On Thu, 08 Sep 2016 08:17:12 -0700, Jeff Liebermann
wrote:

On Thu, 08 Sep 2016 17:15:16 +0800, pedro wrote:

It's fair bet that any I source nowadays will have the
ROHS curse on them.

Ok, I see you've done your homework. Yes, ROHS is epidemic
everywhere.

I did some Dremel surgery today on one of the dead coils. Despite
having differing copper resistivity figures at hand (no consequence)
and the wire diameter testing my micrometer, from known ohms of a good
coil and some spreadsheeted maths I guesstimate it is 33 or 34AWG.

That's rather thin and could easily fuse. According to the wire
table:
https://en.wikipedia.org/wiki/American_wire_gauge
#34 will fuse at 5A. What do you measure for coil DC resistance?

Maybe Chris' suggestion is worth investigating. The surgery wasn't
enough to find the dodgy termination, but thicker winding wire *may*
get a better joint.

I agree. Heavier wire may solve or delay the problem. Methinks it
would be interesting to know what temperature the solder joint is
experiencing. A thermistor or thermocouple glued to as close to the
solder connection as possible might provide some interesting numbers.

Indeed. It would be handy to have a thermoprobe with logging on it.
Note to self: see where I can borrow one.

If the position of the solenoid above the oven is the problem, that
will show it. Perhaps adding a metal heat shield between the coil and
oven?

The dual valve mechanism is bolted to the oven. The coils (see ebay
item# 322017672259 for the ones we "consume" - but at a much better
price than he's asking ...) sit over the metal housing which encloses
the valve plunger (armature) and has a metal spacer each end for
location and ?thermal separation?. So it would be necessary to
elevate the entire incoming gas pipe/valve syhstem to relieve
conducted heat to any significant degree. And if it IS the cycling -
rather than the actual temperature reached - which causes the failures
then reducing the latter may achieve nothing.

For fusing, perhaps a small value resistor or "surgistor" in series
with the coil might reduce the peak current enough to let the thin
wire survive. With an inductor, there should NOT be an inrush peak,
but I'm thinking some kind of glitch, spike, or peak might be arriving
via the 240VAC line.

Dunno. Could do (but it'd be 18 months before I'd know if it made a
difference).