Calling all electricians...Electronic Ballasts + GFCI = snap!

[outsiders71]

I've been growing my seedlings for the last few years with some overdriven GE T8 electronic ballasts. Originally the ballasts were meant to light four bulbs, I overdrive 2. Anyways after I moved last year to my new home, I've setup my grow shelf and lights in the basement. I'm stuck with using a GFCI outlet which last year gave me problems with my timer. Every time the timer turned the lights on the GFCI would shut off the outlet. I didn't know much about GFCI and figured maybe timers weren't compatible with them. So last year I manually turned the lights on and off.

Fast forward to this year, I've increased the amount of shoplights from 4 to 8. I got four shelves so 2 per shelf. The GFCI goes off all the time at random times. It could go off within 20 minutes, a few hours, or not at all. Lately its been going off within an hour or two.

So I busted out my multimeter and started checking the continuity of all the bulbs. They all checked out fine. I've checked and out of OCD rewired some of the fixtures. All the grounds have continuity. I even have a 14 gauge wire attached to my metal rack that I have hooked up to the 3 prong that goes into the wall to ground the rack itself. I did some searches online and it appears this is not an isolated incident. I was "daisy chaining" two power strips together and thought maybe that had something to do with it. So I bought a nice power strip with a built in circuit breaker and 8 grounded outlets. Same results.

I'm running out of ideas and am becoming frustrated. If I plug these shoplights into the GFCI receptacle individually they will not throw it off. It's when they are all plugged in the problem occurs. So based on that and the continuity checks I'd have to say my wiring is correct and safe. What are my options then? There's only two outlets in my basement (gj to whoever built this house) and the other is a non-gfci outlet but is far away. I assume running a grounded extension cord would not be a good idea?

The other option is to install a new circuit breaker, and outlet, since the fuse box is right next to the grow shelf. However I don't have any experience doing this and would prefer to not take any risks if not necessary.

All I can do right now is run some of them all night and swap them out for the others during the day. I'm out of ideas...

[dice]

Something leaks current. It could be a ballast shorting to
the fixture, a light socket shorting to the fixture, wiring
inside shorting to the fixture, or even wiring shorting
between neutral and ground inside a 3-prong plug that
is plugged into the gfci. The gfci itself could be shorted,
too (like between the ground and neutral connectors inside
the female socket, or an electrical equivalent of that on a
circuit board). Usually it is some device plugged into it rather
than the gfci itself, but they can fail.

(faqs)
http://www.faqs.org/faqs/electrical-...ection-24.html
http://www.thecircuitdetective.com/gfis.htm

[outsiders71]

Well that narrows it down

[outsiders71]

Another relating question:

There's 16 T8 bulbs in my setup. I didn't replace all of the bulbs, just the ones that had dark ends.

Could a single defective bulb throw the GFCI? Besides physical appearance and continuity testing, how else can you test these lamps for defection?

Thanks.

[dice]

Well, you could try the one at a time method and see which
light is shorting to ground, if that is actually the nature of
the problem. (Plug the other ones into your two prong plug
with an extension cord temporarily while doing the gfci tests.
The circuit breaker should still protect you from problems
with those lights even without a grounded outlet. Keep a
fire extinguisher handy. Don't walk around on wet concrete
while testing. Wear rubber soled shoes. Etc.)

It is interesting that they may need to run for awhile before it
happens, meaning that the short is produced after something
heats up sufficiently, or that vibration in the home lets it
happen.

[beeman]

Would it not be simpler to swap out the GFCI for a regular outlet? Unless of course you have lots of water around?
I assume the GFCI is installed correctly. I remember when I wired my green house it was most particular about the initial proving test.

[dice]

The bulbs don't matter. They use X amps of current,
and if the wiring is all kosher and the GFCI is not
defective, then that amount of current goes out
through the hot wire plugged into the GFCI and
comes back through the neutral wire. It does not
matter whether they are using more or less current
than they are specified to use, as long as the current
leaving the GFCI matches the current returning to the
GFCI.

In this case it apparently does not match, at least not
all of the time. There are two paths to ground at the breaker
box, through a neutral wire and through a ground wire.
A ground wire only carries current if there is a short
somewhere and the device with the short is grounded
someway, usually with a ground wire and 3-prong plug,
grounded metal conduit, whatever. (Current can also leak
to ground through a concrete floor, a cast iron or copper
pipe that connects to ground outside, etc.)

[outsiders71]

Originally Posted by dice

In this case it apparently does not match, at least not
all of the time. There are two paths to ground at the breaker
box, through a neutral wire and through a ground wire.
A ground wire only carries current if there is a short
somewhere and the device with the short is grounded
someway, usually with a ground wire and 3-prong plug,
grounded metal conduit, whatever. (Current can also leak
to ground through a concrete floor, a cast iron or copper
pipe that connects to ground outside, etc.)

Well this metal grow rack is in the basement sitting on a concrete floor. Does this have anything to do with it?

[dcarch]

Electronic ballasts can generate RFI, especially if you over-drive them and RFI can cause GFCI to trip.

If you have ferrite chokes, try sliping one on the power cord to eliminate RFI to the GFCI outlet.

dcarch

[dice]

Well this metal grow rack is in the basement sitting on a concrete floor. Does this have anything to do with it?

