The #1 criteria is to sort the crystals by frequency. Use an oscillator and an accurate frequency counter.
Be sure to measure them all exactly the same way. Choose the set of 5 which are closest in frequency, certainly within
50 Hz of each other; the closer the better. Choose the closest 4 for the filter (and a fifth for the BFO). If making a G3UUR 8 crystal filter, use the same criteria
for 8 crystals. The rest is mostly finding the proper capacitances to make the bandwidth you want. This method will produce
a perfectly good filter, though it is not nearly as rigorous as those usually suggested.
The exact frequency doesn't matter a whole lot; it will be somewhere around 12 MHz. You will choose or make a IF to match it.
But the bandwidth does matter. Really, one should make 2 filters if using SSB and CW. But I have seen good results by using varicaps
to change the bandwidth, and also changing slightly the IF by loading the BFO crystal a little differently. With a DDS, it is a relatively
simple matter to match frequencies to the filter...but the problem of wide or narrow bandwidth for SSB and CW remains. Note that
a crystal filter does not have a 50 ohm input and termination by itself (closer to 200 ohms). Usually one adapts a capacitance or transformer interface to
match the rest of the circuits.
You should measure at least 50 or so crystals to insure getting a good match. I purchased a set of 200 KDS 11.095 MHz crystals and found
at least ten good matches, to within 10 Hz. The results with smaller crystals or "generic" ones is not so good, but possible.
Mainly, match your crystals using the same oscillator conditions, and you should be ok.
Hope this helps.
john
AD5YE
