In its original state the variable xtal filter has rather broad shirts. This is caused mainly by (a.) the filter being variable and (b.) the filter itself not being separated from the RF-board and thus not being contained in a separate shielded case. The upside is of course that you get an xtal-filter that is inexpensive and easy to solder for the builder.

In order to improve the xtal filter response I begun by implementing the following two modifications.

As C166 is not needed any longer it was removed, resulting in the following:

In addition I experimented with the grounding around the xtals. After many tries, comparing the results on Spectogram, I ended up with what is shown on the pictures below.

The inductor shown in the pictures is T7A.

The graph below shows what the six 0.47uF capacitors do to the filter curve.

The gray curve is before and the black is after these have been instaled.

Bandwith 200 Hz, AGC off, Preamp on, input signal from Elecraft N-gen.