Re: NST and Bridge rectifier to stop res rise

> If the cap is connected across the rectifier output there will be a
> problem.  The problem arises when the gap fires and the cap voltage
> tries to swing negative,  forward biasing the diode bridge.  This
> will drive the diodes into heavy conduction and spoil the primary
> ringdown. You could put a small RF choke in series with the
> rectifier output but I guess that is defeating the object.

And it will destroy the diodes with the high forward current.. Been there,
done that, destroyed $50 worth of microwave oven rectifiers.

If you do this, make sure you have a resistor or choke in series to limit
the forward current in the diodes.

> The diodes stacks would have to be suitably rated to hold off the
> peak HV voltage, and also to withstand the peak tank current I think.
> (I hope I'm not stating the obvious here, but the diode chains
> should have equalising resistors and equalising caps accross them.)

Modern HV diodes,when bought from the same mfr lot, are consistent enough
that equalizing resistors and caps probably aren't necessary, and may
actually aggravate the problem. Bear in mind that 6kV, 200 mA diodes cost
about $0.50 each (less in quantity) so you would probably use them as if
they were rated at 3 kV (use the string as if it had a PIV of 3*N kV, where
N is the number of diodes in the string).

Here is why:

There are two related problems that the equalizing resistors are designed
to fix:

1) unequal reverse resistance causing uneven voltage distribution.  If the
diodes come from the same lot and are at the same temperature, they will be
much more closely matched than any resistors you are likely to come across.
 Even if they vary a bit, if you underrate them by 50%, who cares, it is
still way below the rated PIV.

2) unequal reverse recovery time. Diodes don't turn off instantly when the
voltage reverses (in effect, they have a small sort of capacitance).
Typical HV diodes take a bit less than a microsecond to turn off. In
theory, when the reverse voltage is applied to the string, some of the
diodes will turn off while others are still conducting, causing the voltage
on those that turned off faster to rise.  The traditional fix was a
parallel capacitor to "slow" the voltage rise until the diode had time to
turn off.

Modern HV diodes are quite consistent in their reverse recovery
characteristics, and, what's more important, they have a "soft recovery"
characteristic, so that they don't turn off abruptly.  And, if you
massively overdesign (as you should), who cares if a couple of the diodes
in a string of 10 don't turn off as fast, the voltage increase on the rest
of the string is negligible. 

Finally, the typical use for rectifying a sine wave output of a transformer
is actually a pretty easy job for a diode string, in reverse recovery
terms. The reverse bias is slowly applied as the voltage climbs to the
peak, so there is plenty of time for the diodes to recover. (unless you are
building a 100 kHz HV switcher, in which case you should already know all
about reverse recovery and so forth)

Why are resistors and capacitors bad?
1) Power consumption - all those resistors draw a significant amount of
power, if they are going to do anything useful.  If you want the resistor
string to swamp all the variations in diode characteristics, you want the
resistor string to draw anywhere from 10-50 times the reverse current of
the diodes. A few hundred uA may not seem like a lot, but at 20 kV, a
hundred uA is 2Watts dissipation...

2) The capacitors are bad because they cause the voltages of transients to
distribute unevenly. Cheap disc caps (the kind usually recommended for
equalizing strings) have a tolerance of 20%, if not worse. A transient
voltage will divide in ratios according to the capacitor values. If you are
trying to minimize the number of diodes, and are runnning them at the
ragged edge, those variations might be enough to push it over.

3) Resistors and capacitors cost more than diodes. I haven't found surplus
HV resistors and caps (rated at >6kV) as cheap as surplus HV diodes.  Using
more diodes is much cheaper than using resistors and caps to try and save

4) stored energy in HV power supplies is bad because when things go wrong,
the stored energy damages things (burns out junctions, etc.). caps store