It could. The metal light fixtures touch the metal grow rack,
the metal grow rack touches the floor, .... Current has to be
leaking to one of the metal light fixtures inside it, though,
or you still would not trip the GFCI that way. I think you
mentioned grounding the metal plant stand itself, so that
would provide a direct route to ground for leaked current
without current travelling through the floor. (I just mentioned
the floor bit for completeness in my description.)

[outsiders71]

Interesting. I don't know if this matters or not but due to the amount of heat made by the electronic ballast from overdriving it, I modified the fixtures so that the ballast is mounted on the outside (top side) of the fixture. I don't see why this would be an issue vs being enclosed as the ballasts are still grounded into the fixture and into the ground wire.

So the hypothesis is that current is leaking through the fixture, into the rack, and then...?

Forgive me as I'm 50% ignorant to this stuff so I'm trying to clarify...

A current leaking out from one of the fixtures has to be the cause of the GFCI tripping correct? Since I have 14 gauge wire attached to the shelf at one end and wrapped around the ground of the 3 prong plugged into the GFCI at the other end, how is the GFCI tripping? Shouldn't all the leaked current be going to ground?

Uggh..

Next question, is there a way for me to check each fixture while under load for current leakage via a multimeter?

Sorry if I'm rambling I'm exhausted. I appreciate all your help.

[dice]

You are imagining that the path the circuit takes in normal
operation is out the hot, through the light, and back through
the ground wire. That would be wrong. The current returns to
the GFCI through the neutral wire (the non-hot side of a
2-prong plug, and that same connector in a 3-prong plug).
The third prong underneath is simply there for safety. If current
actually does return to ground through that wire, then
something is wrong.

How to measure current:
http://www.electronics-radio.com/art...re-current.php

This site has a description of testing a gfci:
http://ecmweb.com/mag/electric_think_gfci/

edit:
One thing that you could do is use a 110 ac light fixture,
or a lamp, something like that, for a current tester
on the ground circuit. Cut your ground wire between
the plant stand and the round prong on the 3 prong plug.
Connect the side of that wire coming from the plant stand
to one side of the test lamp, and the other side of that
wire to the other lead on the test lamp. Then turn off all
of the light switches on the flourescent fixtures and plug
in the cord to the plant stand.

If the test lamp glows at all, there is a short upstream
of the light switch on the first flourescent light fixture.
If not, turn on the flourescent lights, one by one, until
you get a glow in the test lamp. The last one switched
on before the test lamp glows has the short. (Test
all of them, even if you find one that has a short, unless
that is the last light fixture.)

This is perhaps easier than wiring up a rig to measure actual
current in the hot, the neutral, and the ground when the cord
is plugged in and the flourescent lights are on.

If the test lamp never glows, then the short is elsewhere,
in your 3-prong plug, in the gfci outlet itself, etc.

(Remember the warm up time, too. You might need to put
a comfortable chair and reading light down there so that you
can wait it out until it trips. Be sure that you can see the test
lamp from the chair with a book or seed catalog in your hand.)

As for dcarch's suggestion, sometimes you can get a cheap
choke to soak up rfi at a second-hand store that has lots
of computer keyboards. One finds them built into old, pre-USB
keyboard cables or the keyboards themselves. People have used
those to clean up the signal on monitor cables, etc.

Here are some pictures so that you can see what he is talking
about:
http://www.radiodan.com/Henry/parts/RF_chokes.htm

[outsiders71]

I've got 6 of the 8 shoplights cohabitating nicely with the GFCI now. I think I have it down to the one ballast that might be leaking current.

Instead of the test lamp, couldn't I just disconnect the ground wire that is screwed to the fixture and hookup the probes, one to the ground wire and the 2nd to the fixture?

Thanks for all your help.

[Worth1]

I have read a lot of misleading stuff here so far.

One, a GFCI does not reference ground to neutral,(neutral and ground share the same buss on many breaker panels, really,) it references ground to hot.

Two, never ever disconnect the ground to stop the GFCI from tripping.
Three electrictiy follows the path of least resistance.

Four if it is your house why not remove the GFCI put in a regular outlet and ground the hell out of your system.
You will still be protected from shock, refer to statement three.
If you have water in your basement don't even think of doing this.

Five for your information refrigerators and freezers do not have to be on a GFCI because they can trip out and ruin your food.

Six a GFCI is a weak breaker of sorts that if it draws too much currant it will trip out.
You probably have a ballast going out which will draw too much currant and cause the trip.
Seven inrush currant is the currant that is drawn when something is first started up.
This goes away after the device is running and the currant goes down this could be the problem.
To solve this IF the outlet has a GFCI on it you can again install a regular NON GFCI outlet and then plug in 2 GFCI cords this will cut the currant drawn on the GFCI in half.
use two lamps instead of one or whatever.

Yes I have worked as a Union electrician and have done electrical for many years I do low voltage instrumentation now.



120 VAC has killed more people than any other voltage ever, this is because it is just enough the make you clamp up on the currant or whatever it is going to and not enough to knock you off.

I have been hit so hard it almost stopped my heart and it did put me on the ground, a buddy knocked me off.

PS get NFPA 70 for reference or just ask me, I have the book and it cost a lot of money

I hope this cleared things up a little.

Worth