K3 roofing filters

[K7TV]

My 2 cents worth:
Since you are primarily a cw operator, you may now or in the future want
the roofing filter that is the ultimate in protecting the DSP in the
toughest possible conditions such as when you are running big antennas
in a contest on the lower bands. This would mean a very narrow roofing
filter. Under other conditions you may encounter conditions that are
less severe, but would still cause the DSP to be overwhelmed if you were
to use an ssb roofing filter. It would then make sense to have a
moderately wide cw roofing filter. Of course, since it is easy to add
filters, for many it makes sense to get the latter filter first and add
the sharpest filter later if you find you need it. But plan ahead; if
you see a possiblity that you will buy a sharper filter later, don't
pick the first filter such that it is too similar to that sharp filter
to come later. (This advice is given based on your main focus on the cw
mode. If you were not mainly a cw operator, it might make more sense to
plan for only one cw roofing filter even in the long run.) So, plan
ahead right now and decide what is that ultimate narrow filter to you.
In practice, that means either the 200 Hz 5-pole or the 250 Hz (in
reality about 325 Hz if I remember correctly) 8-pole filter. In other
words, what is more useful: the narrower bandwidth at the peak of the
5-pole or the steeper flanks of the 8-pole? I have not made a practical
comparison, but my choice was the 200 Hz filter based on a rather
simplified argument about what the roofing filter's job is: Most of the
time even an SSB roofing filter is fine for cw operation. The need for a
narrower roofing filter appears when there are too many, too strong
signals in the gaps between the roofing filter passband and the narrower
DSP passband (the portion of the band that you are actually listening
to). Things are fine until the combined voltage of those unwanted
signals as well as the wanted signals (within the dsp passband) exceeds
the input voltage capacity of the analog-to-digital converter (or
actually the threshold of the hardware agc that will prevent that from
happening). While in most casual operation situations there will be no
problem even with an ssb roofing filter, most of those situations that
require a cw roofing filter will only need the signals in the gaps to be
attenuated by a moderate amount, say 10 or 20 dB, to get the job done. I
don't see any direct advantage in picking a roofing filter that provides
an ultimate attenuation of, say 80 dB over one that provides just 60 dB.
I don't remember the ultimate attenuation numbers for the 5-pole and
8-pole filters, so I am just making a general argument here when I
suggest that ultimate attenuation is not a reason to use 8-pole filters
for the particular reason of ultimate attenuation in selecting a very
narrow cw roofing filter. OTOH, the steeper flanks of an 8-pole filter
are obviously helpful. How helpful? That depends in the particular
situation on how the offending signals are distributed in the passband
gaps. Also, while the flanks of the 5-pole filters are not as steep,
they may provide better attenuation on a given signal due to the fact
that the curve starts closer to the primary wanted signal. Without the
benefit of actual comparison, I simply figure that the competition
between the filters w.r.t. gap attenuation may be a wash. But remember
that signals within the wanted passband also contribute to potential ADC
overload. It is therefore very helpful to crank down the dsp bandwidth
as long as the roofing filter bandwidth is correspondingly reduced. If
we can go from a 325 Hz (nominal 250 Hz) roofing filter to a 200 Hz one,
we are reducing the onslaught on the ADC very substantially, by reducing
the wanted passband (I assume that the dsp setting changes in the same
way), and this in addition to what may happen in what I have called the
gaps. Bottom line: I think the 200 Hz filter (and not the 250 Hz one) is
the ultimate narrow cw filter. While other list members may have more
experience with practical comparison, I don't remember reading about
anyone claiming otherwise, for strict cw operation. Of course if you
just can't stand listening to just a 200 Hz slice of the band, even in
the toughest contest situations, your choice could still be different.
Ditto if you are not so much a cw operator as an RTTY operator.
Anyway, my own choice was the 200Hz, and I have not regretted it. Now
getting back to that other cw filter, which may be the only one you buy
to begin with. As many people have pointed out, it makes sense to have
significant differences between the roofing filter bandwidths. If the
narrowest is 200 Hz, then I wouldn't consider anything narrower than the
400 Hz for the "normal" cw roofing filter. You probably want something
wide enough that you can hear who is next to you. Many people want a
really wide cw listening bandwidth for everyday casual operation. The 1
Khz filter seems at first to be intended for those people. However,
while I haven't tried that filter myself, it is my understanding that in
practice the K3 does not allow you to listen to say, 200 Hz to 1200 Hz
cw tones anyway. If this is so, then the unused bandwidth of the 1kHz
filter represents just the potential for such problems as the roofing
filter is supposed to prevent in the first place. For people who want
the widest cw listening bandwidth, the 700 Hz filters from an
independent source may be the ideal choice. In my case I just got the
500 Hz 5-pole filter, as it wasn't clear to me that the sharper flanks
of the 400 Hz filter would be enough of an advantage against the reduced
bandwidth. Since I occasionally use digital modes, I do see an advantage
in the flatter passbands offered by 400, 500 and 700 Hz 8-pole filters.
Some digital modes may actually use actual bandwidths of 500 Hz or more,
and the 400 and 500 Hz filters are obviously not perfect for such
applications, but I am answering your question as coming from a cw
operator. Finally, you have read test results where 8-pole filters have
been shown to provide slightly better numbers than the 5-pole filters. I
am sure those test results are valid, but remember that those tests are
are conducted with an unwanted signal at a single location such as 2 kHz
away from the wanted signal, and this is very different from the real
world contest situation with multiple nearby signals, many of which are
even closer. More importantly, the K3 with any of its narrow filters, as
well as the K3's top competitors such as the KX3, are producing numbers
that are so good that they are way beyond what you are likely to need in
practical operation. Thus I have no reservation in recommending the 200
Hz and 500 Hz 5-pole filters. For SSB I like the 1.8 kHz filter in
addition to the standard or 2.8 kHz.
73, Erik K7TV
______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Edward R Cole]

Excellent answer from Don, W3FPR.

I fit the mostly SSB type with sub-Rx, so I use diversity
reception.  I chose for the main Rx:
13-KHz (for AM and FM)
2.8-KHz
400-Hz

The sub-Rx has just the 2.8-KHz filter

I chose the 8-pole for ease in set up for diversity Rx and for the
steeper skirts (which I may not really needed up here in AK, but I
figured made the resale value better).

The DSP bw control works fine and I often narrow the high split on
weak or noisy SSB to improve reception.  I keep the low split at
200-Hz.  The 400-Hz filter makes CW reception of a single station
possible.  400-Hz filter is really amazing.

I have not used it on CW eme but expect it will provide some
advantages in narrowing the noise bw.  Typically, I run 100-Hz for
eme CW once I have the signal tuned.  Scanning for eme signals I use
the wider SSB bw and waterfall displays to detect the signal.

Most of my eme on 2m is using digital mode so no filter is used as I
export the IF of the K3 to LP-Pan and therefore to a
soundcard.  Digital eme is watched not heard.  I actually have the K3
select an empty filter slot when running DATA-A which eliminates a
center frequency gain "suck out" (black zone on zero freq display).

73, Ed - KL7UW
http://www.kl7uw.com
[hidden email]
"Kits made by KL7UW"

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html