I have been working on modifying my ATS-3, to flatten it's
crystal filter response. It looked to me that the original terminations of the
filter were causing a large amount of ripple in the passpand. The 
filter is driven by a transformer and looks like a 60 ohm termination. The
output termination of the filter is the input of a NE602, and looks 
like 1.5k ohms.

My modification added an L network to transform the 60 ohms up to 300
ohms and I added a resistive termination (ac coupled) to the output 
side to drop the 1.5k to 300 ohms. I also dropped the value of the 150 pF
caps to 100pF, and now the filter has a  -6dB BW of 300 Hz with 0.3dB
p-p ripple.

The differential lowpass stage following the '602 had a peaked response
and was tuned around 400 Hz. It's response was also flattened and its
rolloff starts at 850 Hz. It's gain was also lowered by 6dB to
compensate for the 6dB gain that was added with the crystal matching
mods.

The final stage is a multiple feedback bandpass filter, and it was
changed to a lowpass with similar rolloff as the differential stage.

The mods are:

crystal filter:
input side-series 3.9uH, and shunt 220pf, added between transformer and
input series cap to filter.
output side-374 ohms in series with .01uF connected shunt to ground at
input to '602.

differential lowpass:
C16 changed to .01uF
C20         "         220pF
R1, R2     "         20K
R3,R4      "         200K
220pF added in parallel with R4

output stage:
R5,C28 changed to 6.81k
C27              "         3.01k
R6                "         .047uF
R7                "         .01uF

Thanks go out to Nick, WA5BDU and Mike, W1MT. Their interest fueled
these mods. Anyone interested in the sweeped response of these mods,
reply to me direct. A kit of parts will be made available if there is
enough interest.

72,
Craig NR4E
Melbourne, FL