Lee,
If you have a 1000 volt power supply, what should the diode stack
voltage rating be?
In principle, just a little more than those 1000V should be enough, as
long as it's a bridge rectifier. That is because with a bridge there can
never be higher voltage across any of the four diode groups, than the
voltage on the filter capacitor. So, basically the diode voltage
required is the peak voltage at the highest line voltage expected, times
the transformer ratio, plus some safety margin. And the output filter
cap will charge to that same level (without the safety margin, of
course!), under those conditions.
But since diodes are cheap, most people use more diodes.
I have heard it is good practice to make it 4
times the voltage of the operating voltage.
That can't hurt, and won't break the bank, but looks rather generous to
me. In my NCL-2000 I used 5 1N5408 diodes in each string. That one works
with about 2500V. So I have only 2 times the output voltage in the
diodes, and that has worked well. This amp uses a voltage doubler, which
poses twice the current load on the diodes, but the same voltage as a
bridge rectifier.
If using a two-diode full wave rectifier with a center-tapped
transformer (obsolete nowadays), I would use a higher diode string
voltage, because due to the imperfect coupling between windings,
transients on the diodes might exceed the filter cap voltage.
Also, would there be a difference when a choke input is
used VS capacitive input?
Yes. With a choke between the diodes and the capacitor, the clamping
effect of the cap is lost, and any transients on the power line will
appear in full across the diodes, as long as they are long enough not to
be absorbed by the transformer's loss at higher frequencies. In this
case, some sort of transient protection should be added - MOVs, Transil
diodes or Tranzorbs. And enough diodes should be used to safely survive
the highest voltage your protective devices might allow to show up, plus
a healthy safety margin.
Now if you use beefy enough avalanche-rated diodes, the diodes
themselves can assume this transient clamping function! A lot depends on
the size of any transients you have to expect. Whether it's just the
transients created by the fridge switching off, or that of a big
industrial motor powered from a long line. And if you are expecting
lightning hits in your supply line, better forget protection, and unplug
the amplifier (and everything else) before the thunderstorm starts!
I remember a funny problem at the job: The circuit breaker feeding a
rack full of control equipment was tripping, very regularly, once every
half hour or so. The current actually consumed by the equipment was just
one tenth of the breaker's rating. The guy in charge was scratching his
head. I was asked to help him fix it. I also scratched my head. The
breaker would trip out of the blue, after roughly half an hour. And
again. And again.
After lots of hours observing, measuring, guessing, finally we
discovered the problem: There was a rectifier/filter connected across
the line, to provide power to a system that was off most of the time.
There was no bleeder resistor, so there was zero current consumption.
That 450V rated filter cap was found to start at about 320V, and over
the course of minutes the voltage would grow, and grow, and grow,
reaching about 600V after half an hour. At that point it arced over
internally, almost killed us with the sudden scare (it sounded like a
gunshot), and at the same time the breaker tripped!
This capacitor must have been slowly charging way beyond nominal
voltage, just from little high voltage spikes periodically appearing on
the line. A 220 kiloohm bleeder resistor across that capacitor fixed the
problem. The capacitor had apparently not been damaged by the arcing and
current pulses, but just to be safe, we replaced it and the diodes
anyway. Neither my colleague nor I liked to be phoned out of bed at 3 AM
to fix a sudden problem caused by an abused and degraded component.
Lesson: Don't forget your bleeders, if there is no other current drain!
Manfred
========================
Visit my hobby homepage!
http://ludens.cl
========================
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|