>I did a bunch of reading on crystal filters and came away even more
>confused then when I started. Most of the descriptions that I read
>either were emperical designs, or the design was referenced as being
>covered by one of the books in the bibliography. "Just build it this way
>and it should work" was a common phrase in these writings.

Mike,
I can sympathize. But handwaving explanations don't hack it
when describing these multi-coupled circuits. So where does that leave us?
It leaves us with the "black-box" situation that you describe. Don't
knock black-boxes too much - it's a very powerful analytical tool
in breaking down circuits.
Xtal-filter is just one of those black-boxes where peeking inside
reveals a lotta nasty math. Do you really wanna know where all those
coupling coefficients come from?
Me - I'm math-adverse. After peeking inside, I'm content to use the
design-tables =:-O
But it IS useful to play about:
What happens if you design for a wider bandwidth?
Coupling caps go down. Terminations go up.
What happens if you use low-Q crystals? In-band attenuation goes up.
Passband edges become rounded.
What happens if your terminations are too low-Z? Passband ripple looks
like a rollercoaster.
What happens when your terminations are reactive? Passband gets skewed.
SPICE simulations can help a lot to show this kinda stuff: use the
book designs as a starting point, then play-about with SPICE.
-Glen VE3DNL

>The thing I am having trouble with is this - why doesn't RFC1 effect the
>filter's shape and why doesn't C12 effect the impedance matching in this
>"swapped" position? Or is it that they do, but not significantly? I am
>intrigued by this because, as a neophyte designer, it isn't intuitive to me
>when one can "get away" with this.
>
Tracy,
These components DO affect impedance match (and therefore filter
shape as well).
RFC1 and C11 form a "L-network" to match U1's output Z of 1500 ohms down
to the filter's required impedance of about 350 ohms.
However C12 is in there to modify Y1's series-resonant-frequency up a
little higher (by 106 Hz) than Y2. It is not part of the impedance matching
network.

I've gone thru the xtal filter design on my elmer101
webpage...Cohn filter
http://epic.mcmaster.ca/~elmer101