"petrus bitbyter" wrote in message
.nl...
"Arfa Daily" schreef in bericht
...
Didn't you walk right into the problem? It seems that by a yet unknown
cause the circuit sometimes runs into some state that makes it oscillate
or in some other way fries the resistors. This kind of intermitted
faults are the most difficult to solve as most of the times there seems
to be no problem. Even your observations with the scope may mislead you
as connecting the probe may change the circuit enough to change its
behavior.
petrus bitbyter
Yes, there may indeed be some kind of intermittent problem. I will put
them all back up again on Monday, and see how they perform this time. As
to my 'scope muddying the waters, it's generally pretty well behaved in
that respect. It's a high quality 100 MHz job, which is always used with
a x 10 low capacitance probe. If that is having much of an effect on the
circuit, then it must for sure be a pretty poor design. To be honest, I
still think that the problem lies with me somehow. Something that I did
differently between the first and second times that I tried them, but I'm
not sure what ...
Arfa
Off course you may have made a mistake. Humans make mistakes, even I
sometimes do 
But if so, what on earth fried those resistors the first time?
petrus bitbyter
Every example that I've seen so far, has the resistors badly discoloured and
the print and substrate scorched. On some, one of the resistors has been
open, so the network has not been doing the job it was put there for, at
all. They are 3 watt resistors, and even when the supply *appears* to be
running correctly, they get hot enough for you to say that they are probably
dissipating a good 3 watts, and maybe a bit more, so I would guess that you
would have to say that from that angle, it's a badly designed bit of
circuitry. I am fairly convinced that the purpose of the network is to
attenuate the big spike that occurs on the leading edge of the switching
waveform. This would tie in with why they have used about the biggest film
resistors they could get, rather than using a higher power wirewound type,
which would have a fair bit of inductance. I would also surmise that they
have used two x 150 ohm resistors rather than a single 330 ohm, to try to
spread the dissipation a bit.
When you replace the resistors, they still run hot, with no visible signs on
the 'scope of any 'RF' on the waveform, so you'd have to say that it *is*
working correctly. What led to this thread in the first place was that when
I was initially evaluating these supplies for the company that wants them
repairing, after I replaced the resistors, they ran very hot when the supply
was loaded, but seemed to just run 'acceptably' hot when it was idling.
Likewise, when loaded, the switching FET got very hot, but was almost cold
at idle.
However, when I next tried them - same conditions for i/p voltage and load,
as far as I was aware - they now seemed to be working much better in that
the resistors were just acceptably hot for all conditions, loaded or not,
and the FET remained cool also. So this has now led me to believe that it
must have been something I was doing differently - and wrongly - when it was
running very hot. So, a mistake ? Yes, probably. As you say, we all make
them, and this has got to be one of the easiest trades for doing it in.
I've just heard from the company that they are sending a bunch more up this
week, so if there's some more of this type amongst them, I'll have some more
'untouched' ones to see what happens this time.
Arfa