[Elecraft] HaroldZ, Variable bandwidth filter, etc

[William Carver]

What is the two tone dynamic range of the K3 when both signals are
inside the roofing filter bandwidth?

Close-in IMD measurements are virtually always done using a CW bandwidth
filter. None of the published data (ARRL, Sherwood Engineering) tests
IMD inside the DSP process without xtal filter selectivity.

My experience with a homebrew K3-like topology, except a strong diode
second mixer (instead of NE602-type mixer), is the limiting factor
INSIDE THE ROOFING FILTER bandwidth can be the DSP performance. If you
put a FLEX DSP processor behind a K3 2.8 KHz SSB filter you would have
FLEX numbers close in, K3 numbers outside the roofing filter. The
advantages of both. 

So rather than bandaid by buying more filters, or push for a variable
filter, if the dynamic range inside the filter is not good enough one
should concentrate on (1) strengthening the post-xtal-filter intercept
of the receiver circuits and (2) reducing the close-in phase noise of
the LO. 

Bill - W7AAZ