[Amps] HV lytic max V ratings..some notes.

[Jim Thomson]

Date: Sat, 10 Apr 2010 22:00:40 +0800
From: Alek Petkovic <vk6apk at bigpond.com>
Subject: Re: [Amps] HV lytic max V ratings..some notes.

At 09:51 PM 10/04/2010, Carl wrote:

>I dont know if it still holds with some caps but they used to have a memory.
>IOW if run for a long spell (weeks, months) at say 60% of ratings and then
>run at max they had a tendency to fail.
>
>Carl
>KM1H


Yeah, that's what I was always lead to believe. I was told that it is 
better to run your caps very close to their voltage rating. Dunno if 
its a wives tale or not.

I usually sail pretty close to the wind. I work on having 1 spare 
capacitor worth of headroom in a bank. For example, if the dc voltage 
is 4500V, then I would use 11 x 450VW capacitors in the bank. Never 
had a cap go bang on me by using that crude formula.

73, Alek.
VK6APK

## When firing up old caps that have been sitting around in box's, or these surplus
ones I have been buying, this is the procedure I use. 

## I insert a fluke dvm  between fwb  and B-  of caps. [using 10 kv rated leads]
 I TEMP bond the B-  to chassis  for safety purposes.  This does not affect the B+
meter, as it's wired directly between B+.. and B-.     Using a small, FUSED variac,
I bring the supply up slowly, over 10 x seconds.. to 500 vdc.. then  let it sit  for
30 mins.   Then repeat the process to 1 kv... then 1.5 kv etc.   

## The DVM  will show any leakage current.    They sit as long as needed,  till any leakage
drops  to zero.   Then increase some more.  

##  I have never had a problem with the... 'memory issue'.. and this is the 1st time I have even
heard of it.   Now, say if the  HB  B+  supply   USED  to be run at 3 kv... and caps  were run at 
75% of their ratings.... and  now  I want to run the caps  at   90%  rating.   4500vdc /4950vdc
= 90%    using aleks  example,   then in goes the variac,  [ step start  temp disconnected].. and the
forming process done again.... but to 90% of the strings ratings.    Then all temp stuff removed,
etc.. and B+  supply now run at the higher vdc.     Never had any problems  using the variac
procedure... regardless of whether the caps   were originally used at 65-75% of their rating...
 then run at 90+%  or if old caps  used, etc. 

## There are a LOT of different series of caps  these days.. in both CD's  on line catalog..
and also united chemi con /nippon chemi con's on line catalog.   Heck, you can still buy the 
old style  CGS  series of hv caps... and also the  old style caps  from other makers. 

## Dunno if the 'memory'  issue  pertains to old style caps.. new style.. or both... or if it's
an anomaly of one style of caps.. from one maker.. or just an  anomaly.   I checked tech notes
and eng application notes  from every cap maker... and also real old tech notes from the same cap
makers... and can find NO references to the ..'memory issue'. 

##  In the past.. I didn't like the idea of running caps at 90%  of their V rating..esp with the
carbon comp eq resistor's drifting, etc.  A transient  off the street could double in peak V.. b4 any 
mov was fast enough to clip it off.   The newer modern style, "high capacity in the can"... caps.. used
with good close tol eq resistors, up to 90% of their ratings is no problem... provided leakage is not
too high.  If it is, then the value of the eq resistor has to be looked at... and probably lowered.    

## These super big uf caps  have higher leakage current than the smaller caps. [4800 uf  vs 560uf] 
Since I have excess capacitance to burn .. and CD's  notes  said for max long term reliability, to run em at 
75%.. that's what I have been doing.    Then the leakage current is through the floor... and  I still  have bucket
loads of uf available. If transients / spikes get past the movs, bypass caps,etc... then the hv lytics  HAVE to
absorb em.  The caps, of course, look like a dead short  to a high voltage transient/spike.. and the caps  will
never charge up to the higher value.   In any event,  you generally  want the bleeder current of the eq resistor's
to exceed the leakage current by a ratio of at least  15-1 .   When I use  CD's formulae  for value of  the EQ resistor
[for caps in series].... it comes out a helluva lot lower than 100 k ohms.   I use 100K anyway. Then I always end
up with the 15-1 ratio... even if the caps  were run at 100% of their  V ratings. 

##  For safety reasons... when experimenting with these high C filters... I wanted PLENTY of vdc overhead. 
[2000-5200 uf variety]  

later........ Jim  VE7RF