the GFCI will not detect any difference.
 
If both conductors leak, the leakage currents will cancel each other and the GFCI will not detect any difference.

Thinking about it, I don't think that it's the correct answer.

The Neutral wire is at almost ground potential, so with the same leakage capacitance to ground as the Hot wire the leakage current will be much less.
 
091201-0931 EST

Pierre:

There is no reason to double the cable length. The branch circuit consists of one grounded wire and one hot wire. Look at dbuckley's discussion above. There is virtually no leakage, resistive or capacitive, from the grounded conductor to ground (conduit or earth) because that voltage difference is essentially 0.

From an electrical circuit analysis point of view the grounded conductor is not grounded until it gets to the input side of the GFCI device. But it is still reasonable to call it the grounded conductor.

Thus, we are concerned with the capacitive and resistive leakage from one 700 ft wire to ground. Further we do not care about any leakage, capacitive or resistive, from hot to the grounded conductor. Only the capacitive and resistive leakage from the one hot conductor to ground is of concern.


steve66:

If I put my test sample of Romex in a 1/2" conduit the capacitance of 25 pfd from hot to EGC when in free space becomes 31 pfd/ft with EGC and conduit connected together. The other conductor is floating. We are not concerned with the capacitance of the conductor I left floating for the reason above. However, if it is also connected to the conduit there is a negligible change, less than 1 pfd.

I roughly agree with your calculation on Xc other than for the use of 1400 ft. Using 700 ft your value would be doubled. This is way above 120/.005 = 24000 ohms.

The Sq-D note is probably a useful value to protect them from a lot of other factors. One of which is the inrush current to this line capacitance when the initial charge is zero and the breaker is turned on at the peak of the voltage waveform. However, I really believe the threshold detector input low pass filter should prevent this problem.


K2500:

You have a decimal point problem. Multiply your Xc by 1000.


iwire:

Yours is probably the most important reason. Provide as much headroom as is reasonable. The GFCI specifications do not mean that tripping may not be somewhere below the 5 MA spec.

.
 
The OP stated # 8 wire so the capacitance will be larger due to the larger wire. Even though the calculation of Xc (guessing at the capacitance value for #8 wire) may not quite give a 5 ma leakage I am betting on nuisance issues if not an out right trip.
 
091201-1048 EST

ELA:

True and I do not have a piece of #8 at home at the moment to test.

Here is a calculation for a coaxial capacitor.

C = 7.36 * dielectric constant / log (b/a) pfd/ft

From p134 of Reference Data for Radio Engineers, ITT, 1960.

Dielectric constant of vinyl about 3 to 4. #8 is about 0.12" dia. Thus, a #8 in 1/2 should be less than
C = 7.36 * 4 / log (0.5/0.128) = 50 pfd/ft
This calculation assumes the wire is centered in the conduit and the air space is fully filled with vinyl.

.
 
091201-0931 EST


steve66:

If I put my test sample of Romex in a 1/2" conduit the capacitance of 25 pfd from hot to EGC when in free space becomes 31 pfd/ft with EGC and conduit connected together. The other conductor is floating. We are not concerned with the capacitance of the conductor I left floating for the reason above. However, if it is also connected to the conduit there is a negligible change, less than 1 pfd.

I roughly agree with your calculation on Xc other than for the use of 1400 ft. Using 700 ft your value would be doubled. This is way above 120/.005 = 24000 ohms.

The Sq-D note is probably a useful value to protect them from a lot of other factors. One of which is the inrush current to this line capacitance when the initial charge is zero and the breaker is turned on at the peak of the voltage waveform. However, I really believe the threshold detector input low pass filter should prevent this problem.

.

After seeing the comment about the neutral voltage being 0, I agree with you that 700' is the correct distance to use.

But as you mentioned, there are other factors that might come into play. One might be the inductive kick that happens when we open the switch for the circuit.

Steve
 
091201-1654 EST

Some additional experimental results.

#4 stranded with Nylon outer sheath. Wire dia about 0.222" and OD of insulation about 0.323". About 0.050" insulation thickness.

In 1/2 thinwall 42 pfd/ft. In 3/4 copper water tubing essentially the same.
In the V of a 1 x 1 Al angle 26 pfd/ft.


#6 stranded with Nylon outer sheath. Wire dia about 0.177" and OD of insulation about 0.252". About 0.037" insulation thickness.

In 1/2 thinwall 34 pfd/ft.

.
 
Status
Not open for further replies.
Top