It might be related to spark plug wire condition,and spark plug
condition;if your wires are breaking down,it would place a greater load on
the coil,and if the plugs were gummed up,slagged,or carboned,that would
affect the coil voltage.
IMO,igniters fail because of transients and migration of metals in the
Darlington and IC.

One could conceivably use an electron microscope and examine failed igniter
semiconductors and see why they failed,but it would be an expensive
undertaking,unless you're a college student with access to an e-
microscope,and doing a thesis or paper.

I don't think the heat is too great for the ignitor. There apparently
aren't
any electrolytic capacitors in it, and those are the big heat sensitive
worries.

The vulnerability of transistors to heat is related to junction
temperature
and to collector voltage. When the collector junction gets hot enough,
the
reverse voltage across it can make the dopants migrate and "blur" the
junction. That spot gets hotter and eventually melts, leading to the
device
being shorted.

But the heat required to melt the junction without the voltage stress
is
much higher than your engine compartment - the melting point of silicon
is
over 2500 degrees F. The silicon chips are typically soldered to the
heat
sink internally, and I have personally soldered a UHF power transistor
to a
large copper heatsink with an acetylene torch.

Mike
--------------------------------------
Yes, but you didn't put power to it until after it cooled down, and I
bet you moved a lot of air across that heatsink to keep it cool. In
service the
transisotr is switching at least 5A. I have the exact value in my
notes,
and yes it is a fast switch, the transistor is driven completly and
rapidly into full conduction so the Vec is about .5V for ~2mS every few
10s of aMsec.
Not a haeavy duty cydle by any means, but it runs, with "perfect"
heatsinking
to the block at ~195F. Those I measured actualy ran about 210~220F.

Somewhere I have the leakage versus temp from motorola and at +200F
we are getting too darn hot. I saw a Ducati pointless ignition system
that
used a flip flop to switch between 2 transitor to cut the Pd in half.
But
that system was expossed directly  to the airstream.  Of the people
I know who have had igniter failurs, Honda or Toyota, it has always
been in >90F weather, most frequently in stop and go traffic on the way
home.
Except for one that failed in ~-25F weather on startup. I can't prove
it, but I suspect if we could reduce the temp 20 or even 10 degrees F,
the failure rate would drop "way down".  Another failure mode that is
only slightly heat related is the piezo effect on the junction when it
switches.
I had a early JBL switch mode power amp that ate switching transistors
at fairly regular intervals. I had built, and still have, a ultrasonic
down
converter and you could plainly hear the transitors screaming at 40KHz.
I supsect that heat, plus current stress, plus the piezo effect could
explain
most igniter failures. One could always attempt to place it remotly, in
front
of the radiator to breath cool air, but I suspect the lead inductanc
would
kill you and the RFI would be "interesting".

When I needed to align my R2000 SW I needed an "ignition like"
noise source "with harmonics extending up too beyond 30MHz"
I pulled my 1991 Civic under my "long" wire antenna, and adjusted
the noise blanker. Worked much better then the puilse generator
at the shop where I work.

Terry

the "blurring" is diffusion - the operating temp doesn't even get close
to melting - but diffusion is powerful stuff and it definitely destroys
semiconductors.

the two factors at play are temperature & time.  if the device was
physically big enough and you solderd quick enough, there's no reason
you couldn't do it - afterall, what temperature does a silcon foundry
work at?  but you got to be /quick/ if you go to high temps.  or best
not at all if you want good service life.