[Amps] .01uf caps across diodes

Sat Dec 17 12:05:43 EST 2022

Date: Thu, 15 Dec 2022 21:17:44 -0700
From: John Lyles <jtml at losalamos.com>
To: amps at contesting.com
Subject: Re: [Amps] .01uf caps across diodes

 The capacitor is there for the transient part of the
commutation in the diodes. All of them switch on and off together, but
any small variances due to the intrinsic capacitance in the die and
package of early diodes could cause a tiny delay in the back bias to
turn on in different diodes. It the package is swamped with the larger
external capacitor, this helps dominate the commutation to level the
playing field so that to the rectified voltage, the diodes all appear to
change at the same time when it crosses zero.

We have a bunch (8) three phase 86 kV DC power supplies rated ~ 1 MW
each, using a full wave bridge in oil.  Then there are 6 boards with
this arrangement. So the PIV rating of a stick is 51 kV, and there are 6
in series to get ~300 kV PIV per leg of the rectifier. This is a
significant safety factor but those things run for years, 24/6,
recharging the capacitor bank of 120 uF, 120 times a second. There are
capacitor and resistor compensation across every diode, and I am happy
with that design. It all depends on your tolerence for risk. In our
machine where downtime is very costly, we use RC networks on the
rectifier diodes, not for noise but to protect the diodes from worst
case events. This is my take on it, to skip the RC if you have only a
few diodes and they are modern, matched in characteristic and you feel
they are running well under their PRV and peak current rating.

<73
<John Lyles
<K5PRO

##  John, if 3 phase is used, the HV filter caps would get charged up 360
times per second, not 120.  (6F, not 2F).

# a few years ago, I measured exactly 900 pf across each of 100  x 6A10
diodes...in a box of 100.
The following I have NOT done..yet.  The plan was to use the battery
operated  Fluke scope meter, and try and measure the transient  across any
one diode in the string.  And done with various loads, like from min to
max.  I want to see what it looks like, esp with a high C filter cap assy
on the output.  Then repeat for 2 or more of the diodes, then the entire
leg. I don't know if the transients every 8.3 msecs add, with diodes in
series.

##  I still can't fathom how any transient  can get past the HV filter
caps. Modern lytics have much lower ESR as freq increases. They have
reduced the internal inductance to almost nothing. (strap used internally
instead of wire per the cap manufacturers). The ESR at 1 khz is less than
1/3 of the same cap at 120 hz.  And through the floor at 10 khz.

##  W8JI insists that a  .01uf cap be installed across each of the 5 x
diodes per leg on his 3 x biggest tube amps.  Those amps only have 5 x
diodes per leg ( 20 in total)..and typ unloaded B+ is 3700-3750 vdc.  That
leaves very little margin / wiggle room.

## KM1H, Carl, used a 2nd method, with a .01uf across each leg of the FWB /
FWD. I have .01uf caps in 5 kv.  You can also get .01uf disc caps  in 10
/20/30 kv.

## Dahl used a MOV across  each 6A10 diode on his FWB assy's. ( 4 x legs on
the single phase FWB...and 6 x legs on the 3 phase FWB). I saw 2 other
commercial manufacturers  of rectifier assy boards, used for broadcast use,
that also used a MOV across each diode, again both single and 3 phase.

##  I bought 2 of the dahl single phase assy's, 1 month before he closed
shop.  8 x 6A10's per board..with mating MOV's.  3 x boards in series for
each of the 4 x legs. 12 x boards in total.   24 x diodes per leg.  He
rated  the entire assy at 8 kvdc @ 12 amps CCS.  (He rated the 3 phase
version at 18 amps).  I assume since the diodes only have a 50% duty
cycle,  is how he came up with the 12 amp CCS number. ( 4 amps is more
realistic). Tech notes state for a 6A10 ( 6 amp @ 1 kv), that 1.1 sq inches
of heatsink required on the trace on any board...( input of 1st diode..and
output of last diode). With diodes in series, it's then 2.2 sq inches at
each combined junction.  I never see that done. In my final config, I
paralleled 2 x complete FWB assys, and done such that the current splits
50-50.

## The place for a MOV would be right at the main 200 amp panel.   Wired
hot to neutral / grnd, and 2 used in total.  And each mov fused / 2 pole
breaker.  If used in a sub panel, or at the input of a plate xfmr, they are
fused..and wired hot to grnd. Never wired hot to neutral.  You don't want
any fault current flowing on the neutral, ( if neutral used).

Jim  VE7RF