Non Linear & GFCI

Status
Not open for further replies.
ptonsparky

ptonsparky

Tom
Location
NE (9.06 miles @5.9 Degrees from Winged Horses)
Occupation
EC - retired
We needed some lighting in a small area we are wiring. We had the permanent fixtures, 2-lamp T8 surface wraps so we put a appliance pigtail on one, and promptly tripped the gfci. Different fixture in another GFCI, same thing. Lift the EG and of course no trip. I called the supplier and told him to come get all 28. He called back a while later and said non linear loads and gfci do not mix well. OK. Seems odd that we have been using 6 - 6 lamp t5 HOs on a GFCI recept for close to 2 weeks now.

Brand name fixtures. We are waiting on reply from the factory. Any one experience similar problems?
 
I did have an issue with Ge afci- no gfci in them- had to change to a different generation gfci to make the T-8 fixtures work. I am not sure whether that afci had gfci in it.
 
121010-0806 EDT

Get a 1000 ohms resistor, anything larger or equal to a 1/4 W.

Put the resistor in series with the EGC from one fixture. Measure the voltage across the resistor with a Fluke 27 or 87 or equivalent with the fixture powered. One milliampere will produce 1 volt. Possbly you want to see not more than 0.1 V as a possible criteria.

Why do this instead of using the current range on the meter. Quarter watt resistors are cheap compared to the fuse in the meter. If there is excessive current the resistor just burns up.

I somewhat doubt that the distorted current produced by the fixture is of a sufficiently high frequency to exceed the bandwidth of the current transformer in the GFCI. More likely there are one or more capacitors from hot and neutral in the fixture to the fixture enclosure. This current test should identify any such capacitors. There may be stray capacitance, but it needs to be about 0.0025 ufd or 2500 pfd for 0.1 MA at 120 V.

My cheap hardware store 4 ft T8 reads 0.19 V across 1000 ohms. An old 8 ft Slimline with noise filter reads 37 V or 37 MA. I used a 1/2 W carbon comp resistor and at 37 V it would burn out fairly quickly, but not in the time for a quick reading.

.
 
121010-1045 EDT

An Advance R-2E60-S-TP magnetic ballast, not in a fixture, has a shunt capacitance of 385 pfd from the white wire to case. 330 pfd from black to case. These measurements are with a GR-1650-A RLC bridge at 1000 Hz.

.
 
One of the sites I care for had the same problem, so the electrician changed to a different brand ballast and the GFCI stopped tripping. He and I concurred there was probably excess leakage to ground but didn't do any experiments other than lift the EGC from the ballast. Of course, this was one bathroom fixture so it was easier to deal with than a lot of 28...uggh.

Thread derail coming...

gar said:
...These measurements are with a GR-1650-A RLC bridge at 1000 Hz....
Click to expand...

You mean one of these?

eqp-gr-1650-a_lg.jpg

Cool! Used to lug a GR-916 bridge out to AM radio towers back in the day. Glad I don't have to anymore...they were beasts.

End of thread derail...:)
 
Last edited:
121010-1517 EDT

grich:

Yes, one of those. It is 50 years old and still works. The mechanical automatic balance mechanism failed long ago, and was removed.

.
 
gar said:
121010-0806 EDT

Get a 1000 ohms resistor, anything larger or equal to a 1/4 W.

Put the resistor in series with the EGC from one fixture. Measure the voltage across the resistor with a Fluke 27 or 87 or equivalent with the fixture powered. One milliampere will produce 1 volt. Possbly you want to see not more than 0.1 V as a possible criteria.

Why do this instead of using the current range on the meter. Quarter watt resistors are cheap compared to the fuse in the meter. If there is excessive current the resistor just burns up.

I somewhat doubt that the distorted current produced by the fixture is of a sufficiently high frequency to exceed the bandwidth of the current transformer in the GFCI. More likely there are one or more capacitors from hot and neutral in the fixture to the fixture enclosure. This current test should identify any such capacitors. There may be stray capacitance, but it needs to be about 0.0025 ufd or 2500 pfd for 0.1 MA at 120 V.

My cheap hardware store 4 ft T8 reads 0.19 V across 1000 ohms. An old 8 ft Slimline with noise filter reads 37 V or 37 MA. I used a 1/2 W carbon comp resistor and at 37 V it would burn out fairly quickly, but not in the time for a quick reading.

.
Click to expand...

Have 1K, will meter.
 
Wasn't there a time when the ballast manufacturers were building in a ground sensing circuit so that fixtures wouldn't work when ungrounded? If that were the case here, you could easily have more than 5 mA shunted to ground.
 
Status
Not open for further replies.
Top