[Elecraft] [K3] 8 or 5 Pole cw filters

[Guy Olinger K2AV]

On Wed, Jan 14, 2015 at 8:09 AM, Gary W. Hvizdak
<gary.hvizdak at cfl.rr.com> wrote:
> 4)  INRAD's 250 Hz filter is actually closer to 370 Hz wide at 6 dB down,
> while their 400 Hz filter is about 435 Hz wide.  That's a difference of
> roughly 21%, which IMHO is too similar to justify installing both.

There is a CW contesting reason for installing BOTH the 8 pole "400"
and the 8 pole "250": Hearing off frequency callers vs. having your
run frequency squeezed. This may not apply to you and Gary is
well-taken in his advice unless the following might apply to you.

In any contest, when running (calling CQ) there will be a sizable
number of people calling well high or low. Some have attributed it to
rigs that have poorly adjusted or defective TX offset. I suppose there
is some of that in older rigs. What I think is at root is something
more akin to color blindness, but for the ear: inability to closely
distinguish two different tones, or some degree of what some call
tone-deafness.

I think there are wide variants of this. All one has to do to be
convinced of this is to listen to a primary school orchestra, and the
variation of tonality around a string instrument note. There is a
gradual spectrum of tone distinguishing from perfect pitch to not
being able to tell the difference between 500 and 700 Hz at all. So
off frequency calling is not going away, and you either work these
folks or you don't. In a contest you work anything that moves. Calling
high or low, so weak you strain to even get a call, you try to work
them all.

When I initially pick a run frequency, I listen to find a space where
I don't hear anyone else in the 450 bandwidth. As things go, there
will usually be squeezing in later, particularly if the band you're on
becomes the main open band. When I get squeezed, I drop to the 350
bandwidth, and use the shift control as needed. With the sharp 8 pole
filters and skirt alignment, the 10 Hz granularity in shift can reduce
the offending signal about an S unit per 10 Hz shift. And for off
frequency callers this still allows you to hear most of the same
"radio real estate" as the 450 Hz setting.

There have been some number of times on 40m at NY4A where a particular
very loud Italian station (same guy contest after contest) would
establish his run frequency up about 300-350 Hz from me. He also had
moderate key clicks. I would switch to 350 Hz width (invoking the
"250" filter), and shift down 20 or 30 Hz, Plus set NB on with
settings DSP t2-7 and IF off, and I was able to copy very weak
stations not dead on my TX frequency in spite of him. Have gone on for
three or four hours like that with no drop in rate. I hear him weakly
at edge of passband, but not loud enough to keep me from copying
in-passband.

The way it worked was that the settings, particularly with the shift
offset, had the offender WELL DOWN ON THE SKIRTS OF THE ROOFING
FILTER, as well as the DSP, and so they could not pump the hardware
AGC.

It should be noted that some contesters will start with 1/3 kHz run
frequency spacing instead of 1/2 kHz apparently using the 350 width
setting.

My Filters:

Main: 2.7, 1.8, "400", "250", 200
Sub:  6.0, 2.7, 1.8, "400", "250"

The way these look in the filter settings:

Main: 2.7, 1.8, 450, 350, 200
Sub:  6.0, 2.7, 1.8, 450, 350

The 600 is for AM BC when it is good, otherwise I use 2.7 or 1.8 SSB
for listening to AM.
2.7 is required by K3 design, used for casual SSB.
1.8 is SSB contesting.
450 and 350 discussed above.
The 200 is for digging out the DX in a pileup, including S&P in a
contest. It has a gradual sharp shape which is good for that, better
than a flat top to the filter response.

Using 1.8, 450 ("400") and 350 ("250") filters I make extensive use of
diversity which needs identical passbands in main and sub RX to work
best. In S&P with diversity will often use 250 width (350 roofer), and
do not want to engage the dissimilar 200.

73, Guy.