Mfcttrf

This is, I am sure, an extremely naive question. I hope that's okay (and I hope it can be answered at a level I'll understand).

The background: My swimming pool has underwater light fixtures, on GFCI circuits. I recently replaced the bulb in one of these fixtures, reassembled the fixture incorrectly, and re-installed it. About 48 hours later the circuit breaker (the one inside the house, in the breaker box, not the GFCI) tripped. I removed the fixture, which was full of water, realized my mistake, reassembled it correctly, re-installed it, and it's now been fine for several weeks.

Question 1: In my naivete, I'd have thought that the water in the fixture would trip the GFCI immediately, so that the circuit breaker in the house would never have tripped. But the circuit breaker in the house did trip. So was my expectation wrong, or does this mean my GFCI is not working?

Question 2: After I re-installed the fixture and flipped the breaker back on, the lights worked, even though I never reset the GFCI. Is that evidence that the GFCI is not working?

Question 3: In pondering Questions 1 and 2, I was led to wonder how I would ever know if my GFCI is not working. Removing the underwater fixtures once a month to hit the test button is not terribly practical. Is there another way to test this?

An underwater GFCI doesn't matter

That is to say, it doesn't perform any useful function underwater. It does nothing to prevent the water from being electrified, which is its one job. Here's how a GFCI is laid out.

As you can see, if water can get to the "Line" side of the device, then it electrifies the water. And the GFCI cannot do a thing about it.

So this is a lost cause. I really don't care whether your friend thought he saw a GFCI down there; if it exists it is useless.

GFCI and overcurrent are different things

A breaker trips on overcurrent, when the total current flow exceeds the breaker limit by a wide enough margin or a long enough time. (breaker trip curves are pretty generous).

A GFCI actively looks for current differential on the two conductors. When it detects a small amount (8ma) of unequal current, that means current is seeking a third path, and it trips.

While they both relate to current, that is the end of the similarities.

That said, if the breaker is a combo GFCI+breaker device, it may seem like an overcurrent trip. You have to pay close attention to its indications to see whether you are dealing with a GFCI or overcurrent trip. If it is a GFCI breaker, it will have a TEST button.

Otherwise if the breaker tripped, it's because either a massive amount of current flowed, or the circuit was mildly overloaded for awhile. This could simply be from too many appliances plugged into the circuit.

GFCI protection is required for pool circuits

As said, the GFCI units underwater don't count. Every circuit within 6 feet of the water (length of a common appliance cord) needs GFCI protection. Fortunately any GFCI device can confer GFCI protection to devices down-circuit of it. At extremes, a GFCI+breaker protects the whole circuit.

So the right way to protect the circuit with the pool lights is find a point along that circuit before it nears the pool, and fit an appropriate GFCI device there. E.G.

• a receptacle not likely to be splashed with water,


• inline in GFCI-only (deadfront) devices,


• part of a GFCI+receptacle+switch combo device


• a GFCI breaker to replace the regular breaker.

Alternately, make the pool lights low-voltage DC

12 volt DC power is incapable of shocking swimmers. Old incandescents draw too much current for that to work with existing wiring, but LED is so efficient that it works fine at 12V on existing wiring. So why have special devices which require constant testing to protect swimmers from shock, when you can moot the entire point by switching to 12 volts DC?

I recommend fitting a 12VDC power supply at least 6' from the pool and running all the pool lighting on that, using LED. This also means you never need to change another bulb. It's even possible to do it using existing fixtures, since they make Edison base LED replacements that are 12 volts DC.

It sounds like your existing GFCI is not working, pool water (if clean is not really a good conductor) but should have tripped a GFCI prior to tripping a breaker (unless the breaker is a GFCI). The type of pool can make a difference also an in ground concrete/ gunnite or similar pool has a bonding grid that should have tripped the GFCI. A fiberglass in ground pool or a poly lined may not trip a GFCI.
The GFCI should not be embedded in the fixture you should be able to push the test button at the breaker panel (monthly) to test if it is working. If the GFCI is an outlet type feeding the light fixtures it should not be within 5’ of the pool but it should also be able to be tested without removing the fixture. The type of fixture can also make a difference if the shell of the fixture is plastic with a glass bezel it is insulating material so even though there was a leak and clean water is not a good conductor the contacts may have been in the water conducting from hot to neutral and an air space so the dripping water did not or could not create a leakage path to trip the GFCI , so with this there are several cases that lighting won’t trip a GFCI but may trip a breaker. I have always used low voltage lights but still provide the transformer with GFCI protection, you should be able to test the GFCI device and this should be done monthly.

If the GFCI is faulty how would I know if the test button is functioning correctly. The test button is not part of the fault sensing circuitry it is an external portion that creates an imbalance in the hot and neutral line and when this imbalance reaches 5ma it should cause the fault or open the circuit. if not the GFCI has failed and needs to be replaced. You should be able to test without removing the fixtures, if you need to remove the fixtures to test the GFCI the breaker needs to be updated to GFCI unless the circuit voltage for the lights is below the low voltage contact level of 15v ac or 30dc (I believe these are the correct values but I don’t have a code book handy).