[Elecraft] K3: KRX3 question

[Bill W4ZV]

Joe Subich, W4TV-3 wrote:
> 
> 
> This is not the case based on careful measurements.  The problem 
> is 'skirt sharpening' when multiple filters are cascaded.   Where 
> a single filter might have a 6dB bandwidth of 225 Hz ("200 Hz" 
> roofing fitter), when combined with the 200 Hz DSP filter the 
> effective -6dB bandwidth of the 200 Hz DSP filter and the 200 Hz 
> roofing filter in cascade is 150 Hz! 
> 

I decided to make my own measurements using an XG1 at 50 uV and the internal
dBV meter in the K3 (for 0.1 dB and 1 Hz resolution).  These could have some
errors due to not being swept narrowband measurements, but they should be
consistent which is sufficient for comparison purposes:

1.  My "200 XFIL" 6 dB BW is actually 203 Hz (measured by disabling all
other XFILs and setting DSP = 500 so there is no cascading effect).  

2.  My "200 DSP" 6 dB BW measures 187 Hz (measured by enabling only the 500
XFIL so there is no cascading effect).  

3.  My "250 DSP" measures 210 Hz 

4.  My "300 DSP" measures 272 Hz.

5.  1 and 2 cascaded measure 157 Hz (i.e. close to the 150 Hz Joe
calculated).

6.  1 and 3 cascaded measure 175 Hz.

7.  1 and 4 cascaded measure 190 Hz.

Ed W0YK said:

"What can make a difference in large RTTY pileups is running the DSP down as
low as 200 Hz which rolls off the outer edges of the tones, but eliminates
enough of the pileup to sometimes be a net advantage."

Ed reported this using DSP = 200 and XFIL = 370 (his "250").  The BW of this
combination should effectively be the 187 Hz of the 200 DSP alone since the
370 Hz filter should contribute little attenuation in its center.  Using
Lyle's formula for DSP shape factor, that's ~487/187 or 2.6.  

If I select 1 and 4 above, that cascaded combination is 190 Hz BW with a
wider shape factor than what Ed used, since it's mainly the XFIL in effect
(Elecraft says SF = 4.0 for the 200).  
I would expect I could copy the same RTTY signal Ed does using the same
bandwidth and an even wider shape factor, and it might actually be better
for two reasons:

1.  It should be to tune because it's shape factor is wider (4.0 for my XFIL
alone versus 2.6 for Ed's 200 DSP alone).

2.  It has better close-in rejection between the 190 BW edges and the point
where the 200 XFIL crosses the 370 XFIL's skirts (which must be well over
400 Hz BW).

Incidentally, if my XFIL were 224 Hz as Elecraft measured, all of the above
BWs would be about 20 Hz wider (but I'm happy mine is 203 instead of 224!).

I'm not sure how critical RTTY tuning is but I believe this demonstrates
it's possible to fit a 170 Hz signal inside a 200 XFIL with <6 dB
attenuation.  It probably requires that you carefully center your 200 XFIL
and it may be on the hairy edge of copy (as I believe Ed said) but if RTTY
contests are like CW contests, that's what you sometimes need when you want
a really narrow filter.  I think the bottom line is that I might want both
the "200" and "250" if I were a serious RTTY contester, and I would use a
set point of 300 for toggling between the two. 

The latter reminds me...to set a wider DSP BW than the XFIL in my cascaded
measurements above, I simply changed my XFIL set point (to 250 or 300
instead of 200).  I normally have both my 500 and 200 filter set points a
little higher than their actual BW anyway (e.g. 600 and 250).  For the 500,
this keeps you in that filter when using DUAL PB (which automatically
selects whatever XFIL you have set for 600 Hz).

One final comment.  For those who think the 8-pole filters have zero
offsets, that may not be a good assumption, especially for the narrow
filters.  I was told by someone who has multiple 400 filters that they are
actually offset by ~100 Hz.  100 Hz is not a big issue for wider filters but
that's significant in a 435 or 370 Hz filter, so you might want to check
your narrow filter centers and tweak accordingly (0.01 increment in the menu
for each 10 Hz of offset).

73,  Bill

 
-- 
View this message in context: http://www.nabble.com/K3%3A-KRX3-question-tp18280281p18340909.html
Sent from the Elecraft mailing list archive at Nabble.com.