[Elecraft] K3 2.7 k Filter vs. 2.8 k

[Bill Tippett]

         This topic has drifted from "2.7k versus 2.8k" to "2.7k
versus 400 Hz".  *Of course* a 4-500 Hz filter is needed for CW
and will have vastly better IMD/BDR than a 2.7k/2.8k.  But I
repeat that there will be very little if any difference between a
400 (actually 435 Hz) 8-pole and 500 ( 565 Hz) 5-pole.  In my
opinion, you are really wasting money to buy 8-pole filters for
improved RX performance, unless the 130 Hz BW difference
is important. In practice, this implies the difference in a signal
spaced 435/2 = 218 Hz versus 565/2 = 283 Hz...65 Hz is not
much difference, and I actually prefer the wider BW to catch
more off-frequency callers.  From Eric's posted data:

Filter            20kHz  10kHz  5kHz  2kHz

400 Hz, 8 pole    100+   100+   100+   95
500 Hz, 5 pole    100+   100+   100+   94

Again in my opinion the 250/200 Hz filters are
redundant and unnecessary if you have a 400/500.
They do not improve 2 kHz IMD significantly as
seen below (1 dB difference is meaningless as
that is well within measurement error):

Filter            20kHz  10kHz  5kHz  2kHz

200 Hz, 5 pole    100+   100+   100+   95
250 Hz, 8 pole    100+   100+   100+   95

You could argue that the 250/200 would be better
for IMD fom extremely close-spaced signals (e.g.
<200 Hz spacing from your TX frequency), but at
that spacing other factors such as the transmitted
signal's phase noise, key clicks, etc. will override
any theoretical IMD advantage (i.e. the IMD becomes
"noise limited" in ARRL terminology).

         Remember also that Passband Tuning can be used
to shift a 400/500 Hz filter if you actually do need
to eliminate a signal spaced at 200-250 Hz from your
TX frequency...not that it would actually do any
good to eliminate phase noise or key clicks.

                                 73,  Bill  W4ZV