[Elecraft] K3: roofing filter configuration settings

[Ed Muns]

"Configuration bandwidth" is the DSP bandwidth at which the given crystal
filter engages in the K3/K3S.  This crystal filter remains engaged as the
DSP bandwidth is narrowed, to the point where another crystal filter's
"configuration bandwidth" is reached.

In choosing the configuration bandwidth, one consideration is the cascade
effect of the crystal and DSP filter bandwidths.  In general, the cascade
bandwidth will be less than either of these two filter bandwidths.  The
extent to which this is true depends on how close the two filter bandwidths
are to one another.

This reduced cascade bandwidth is the underlying reason for the KFL3A-250
crystal filter from INRAD being called "250 Hz" but actually measuring about
370 Hz at -6 dB and about 310 Hz at -3 dB.  (See
http://www.elecraft.com/K3/K3_filter_plots.htm.)  This filter was originally
designed for the Yaesu first IF and intended to be used in conjunction with
another "250 Hz" filter in the second IF.  Each filter is actually wider
than its marketing name because when the two are cascaded, the resulting
bandwidth is about 250 Hz.

The cascade effect of the K3 crystal and DSP IF filters is less than the
Yaesu and Kenwood cases where both filters were crystal filters, each with a
somewhat "round" shape.  The K3 DSP filter is more "square" in the
transition between passband and stopband.

Since the K3 DSP bandwidth is continuously variable, it becomes more
important to consider this cascade effect because it may or may not be
significant depending on how close the crystal and DSP filter bandwidths are
to one another.  It is also more important with the K3 to be aware of the
actual crystal filter bandwidths.

For RTTY operation, the minimum cascaded bandwidth should be about 400 Hz.
This is because modern software decoders, e.g., MMTTY and 2Tone, use
algorithms that make use of the sidebands of each tone in order to minimize
error rate.  Using narrower IF filters, including the DTF, in the radio
attenuates these useful sidebands.

For the same reason, the K3 (and, Icom) dual-tone filter removes sideband
information that could be used by the decoder to improve copy.  Anecdotally,
in pileup and QRM situations such as DXpeditions and contests, I've found a
cascaded IF bandwidth of 400-500 Hz to be superior to the narrower
bandwidths, including the DTF, that I used several years ago.

Accordingly, I've chosen the INRAD 500 Hz 8-pole filter for CW and RTTY,
setting my DSP bandwidth to 400-500 Hz as desired.  I think this gives
marginally better, and more versatile, filtering than the KFL3A-250.

For very heavy QRM CW situations, e.g., 160 meter contests, the 5-pole 200
Hz crystal filter is useful.

Ed W0YK
____________________________________________________________________________
___

Dick K9OM wrote:

A question regarding roofing filter "configuration bandwidth"  settings on
the K3:
 
The factory default roofing filter "configuration bandwidth" setting  is to
set the 250hz filter at a setting of "250".  Therefore, the  250hz roofing
filter will engage at the same time as the 250hz DSP filter. 
However, since the 250hz roofing filter has an actual BW6 of 370hz,  aren't
there times when it would be more beneficial to have  the "configuration
bandwidth" of the 250hz roofing filter set to 350 or  400?  
I'm thinking this may be especially true when operating RTTY since the
average RTTY bandwidth is aprx. 370hz.  Perhaps there are times when  this
would be beneficial when operating other modes as well?
 
Likewise, the 400hz roofer has a BW6 of 450hz, so perhaps for certain modes
a "configuration bandwidth" setting of 450 rather than 400 could be
beneficial as well?
 
If anyone is wondering where I'm getting the term "configuration bandwidth" 
 from, it's a configuration heading on the K3 Utility programs filter
configuration page.