GFCI's

[Don Y]

There are no configurations that "don't fail" -- except the configuration where there is no load present (and, thus, no need for power!).

My approach will be the opposite: remove things until it starts working.

Recall, everything worked last year. The extension cords and lights have been stored "indoors" for all that time. OTOH, the outlets have been exposed to the elements for the ~9 months (including a Summer and a Monsoon) since then. So, any "changes" are most likely manifest in those items!

Wiring inside the block wall is likely not at risk as it is safely hidden away (unless something likes chewing on vinyl!)

First, verify SWMBO's assertion that the toaster oven resulted in a similar behavior. If that's true, it *tends* to rule out the lights and the extension cord as possible problem areas (no guarantee, there, as they could have other problems, as well). Then, reattach the lights as the "nominal load" -- cuz I have lots of experience with them failing (I don't want to end up with a situation where things SOMETIMES work and that confuses any deductions made later).

[The idea of leaving an item used for food prep outside just doesn't appeal to me! :> ]

Explore the upper limits on what the breaker will HOLD. If it trips (when it would otherwise have held, based on prior observations), I have another data point regarding the circuit's performance. Is any loss of capacity a likely effect of aging? Or, is it dramatic enough to suggest a fault, somewhere (keeping in mind that I'm using the replacement GFCI, presently). Can a regular (20A) branch circuit carry the same load without incident? (those breakers are OLDER, yet!)

Then, open all the Jboxes and have a peek inside. Any signs of "wildlife"? Moisture?

Beginning at the box closest (electrically) to the panel, look at hot-neutral, hot-ground, neutral-ground voltages under that fixed load (with breaker in the "holding" state). Any IR drops should scale linearly; regardless of how much wire there is between Jbox #X and Jbox #Y, there should be an identical amount of neutral, hot and earth conductors! I can do this from the exposed side of each receptacle -- no real effort required.

Then, start removing receptacles (in the hope that something may have set up shop INSIDE one). At the same time, examine the contacts on each: anything loose? Any signs of oxidation/corrosion?

Any changes in voltage readings when probing conductors instead of outlet contacts?

Ultimately, I'll have to remove the outlet into which the extension is plugged.

If this turns up nothing, start moving the load upstream. And, after verifying operation at each such point, isolate the downstream portion of the branch circuit (open wire nuts).

Eventually, I'll have a wire in the wall that connects the GFCI to the *first* -- and ONLY -- receptacle.

[Don Y]

How do you define "paying"? If I spend an hour doing something that doesn't yield any results, I've lost an hour of my time. No amount of money can buy that hour back, again! (this was the rationale that I used to stop letting other people -- bosses, clients -- waste my time "for a paycheck")

I chuckle when I hear folks claim that their time is "worth" $X/hour. "Yeah, *if* someone is standing nearby with $X in their hands, willing to give it to you for the next hour of your life!" :>

Working for myself, it is very easy to put a number like that on my time; how many MORE hours do I want to bill, this month? I.e., there are folks who *will* trade their $X for that hour -- so, I know the hour is worth AT LEAST that much.

Whether it is worth *more* I want to think it may be worth is always debatable. But, if you're not starving, you can elect to horde your hours and "spend" them on what YOU find enjoyable. I choose that approach at this point in my life (no one ever died wishing they'd worked a few more hours!)

[gfretwell]

So if you plug something into the outlet right next to the house it fails?

[Don Y]

(All of the outlets are "on" the house; presumably, you meant "closest to the electric panel"?)

I've not tested each of the 5 outlets. They are spread over the length of the house -- as well as around the side. I.e., I'd have to *lengthen* the extension cord to reach them all.

I don't plan on continuing the "willy nilly" approach of jumping to "try this", "now try that". Instead, I will approach the problem in a more structured manner -- so each result adds to the data set in a more organized fashion (instead of taking pot shots in different areas).

But, that will wait until at least the weekend. I have many other things that need to get done in the short term and the forecast suggests I've got at least a week before temperatures begin to approach the lows that are troublesome...

[Don Y]

Of course it's do-able. The question is one of where that sits on the list of things that *must* get done and *should* get done.

The decorative lintels over the garage and all windows will need to be replaced. Cracks in the stucco need to be patched. Both of these before I can paint the house.

We'll need to give some serious consideration to replacing the roof in the next year or two.

SWMBO wants to replace some more windows in the front of the house.

I want to install a french drain along the back side of the house to harvest rainwater -- and a large cistern to store it.

We've yet to select the water softener.

And, I still have a lot of automation to design and install.

Replacing an electric panel that isn't causing any problems sure seems like ASKING for work!

We've seen one panel "fail" in the neighborhood -- due to corrosion on the feed. Every other panel-related activity has been "elective" or mandated by tariff changes from the utility. All the homes were built within a year or two of each other so all "suffer" the same sort of wear. If there was a pending problem, it;s likely we'd be seeing some signs of it *somewhere* in the neighborhood.

