[Amps] Disc ceramic capacitors

[Jim Thomson]

From: Jim Thomson 
Sent: Tuesday, April 23, 2019 7:15 AM
To: amps at contesting.com 
Subject: disc ceramic capacitors

> What parameters or characteristics should one seek out when selecting 
> disc ceramic caps for power rf applications such as plate coupling and 
> padder caps used on the pi network.
 
##  I have yet to see a disc ceramic that even list a RF current rating. 
Typ, larger diameter  disc ceramic caps  will also handle more  RF current. 
I just bought 2 dozen  .01 uf  at  1.3 kv  rated  disc ceramics.  These are 
25mm  diameter types.  They are also thicker bodied....vs the smaller  15mm
diameter   .01 uf at  1 kv types.   How much current a  disc  ceramic will  safely
handle is largely determined by  experiment.  Alpha uses several  paralleled  6 kv rated  disc ceramics
for their plate block caps..which will handle the current on 10M. 

##  For  coupling and bypass  caps,  temperature coefficient  and actual value
is a non issue,  provided the value  chosen meets specs.  IE: sufficient C used for the
lowest freq involved.   For bypass caps, typ a higher value is used for the lower bands, to ensure
low XC. 

##  For  plate block caps, they have to be able to handle both the  DC  bias.... IE B+
AND the AC  voltage drop across the plate block cap.   Plate block caps have far more  current 
through them on the highest bands, like  10m  +  6m.... vs  160m.     All the  current that flows through the 
anode to grounded grid in the GG triode, also has to flow through the plate block cap.  You also have to
factor in all stray C, from anode to chassis.    RF  voltage across the....  anode to grounded grid  is assumed
to be  aprx  60%  of key down loaded B+  value.   Once you measure the total C  from anode to  grid..with tube
in socket, then its an easy calculation to determine  the XC...on  band extremes...like  160m..and also  say 10m. 

##  On 10M, the tube C  will make up most of the  total  tune C.   On 10m, the C1  tune cap is basically  a tiny,
variable padder, in parallel with the  tube C.  Problem is, the plate block cap is in between the  tube..and the  C1
main tune cap.  ALL the  calculated  RF current that flows through the tubes  C, also has to now  flow through the
plate block cap.   Any plate block cap has to be sized to handle calculated  rf current on 10m...for modes like FM, and  RTTY.

##  actual pf value for the plate block cap is a moot point.   100-170-500 pf will  work just fine on 160m, provided that it will  also handle 
the calculated current on  160m.     You cant go by  KVAR  ratings either.  You can see this on  the  HEC  ( high energy corp)  site. 
Use their  charts and graphs to determine  the biggest bang for the buck.    For  example, a  200 pf  HT-50  or  HT-57  cap  will handle a lot
of current.  Henry radio used  4 x 200 pf caps in parallel on its  4 k ultra, which is a 80-10m amp.  A single  200 pf    HT-57 cap will handle
15A on the upper bands.   On the low bands, the saving grace is there is very little current flowing through  a plate block cap on 160m.
Even with lower values of C,  on 160m, the  RF  AC  V drop across the plate block cap is very little.    Calculated XC  X   current =  V drop
across the cap.   That calculated V drop across the cap  + ....  B+  voltage, can not  exceed the  V rating of the plate  block cap.  

##  Padding caps for padding say a tune cap..or a load cap,  different ball game.  For padders, they are now part of a tuned circuit. In this case
we dont want the padder to start drifting in value.   Here  temp  coefficient has to be taken into account..and also the current rating at the freq in use..  
Padders  are typ used for padding a load cap on 160m.   For a simple PI net,  1st calculated the  total  required  load C.   Use the largest practical
value variable load cap you can get your hands on..that will also fit into the amp, for say a hb amp. .  That will minimize the required load C  for the 
worse case, typ   1800 khz..and some swr. Peak V across  both the variable load cap and also the padder, will be the same as the peak V across 
the  dummy load.    Calculate the XC across the padder cap.  RMS voltage across the dummy load...divided by the calculated  XC  of the padder
= the current  flowing through the padder.  Knowing all of that, then its a simple matter to figure out how many paralleled padders  required.   A single
padder might barely suffice in some cases. In other cases,  2 or more padders..even though their  TC is not as good, will  easily work well. With padders that
are one step worse than a NPO cap, their current rating has to be examined.   IF each of the the paralleled padders is handling  minimal rated current, their  TC
is not an issue, they wont drift, cuz they are not heating up in the 1st place.     

Jim  VE7RF