On 12/3/2015 12:44 PM, dpb wrote:
On 12/03/2015 1:21 PM, Don Y wrote:
....
Sorry, I don't mean to sound ****y but "try this" is not what
I'm looking for. I want to approach the problem logically
not willy-nilly. I'd hoped someone might have *definitive*
information of problems like this instead of a litany of
hit-or-miss attempts. I.e., an explanation that reconciles
ALL of the observations I've posted.
(sigh)
One quick one is the possibility of the extension cord having a fault
(particularly as noted, if it's one of the pretty-common type with the
indicator light in the plug). It's a common element in all the combinations
done to date and is simple as compared to many of the other things you've tried
already or proposing to eliminate.
But, it has to be a "leak" -- as the problem doesn't manifest when
plugged into a non-GFCI circuit.
And, that "leak" has to somehow be healing itself, then reappearing
(based on whether or not the lights have been on or off and for how
long!)
"Willy-nilly" is pretty much what you've done to date; the above is, as noted,
one common element in all. It may _not_ have any bearing but then again, if it
just turned out to be so, then all the other pontificating is of little value.
I've been "willy nilly" in an attempt to be polite to folks offering
suggestions here. It's not how *I* would have tackled the problem.
But, having asked, it would be rude to dismiss suggestions when
(in MOST cases -- change breaker, change load, plug into different
outlet, swap extension cord, etc.) I can just do the test to humor
the respondent.
But, the "try this" can go on indefinitely. And, you can always
rationalize a semi-legitimate justification for each such attempt!
(e.g., ALL of the lamps have been out in the elements; perhaps
they've *all* been affected in some way! So, it doesn't matter
which strings you try, the problem will persist -- cuz it's
in ALL of them!)
[I worked on a piece of test equipment many years ago. There was
a short between two of the power supplies -- on a "circuit board"
that was 2 ft wide, 6 feet long and contained *thousands* of
components! I.e., the short could be in any of those components.
Or, in the board, itself. Troubleshooting all those potential
connections -- considering that the short was an UNINTENDED
connection -- would be a daunting task. Better to replace the
board and start over... after several months of delay!
I hypothesized that a capacitor could be the "short" -- they were
all over the (brand new!) board. Of course, *which* capacitor
was the problem!
So, I simply picked one AT RANDOM, removed it from the circuit
and tested it.
Shorted!
My colleagues looked at me like I had just walked on water:
not only had I come up with the "problem" but also managed to
find the defective capacitor, on my first attempt!
Boss, quickly turned that optimism into an accusation: "There's
no way you could have known that THAT capacitor was the failed
device! You *put* that there!!"
frown Sort of unfair -- given that I'd just saved his *ss!
I came back later to be greeted by news that *all* of the capacitors
were shorted! (WTF??) Turned out, the board manufacturer had
installed capacitors with the wrong voltage rating and, as soon as
we had applied power, they all went "pffft!"
So, there was no "luck" involved in my finding THE shorted capacitor.
OTOH, there was *genius* in my REASONING that it could be *a*
shorted capacitor -- and not a wire that was misplaced (of the
tens of thousands on the board)!]
(PS) And, no, on the GFCI circuit breaker I'm not specifically aware -- don't
have any and plan to live out my life that way. 
If it's not feasible, then
ok, so be it.
My problem *has* to be related to the GFCI characteristics of the "circuit".
And, not the GFCI detection itself (faulty breaker) but, rather, some
aspect of what it is testing that the current configuration happens
to "tickle" -- and, only in those cases where it *does* tickle (also
accounting for those cases where it *doesn't*!)