torturing caps

Impedance exposure

I think he should be let off with a conditional discharge, after all, JL's a busy man with a lot on his plate.

That could be taken to be negative.

John

...

um

if

ical

ey

There are several tests they do, 2 of which were performed on 100% of parts. Other tests were performed on a percentage.

not really relevant, just a red herring.

This isnt the first time I've seen a ref given that contradicts what the reference giver is claiming. We all make mistakes, accept and move on.

NT

I guess Di elected to let him off the charge. Such capacity for compassion.

NT

..

the

ry

ack

at

is

ly

use

ne

is

ou

ou

he

or.

t's

not

s a

al

Wiki has masses of expertise, but when it comes to electronics its frequently wrong, not something I'd rely on in this subject. the wiki diode breakdown graph in that article should help clarify the situation somewhat.

At teh end of teh day, people did explain.

NT

It's a general article mostly concentrating on breakdown of dielectrics rather than semiconductors. It is a brief overview and doesn't contain info on subjects of recent interest such as partial discharge testing of dielectrics.

Indeed: "As long as the current is limited, exceeding the breakdown voltage of a diode does no harm to the diode."

Repetitively doing it eventually causes enough ion migration to soften the breakdown, ultimately destroying the junction. Heat accelerates the process. ...Jim Thompson

OK as long as it wasn't in public, what one does in their own 'castle' is their own buisness.

George H.

=20

Not ceramics. They normally will not fail below 3X rated V at 125C. At room temp over 10X rated V or more is normal. About 30 years ago i participated in a breakdown study on MLC chip caps. Lots of 50V and 100 = V parts were making it to 1500 V to 1800 V and 2.4 kV to 2.8 kV respectively, using a Biddle tester and a HiPot tester.

Change in C with voltage and temperature is a whole different matter. I each and especially in combined environmental conditions it could take 30 minutes to settle out. (0.1 % measurements and better.)

That depends a lot on what you have for measuring equipment and means (which may be built into the measuring equipment) to segregate the bias voltage and the measurement signals. =20 Earlier in life i was decent test engineer. What is there to work with currently (measuring equipment and fixtures). We can discuss what can be done with that and perhaps a modest effort and cost (preferably below = $100 materials) in developing fixturing with added ability to segregate bias voltage and test stimulus and measurement signals.

This is only an invitation to proceed with the investigation.

the

and=20

=20

bad=20

=20

Nearly. They test a few diodes per wafer and type the wafer full based = on that result (worst case). Most diodes (non-Schottky) will take 2X to 3X more Vrr.

the

and=20

say=20

bad=20

could=20

=20

Nope. Not more than a few percent. Straightforward statistics. That is still scores to hundreds per wafer (hundreds to thousands of diodes) = which is enough.

part,

the

say=20

a bad=20

could=20

statistics...=20

or whatever.

Not even for Military do they do that any more, nor have they for = decades. See Mil-Hdbk-217.

but=20

=20

statistical=20

would=20

issues=20

Wonderful, but the production equipment is spitting out 1 million diodes an hour. You can test about 4 percent, which is way, way more than = needed for statistical process control.

but=20

Nope, just one machine (maybe two if here are two lines in that space), not even running all of the time.

=20

statistical=20

They haven't tested 100 percent of anything produced in volume since the Shuttles were built. They didn't test most satellite parts 100% even 20 years ago. I did work for Hughes Space and Comm back then for a while.