[HBR] That General Coverage HBR Project -- 4

[Walt Hutchens]

The crystal filter is finished for now.  It's a ladder design, with five 
crystals at 4096 kcs.

A ladder is just a few crystals of the same frequency in series, 
capacitors couple from each end to the input and output load resistors 
and from each junction between crystals to ground.  All six caps in this 
one are 22 mmf.

I bought 20 crystals of the same frequency (about $10 worth) and got two 
sets of 5 that were within a range of 7 cps of the same.  The caps are 
within about 1/2 mmf, matched with a capacitance meter.

There's a picture at:

http://www.timbreblue.com/radio/crystal_filter.html

The colors aren't quite right and the reflections on the back wall make 
it a bit confusing but I think you can see the ladder circuit diagram. 
It's 2" long, 7/8" height and depth, made from tin can stock.  The end 
flaps bend up and I'll cover the top and front with another tin can 
stock section when the receiver is done.

I drilled holes in the back wall and soldered the cases of the five 
HC-49US (short HC-49/U) crystals in the holes, then bent the leads 
together in a zigzag pattern and soldered the caps from each to ground. 
  The two bluegreen feed throughs on the ends are threaded and will be 
used to mount the unit.

The best explanation of experimental design of ladder filters that I 
found is the QST article by Wes Hayward, July 1987.   He gives all the 
info you need to tinker one of these into existence: more crystals = 
steeper sides and fewer bumps, smaller caps = wider bandwidth and higher 
impedance, higher terminating resistors = smoother passband but less 
steep sides.

Use of equal caps makes this a Cohn filter -- minimum loss, a compromise 
on bumps in the bandpass.  Other articles discuss other designs with 
minimum bumps or steepest sides.

I did a crude check with another receiver: it does pass a signal and 
with 4.7k termination resistors the passband is pretty smooth but using 
a grid dip meter as a signal source that's about all I can tell. I'll 
probably add shields inside the filter housing at each end so the input 
and output connectors can't 'see' each other -- at 60-80 db down, even 
small amounts of coupling matter.  There's no problem under the chassis 
because the filter input is in the 2nd mixer compartment and the output 
is outside it.

I'm sure the terminating resistors will need to be changed once it's in 
the set.  But it's good enough to proceed with the BFO and other work!

Walt
KJ4KV