Towertalk
[Top] [All Lists]

Re: [TowerTalk] Capacitors To Tune 160 Vertical

To: David Gilbert <xdavid@cis-broadband.com>
Subject: Re: [TowerTalk] Capacitors To Tune 160 Vertical
From: Jim Lux <jimlux@earthlink.net>
Reply-to: Jim Lux <jimlux@earthlink.net>
Date: Thu, 29 Nov 2007 20:09:30 -0800 (GMT-08:00)
List-post: <mailto:towertalk@contesting.com>

-----Original Message-----
>From: David Gilbert <xdavid@cis-broadband.com>
>Sent: Nov 29, 2007 11:15 AM
>To: Bill Turner <dezrat@copper.net>
>Cc: towertalk@contesting.com
>Subject: Re: [TowerTalk] Capacitors To Tune 160 Vertical
>
>
>I don't understand that comment.  Murphy's Law has nothing to do with it 
>... Ohms Law does.  As N6RK said, where high voltage DC is present, 
>capacitors in series are typically bridged by high value parallel 
>resistors to make the voltage divider dependent upon the resistors 
>instead of the much higher (and indeterminate) DC resistance of the 
>dielectric.  For RF, the impedance of the capacitors themselves provides 
>that same shunting action as far as the bulk resistance of the 
>dielectric is concerned. 
<snip>
>
>All that being said, not many inexpensive RF capacitors handle several 
>amps very well, so for high power applications we probably end up with 
>some configuration that includes both series and parallel devices ... 
>and I'd probably shoot for more margin on the parallel portion than on 
>the series portion.
>
Back when I was building high power tesla coils and Marx banks, I discovered 
that capacitor prices (and sizes) tend to be proportional to energy storage.  
Say you needed 1 uF at 10kV.. You could parallel 10 0.1 uF caps rated at 10kV 
or series 10 10 uF/1kV caps to get there, and the physical size and cost would 
be about the same (within limits, and assuming the same basic capacitor 
technology).

In fact, high power pulse capacitors from,e.g. GA/Maxwell, are made of subunits 
hooked in series parallel to get the required ratings.  The voltage rating 
tends to be determined by the dielectric/electrode system and there's a certain 
sweet spot for thickness vs capacitance vs breakdown (thinner dielectrics have 
higher breakdown fields in a Volts/mil sense).

These days, the best all around performance for largish values (tenths of uF) 
at medium voltages (1-2kV) is with extended foil polypropylene dielectric (e.g. 
CD 942P series for example).. they're very low ESR intended for things like 
high frequency switching supplies.  Probably pretty good up to MHz-ish (I 
haven't measured one at HF... I'll have to get around to it), but their 
outstanding characteristic is the ability to handle high rms AC currents (amps) 
without melting or blowing sections of the foil off.

There's a similar PP dielectric system where the electrodes are metallized onto 
the film as a series of plates. These have a little "fuse" (actually, a very 
thin trace) between the plate and the bus along the edge, so if a fault 
develops, the fuse blows, and the capacitor keeps on working, with a slight 
reduction in C.


In more traditional RF caps, you're looking at things like doorknobs which are 
barium or strontium titanate ceramic from companies like TDK or Vishay/Ceramite 
 .These too have DC and RF versions that look a lot alike, but are VERY 
different internally.  The RF ones are a lot less dense (pF per volume, pF per 
dollar, etc.).  The internal differences (aside from the intrinsic loss 
differences in the dielectric) are much like those for film capacitors with the 
DC ones not being able to tolerate as much RF current and having higher losses.

However, even better these days are the stacked caps from companies like ATC.  
Surface mount style with tabs, not wires, very low loss, etc.  Pricey, but work 
really well, and if you only need a few (as for a DC block in a kilowatt amp) 
they might be worth it.  As with others, these seem to be sized and priced by 
volume, which in turn corresponds to stored energy (double the voltage rating 
is 4 times the size, etc.)

Jim, W6RMK
_______________________________________________



_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk

<Prev in Thread] Current Thread [Next in Thread>