Re: Diode Strings

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: tesla-at-lists.symmetric-dot-net 

On Wed, 18 Aug 2004, Tesla list wrote:

 > Original poster: "Richard W." <potluckutk-at-comcast-dot-net>
 >
 >
 > Hi all,
 >
 > I have 200 1000V 6A diodes (6A10). Recovery time is ~2.5uS
 >
 > 
<http://pdf.searchdatasheets-dot-com/pdf/266/266448.pdf>http://pdf.searchdatasheets-dot-com/pdf/266/266448.pdf
 >
 >
 > I hope to use 30 in each leg of a bridge to rectify 14.4Kv from a pig.
 >
 > 6A10s are general purpose diodes so I doubt they're manufactured with
 > comparatively tight tolerances. Seaching the mailing list I've read
 > something that the first diode to turn off can take the full applied
 > voltage and could cause failure if the diodes used aren't at least close to
 > being matched.  This seems plausible if the diode turned off at the peak of
 > the AC voltage swing but that doesn't seem likely. If the diodes begin to
 > turn off relatively close to zero crossing then I don't see that it would
 > matter so much. Say I have 30 1000PIV diodes in a string with 15Kv applied.
 > Close to zero crossing the the collective voltage drops shouldn't even be
 > close to individual diode PIV. I know there are many that have had success
 > using 1N4007s but I'm just curious and maybe a bit cautious.

You're going to want double the peak input voltage per leg. 14.4kV comes
to just under 22kV, so you're going to have problems.

As for recovery time, what I found (the destructive way of course) is that
when you rectify the input for a tesla coil, you aren't dealing with a
simple 60hZ setup. You will get RF back into the rectifer assembly and it
will burn up pretty fast withough compensation, fast or ultra fast
recovery diodes or some sort of resisitor between the rectifer assembly
and your tesla coil, to drop the Q and resonant frequency (of the power
supply itself) so and ringing doesn't destroy your diodes.

I destroyed a rectifier assembly (no, not using the Sanken diodes I'm
always trying to sell) rated 50kV or something close per leg as I recall
with a 15kV neon sign transformer. I'm going to try again with a
compensated rectifier stack rated 48kV per leg, and a output resistor to
drop the Q. When and if that works, I'm going to test out rectifiers made
of ultrafast recovery and avalanche diodes. Neither should be compensated
as I understand, especially when using avalanche diodes. We'll see what
happens.

KEN