Peter,
> The guys who manufactured and did the applications on diodes, SCR's,
> triacs and IGBTs at my employer's plant in Lincoln (before it was
> sold it to someone!) told me that it's OK doing without equalising
> resistors if the diodes are from the same diffusion batch, but warned
> that diodes from different batches shouldn't be mixed without
> equalisation
Yes, this is of course a point to keep present. It would never occur to
me to make a series string of different diodes, and use them without
equalization! But for that matter, I wouldn't use a series string of
different diodes at all, not even with equalization!
> because the avalanche voltage can be high enough in some
> of them for the dissipation to become a problem when they avalanche.
I'm using 1N5408 diodes, and the maximum average forward current they
have to conduct in this amp (in a voltage doubler configuration) is
close to one half ampere. So there is a lot of dissipation capability to
spare! As long as a diode won't dissipate more than about 2 watts (a
very conservative rating) due to avalanche, it should be safe. And even
if one of the diodes takes the full voltage (very unlikely!) of 2.5kV,
that allows almost 0.8mA of avalanche current before that dissipation is
exceeded, and of course the other 4 diodes in series with it will NOT
pass that 0.8mA reverse current with a negligible voltage drop! They are
rated for a worst case reverse current of 0.5mA at their maximum voltage.
So I think in this application the diodes are quite safe.
> Now this was from people making diodes all the way from microwave
> detector diodes through small signal, small rectifier and up 5000 Amp
> devices for electric trains....
I can imagine that when stacking 5kA diodes to make a 22kV rectifier,
these issues become quite a bit more critical than in a "low power" ham
radio amp! ;-)
> This suggests that if you have to
> replace a modern diode, you should replace the string...
Yes, that's advisable in any case. And they are so cheap that it doesn't
hurt to do so.
Steve,
> Interesting comments, Manfred - but what types of resistor are
> failing on you?
Those in the oscilloscope are all carbon film, ranging from 1/4 watt up
to 2 watts. Resistance values range from 560k to 10M. The voltages per
resistor go from as low as 95V to about 500V, with the higher voltages
on the larger resistors. So I can't see an obvious design mistake. The
resistors are not used at higher voltages than their rated ones. But
still they fail a lot! I have been replacing the smaller ones by higher
power, physically larger ones, whenever possible, and that has reduced
the failure rate, but even some of the larger replacement resistors have
failed!
In the amp, I'm not sure what technology was inside those resistors.
They were cylindrical with end caps, no visible wire winding on them,
covered by brown lacquer, about five or six cm long and about 8mm thick,
if my memory isn't kidding me. I threw them away, so I can't check now!
They MIGHT have been carbon film too! Otherwise they could have been
metal oxide. They must have been rated at 5 watts or so, and resistance
value was, I think, 50k. They were working at about 320V each.
Manfred.
----------------------------
Visit my hobby website!
http://ludens.cl
----------------------------
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|