[Amps] Re: L networks

Ian White, G3SEK G3SEK at ifwtech.co.uk
Tue Jan 11 09:13:57 EST 2005


RZP wrote:
>
>In a message dated 09/01/2005 23:05:35 GMT Standard Time, 
>G3SEK at ifwtech.co.uk writes: The minimum possible *loaded* Q that is 
>required to transform between two resistances R2>R1, using any possible 
>circuit, is sqrt(R2/R1 -1).
>
>I think this is true for a network of two compnonents. If you allow 
>more components (e.g. my transformer), it no longer applies.

Thinking about it some more, Peter is right. A network that includes a 
transformer obviously can break that limit - an ideal transformer can 
match any two resistive impedances with a loaded Q of exactly 0.

>The final extension of this lies in the use of multi element Tchebychev 
>networks for wide range matching over several octaves.

Also true, and this time without the use of transformers. So OK, there 
is no general rule.

I'm still pretty certain that Rich is right about the L-network having a 
lower loaded Q (and hence in general lower losses) than a T- or 
Pi-network transforming between the same two impedances... or at least, 
in most practical cases.

But just because an ideal transformer can match impedances with an even 
lower loaded Q, we cannot automatically assume the same will be true for 
the link-coupled windings in a typical Matchbox. Link coupling makes a 
far from ideal transformer, so it's very hard to be certain that some 
other network couldn't do the same job with lower losses.


>The definitive work on this was published in Russia in the early 
>1960's, and I don't have the reference here.
>
Shame on you, Peter! I bet you left your copy of Zverev at home too.


-- 
73 from Ian G3SEK         'In Practice' columnist for RadCom (RSGB)
http://www.ifwtech.co.uk/g3sek


More information about the Amps mailing list