[Elecraft] K2 Filter Shape Factors

[Bill Tippett]

	I just went through filter realignment using my
preferred Pitch setting of 400 Hz.  I did this simply by
using the XG1 as a signal source and adjusting all CW
filters for equal passband difference on both sides of zero
beat.  I used the K2's meter and tuned for one display
bar down (S7 with no flicker) on both sides of the XG1's
calibrated S9 output.  I have no idea how many dB down
S7 is but that's not critical to centering the passband of
the filters.

	It seems to me that the filters have some unusual
shape factors, and I wonder if anyone has noticed the
same?  For example, here are the 2 S-unit bandwidths
I measured after adjusting my BFO settings:

Filter	S9-S7 BW*

1.50	190
1.50R	190
0.70	340
0.70R	350
0.40	210
0.40R	200
0.10	190
0.10R	180

(where "R" means CW Reverse.  )

*For the 100 Hz filter, I used the S7-S5 transition since
the XG1's S9 output is attenuated to S7 by that filter's
higher insertion loss.

	Each filter has narrower ultimate rejection than
the previous, which is easy to determine by tuning off-
frequency and scrolling through the filters, but the 1.50
must have a strange shape factor since its 2 S-unit BW
is actually narrower than both the 700 or 400.  Has
anyone measured the 1.50's shape factor?  (Not on
KI6WX's CW Filter Mod page).

	I also noticed on KI6WX's page that the post-mod
400 Hz filter has a slightly better shape factor than the
700 Hz, even though their 6 dB BW is nearly identical
(330 vs 340).  This makes me wonder two things:

1.  What bandwidth setting optimizes the shape factor
for the K2's filter?  It must lie somewhere in the ~400 Hz
range but does anyone know for certain?

2.  Would I be better off to choose only that "optimum
shape factor" filter and only have three choices (wide/1.5,
narrow/100 and best shape factor around 400 Hz?)

	I noticed from ARRL's Expanded Test Report that
the K2's excellent 5 kHz IMD performance falls apart for
narrower spacings (~90 dB at 5 kHz versus <70 dB for
1 kHz spacings).

http://www.arrl.org/members-only/prodrev/pdf/lab/k2.pdf
(top graph on page 20, using K2's 700 Hz setting, actual
6 dB BW ~350 Hz.)

This makes me feel I should find the best shape factor
setting in order to optimize IMD performance for close-in
interference.  Maybe the next-generation K2+ will use
better filters which should improve IMD performance at
closer spacings!

	Hopefully someone has been down this road
before and knows which BW optimizes shape factor.

				73,  Bill  W4ZV

P.S.  Just for fun, I tried to work 3B9C on 160m last night
using my mighty 14 watts to a 3 element vertical array.
Humbling experience but I'll keep trying!  :-))