[Elecraft] Decoupling circuit -- why is it like it is?

[Bill Coleman]

On 10/13/02 12:15 PM, Vic Rosenthal at rakefet@rakefet.com wrote:

>In the KPA100 the positive line to the transistors is bypassed by numerous
>decoupling capacitors:  C52-C57, each of which is 0.047 uf, and a 470 uf
>electrolytic.  My questions are:
>
>1) Why six capacitors instead of one 0.3 uf capacitor?  Obviously these 
>are to remove RF from the +12 line, by why use more than one?

Vic, I'm just guessing, but I would suggest that perhaps it was easier to 
include three caps of a given value that was already used in the design 
than one of a new value (that would have to be ordered, stocked, 
inventoried, etc)

>Is it due to RF current ratings?

That could be a factor, as well as the inductance of the leads and 
internal construction of the hypothetical .3 uF cap.

>2) Why a large capacitor like 470 uf?  What does this one do?  Certainly it's
>not for RF!

One problem that some bipolar transistor RF designs have is low-frequency 
oscillation, or motorboating. This is because the transistors have a lot 
more gain a much lower frequencies than at HF. 

Big caps like this help to quench these low-frequency oscillations. If 
the cap value seems big, remember that these transistors are very 
low-impedance devices, so it takes more capacitance in the bypass to have 
a low enough reactance to be effective than it would in a 
higher-impedance weak signal or tube/FET circui.

>Why am I asking?  Well, there's nothing wrong with my KPA100.  But it is more
>intelligenly designed than my high-power amp.  This unit has four boards like
>the one in the KPA and puts out from 550w to a KW on various bands when it's
>working.  It has only a 100 uf electrolytic in this location -- and it blows
>these electrolytics like firecrackers on 15 meters!  I have tried 0.1 uf 
>discs
>across the electrolytics to no avail, but looking at Wayne and Eric's 
>design I
>see why!  I'm hoping someone can explain this aspect of the design to me so I
>can understand how to fix it right (the manufacturer of the amp thinks I'm an
>idiot and won't talk to me).

There's some discussion in the ARRL handbook about bypassing transistor 
RF power circuits with a multitude of capacitors, typically with values 
of 1x 10x and 100x, all in parallel. These are done with minimal lead 
length (to avoid stray inductance). On the surface, it may seem wasteful 
to used three instead of one 111x component, but the smaller value bypass 
tends to take care of higher frequencies which the larger cap may not 
pass, due to the inductance of its internal construction.

If the 100 uF electrolytic keeps blowing, it's probably carrying too much 
current, or you're exceeding it's voltage ratings. Using several caps, 
with higher voltage ratings, can help.


Bill Coleman, AA4LR, PP-ASEL        Mail: aa4lr@arrl.net
Quote: "Not within a thousand years will man ever fly!"
            -- Wilbur Wright, 1901