[Amps] Filter Capacitors

[Jim Thomson]

Date: Mon, 16 Apr 2012 09:11:38 -0500
From: Rob Atkinson <ranchorobbo at gmail.com>
Subject: Re: [Amps] Filter Capacitors
To: Jim Thomson <jim.thom at telus.net>
.
>
> ## ?when playing with conventional choke input supplies, not the resonant types that
> henry radio used, ?you are in for an eye opener. ?When you 1st turn it on, ?you get this
> huge ?yo-yo oscillation on the P-P ?v waveform, that finally settles down. ? what psud wont
> simulate is a varying load, like ssb-cw. ?Once the supply is up and running, and no more
> soft start involved, is when the trbl starts up. ? ? Every time you hit the key, ?you are slamming this
> big load on it...and you end up toggling between full and no load. ? Every time you hit the key,
> that oscillation starts up, and P-P V is sky high, till it settles down. ? On cw, that?s next to impossible.


You have to use bleeder resistors so there is always a load on the
supply, or swinging choke on the input.

### I realize that.  Put a scope across the  B+ and B-  and you will get a real surprise
when keying on CW.   Even with a massive bleeder load, they don’t work right. 




I suspect modern day amp manufacturers avoid using a two stage LC
filter because the iron costs more and adds weight. Nowadays even
good 330 uF electrolytics, a string of 8 let's say, is cheaper and
smaller. I think that once you start getting above 50 uF @ 3 KV the
stored energy gets high enough to IED on a fault. CW rigs used choke
input filters for years--actually the handbooks say you can get by
with more ripple with a CW rig--maybe it was okay because they were
class C? I'll find out eventually--I have a single 813 with a 1500 v.
supply from a pair 866As and dual LC filter to try out.

73

rob
K5UJ

##  any 2 x stage LC  filter in a power supply these days is plane nuts. Any $$  spent on
fixed and swing chokes is a complete waste of $$  imo.   Even the resonant chokes used in the
henry amps is a waste of effort.  

##  For XXX  amount of $$$  spent, you are much better off to just buy the biggest hypersil
C core xfmr you can get your hands on.....or an EI type  with real low loss plates.  Then use enough
C to get the ripple down to at least 1% or less.   Then you end up with superb regulation, both static and 
dynamic.  On a  2 x GS35B 6m amp my buddy built, he tried using the bigger plate xfmr, and just changed the
taps on it.    With 5 kw   CCS CXR out on 50.125 mhz, the B+ drops exactly 50 vdc  between  RX and TX.  
Ripple is real low, like .3%  with the 162 uf  total C used.  Ripple is  zip on 120-240-360-480 hz.   
The combo of high C + a bigger plate xfmr will result in good V regulation.  
The low ripple is just the side effect of the high C...and amounts  to a free lunch.   And no, it wont cook your xfmr either. 

##  You wont get any IED  effect with any fault either.  The 50 ohm glitch simply limits the fault current to a safe value. 
The HV fast fuse in the B+ lead will open off in less than 2 msecs.   The filter C  + 50 ohm glitch R  will  result in a
RC time constant.    The RC time constant is long enough to do some serious damage.... hence the use of the HV fuse. 
3600 vdc  divided by a 50 ohm glitch R = 72 A of fault current.   And 72A will open off a fast 1A rated  HV fuse pretty
quick.  Believe me,  you can cro-bar this thing all day long if you want with no ill effects.  HV fuse opens up, event over .
Replace HV fuse, and repeat  until you run out of HV fuses. 

later... Jim  VE7RF