[I walk the entire neighborhood -- 3.8 mile loop -- daily. So, interact with folks in enough different areas to have a pretty good feel for what sorts of problems people are having. I can tell you who's had plumbing problems, who's had their furnaces replaced, who's had problem with their PV arrays, who's water meters have failed, who's had pipes freeze, etc. While some folks are interested in the salacious trivia, I'm more concerned with things that are likely to cause me problems down the road; things I can anticipate before they force me to act! :> ]

[makolber]

A partial short between the ground and the neutral in the extension cord or elsewhere could cause the OPs symptoms.

Mark

[Don Y]

What causes it to *heal*? And also reappear?

"Cold flow" from a neutral conductor to case can likewise cause load-related GFCI trips. But it doesn't "recover" and also "recur" -- not in a matter of tens of seconds.

[gfretwell]

You can't get more "structured" that just plugging a comparable load into the receptacle that fails (without involving extension cords tree lights etc)

[Don Y]

What if the nature of the load has an impact on the results? E.g., say I plug a large power supply for that has an insane turn-on transient in but idles at just a few watts (switching losses). Smaller load but bigger turn on transient.

Or, drag out a longer spool of wire to use as the extension cord?

Or, try the toaster oven on one and lights on another?

Don't add variables to the analysis. Just come up with a consistent test strategy and apply it consistently.

E.g., the second set of 3 strings that I dragged out to mimic an "equivalent load" (as the first set that are presently IN the tree) is only conceptually an identical load. Making observations with one set in one case and another set in another case isn't applesapples. Even though it might (and quite probably is!) appear to be so.

[gfretwell]

For the purposes if this discussion, a hair dryer would do just fine. If it trips, you know you have a problem in the wall or receptacle string. Then try it at the end of the extension cord. If still no trip, there is a problem with your lights.

The idea that an incandescent light will trip a GFCI in normal operation (no faults) is ludicrous. That tiny surge while the filament is heating up is well inside the trip curve of any breaker I have ever seen. It would certainly be less than the heat up time of a hair dryer element.

[makolber]

If it is a partial short from neutral to ground, the GFI will trip only when there is a large current flow, for example during the turn on surge of cold bulbs.

If even a small percentage of the current flows through the ground, it will trip the GFI.

[Don Y]

You've not been paying attention to the numerous "experiments" I've already conducted.

E.g., different extension cord --> trips. Different light strings (same cord) --> trips. Different GFCI breaker (same branch circuit) same extension cord, same lights --> trips. "Cold" lights --> trips. "Warm" lights --> no trip.

Extension + cold lights plugged into non-GFCI branch circuit --> *no* trip. Ditto for different GFCI branch circuit, no trip.

I.e., nothing wrong with extension cord *or* lights.

Toaster oven (reported by SWMBO) plugged into same outlet WITHOUT extension cord --> trips.

Problem is *clearly* with the wiring in the wall -- the only thing common to all fault cases and NOT present in any of the non-fault cases! And, only manifests when a load is present. Furthermore, only when the load is significant ("cold" lamp strings)

The whole point of the discussion is that there *is* a fault! Note that a much larger load (3X) had been present on the same branch circuit last Winter with *no* problems for the entire season!

Forecast was for a cold night, tonight. So, rushed to get *something* working -- regardless of an "explanation" -- in the hour I had before sunset.

As it was obvious that the problem had to be in the branch circuit, I removed all receptacles, examined the routing of the conductors within each Jbox, inspected all wire nuts, screw terminals, grounds, etc. and, then, reassembled everything. Washed the "wet use" covers just to make things look pretty (continuous sun exposure turns the exterior paint to a powdery substance).

No "obvious" problems: no "bugs" falling out of the receptacles, no corrosion on terminals, no exposed wires under wire nut skirts, no nicks in insulation, no moisture in boxes, no mounting screws pressing on conductors, no cables pinched in clamps, plenty of room in each box (35 cu in), etc.

Circuit has been holding without any problem -- even in the (unexpected!) rain that's been falling (and the "false alarm" for the cold weather!). Added another 2 strands of lights (with a second extension cord) just to push my luck...

Took the (inexpensive -- $1) precaution of fitting "child proof" plugs to all unused outlets to ensure nothing *can* crawl into any of the outlets in the future.

No way to ensure I can recreate the problem -- as I have no idea what it *actually* was -- so I'll leave well enough alone and wait for the next hiccup. Maybe buy some 20A receptacles to replace these when I next have to go poking around in the Jboxes...

Case closed.

[Muggles]

Perhaps something wasn't plugged all the way in and you fixed it when you checked everything?

[gfretwell]

More likely that there was a neutral fault that was cleared when he pulled everything apart and put it back. Lots of mystery GFCI problems are fixed by an inspection that didn't actually find a problem. Who knows what it was?

[clare]

"It's all in the way you hold your mouth"

[gfretwell]

A very unrecognized skill in this business ;-)