Weird: spa light trips GFI when water is heated

[generessler]

Replaced the bulb in our spill-over spa's old school halogen Pentair Aqualite about two months back. Worked great until recently. Then one night I heated the spa for the kids and the GFI breaker tripped. Reset a couple of times. Kept tripping. Next morning, all was fine.

This pattern repeats. I have a guess about what's going on, but wonder if anyone has seen this before.

I used a new gasket: highly rated third party. The fixture is only 3.5 years old. No sign of water ingress. Pool lights work well throughout. They're on their own relay, same breaker, so it's easy to be sure the spa is the problem.

Before I open up the fixture again, any ideas how this happens? I'd think if there's water in there, it shouldn't be able to "fix itself" when the spa cools down.

Many, many thanks for all help, past, present, and future.

 

[PoolGate]

Maybe try replacing the GFI with a new one. They do wear out.

 

[Mdragger88]

With every trip they get weaker and more sensitive.
You obviously do want to rule out any safety issues with the heater or light first.

 

[JamesW]

Is the light 120 or 12 volts?

LED or incandescent?

Light Model Number?

Turn off power, disconnect the power wires and check power wires to ground for continuity.

Note: Only do what you are 100% sure you can do safely.

 

[generessler]

Is the light 120 or 12 volts?

LED or incandescent?

Light Model Number?

Turn off power, disconnect the power wires and check power wires to ground for continuity.

Note: Only do what you are 100% sure you can do safely.

Thanks. Some of this is in the OP:

• Pentair Aqualite model 77362000
• Halogen: 120 vac, 250 watts (T-4 bulb).

Thanks. I studied EE and have done quite a bit of work on home wiring.

 

[JamesW]

Turn off power, disconnect the power wires and check power wires to ground for continuity.

 

[generessler]

Thanks as always for the idea. I was unable to work on this until today. But just did what @JamesW said.

Fwiw the problem is now more consistent. It trips randomly when cold, too.

From light hot (black) to ground (green) initially reads 140k ohms then over about 30 seconds increases to 450k. I guess this is the capacitance in the cord charging up? Neutral to ground is the same, which makes sense.

I read GCFIs are supposed to trip at ~5 milliamps. Even the low end 140k would pass less than 1 milli. Right?

I also noticed that the disconnected incoming line neutral to ground gives 11 ohms. That seems high, no? Maybe that's the problem? At outlets I typically see 2.5 ohms.

 

[JamesW]

You should read O.L from power to ground.

Any continuity is a ground fault.

 

[JamesW]

From light hot (black) to ground (green) initially reads 140k ohms

I read GCFIs are supposed to trip at ~5 milliamps. Even the low end 140k would pass less than 1 milli. Right?

120 volts at 140,000 ohm would be about 0.9 milliamps.

However, the voltmeter uses a low voltage, so it might not be accurate.

A megohmmeter would use a higher voltage and this would probably give a more accurate reading than a standard multimeter.

In any case, any continuity reading is a fail and points to a ground fault.

 

[JamesW]

Note that the neutral is tied to ground at the main panel, so the light neutral has to be isolated when testing to ground.

 

[JamesW]

I also noticed that the disconnected incoming line neutral to ground gives 11 ohms. That seems high, no? Maybe that's the problem? At outlets I typically see 2.5 ohms.

It depends on the wire size and length.

For example, a 100 foot #14 AWG wire should read at 0.59 ohms for the round trip from a 100 foot neutral to the main panel and then a #14 ground back to the test point.

You will have some resistance for connections, so maybe 1 ohm total would be expected.

I think that 11 ohms is too high, but it should not trip the GFCI because the problem is on the Line Side of the Power.

It probably makes a problem worse because the high resistance causes more current to go through the ground fault than if the neutral was good.

I would probably replace the Line wires, the GFCI Breaker and the light.

 

[generessler]

Okay thanks guys. Based on the "fail" assessment I opened the light. No obvious water, but there was a thin coating of white inside. (Salt or calcium?) And the lamp base was cracked, though the lamp was still working. So sure looks like there was tiny seepage. Just for grins I'll blow some hot air at it to make sure everything is dry and try the ohmmeter again.

Any tips for getting a perfect seal? I was clean and careful, but I guess not enough. The metal case has a double ridge where the gasket ring sits. The outer ring had some roughness. Probably corrosion. I worked that out with 600, then 1500 grit emery. It's like a mirror now. Maybe add silicone lube? RTV? I will order the OEM gasket. Maybe the third party one was no good, though it looked fine.

 

[PoolGate]

Gasket lube and don't over tighten.

 

[generessler]

It depends on the wire size and length.

Thanks so much for taking a look at this.

The 11 ohms is measured at the pad, which is maybe 40' from the main box where the neutral/ground tie ought to be. The wiring I can see is 12 gauge, but even if it's 14 the wire should be way less than one ohm. I will try back-tracking the neutral tomorrow. I have already had a couple of failures due to bad wire nut installations. Guessing it's that again.

The breaker might be weak, but I don't think it's terrible because it also powers the pool lights, which work fine. I'll replace it anyway.

After I cleaned and dried the light parts I get infinite ohms to ground on the neutral side and 5 megs on the hot. So still something off. Maybe that white coating is slightly conductive. I can't get way down in the bottom of the socket to clean there.

The prospect of replacing the light makes me sad. I already did that just 3.5 years ago when the bulb blew. We seldom use the lights, but when we need them they're important. Was trying to avoid this kind of try-fail cycle. Oh well. All the threads about lights not lasting very long now include mine.