"Floyd Sense" wrote in message
...

Dale - regarding homebrew crystal filters - I wonder if you or any others
here have looked at Elecraft's crystal filter implementation in the K2?
It's built with discrete crystals and the bandwidth is variable over a
large
range. Haven't seen that design exploited elsewhere, but don't know why.
It sure works well.

K8AC

That filter topology was, I think, first used by Ten Tec in their Scout.
From a purist standpoint there can be problems:
1. Asymmetrical skirts
2. Changing IN/OUT Z with BW
3. Shifting center freq as the BW is narrowed ( capacity loading)
4. I am guessing here, but I would think the group delay would suffer at one
end or the other
5. Passband ripple can become excessive as BW is increased.

I looked at this approach, along with a method described in QEX last year
for dealing with the center freq shift, but came to the conclusion that I
could design a much better filter if I optimized each BW- even though it
took 3 filters.

One aspect of filters that has long been either ignored or overlooked (
even though W7ZOI details it in EMRFD) is group delay. A friend (N2PK)
recently sent me 2 WAV files of a Yaesu 1000 stock 250Hz filter and his
homebrew 250 Hz filter. The difference was remarkable- his filter was much
cleaner and crisper sounding, no ringing. Great care was taken with group
delay in his design.

Dale W4OP