[Elecraft] K3 5 pole vs 8 pole filters - attenuation in the passband

[Joe Subich, W4TV]

The Elecraft/Inrad "250 Hz" filter is actually 370 Hz at -6dB where the
"400 Hz" filter is 435 Hz.  That's not enough to make a real difference
*as a roofing filter* since the primary selectivity of the K3/K3S is in
the DSP.

370 Hz is the minimum necessary bandwidth for a 170 Hz shift 45.45 baud
FSK signal ... any tighter than that and the loss of sideband energy
creates increased inter symbol interference (one bit gets "smeared" in
to the next).  The 400 Hz filter is really the one to have for RTTY -
even if the DSP is tightened below that on occasion.

It is a shame the Elecraft 200 Hz filters are no longer available -
they were superior overall for very tight CW (tighter than the 250 Hz,
8 pole filters all the way to -40 or - 50 dB) and provided a real
difference from the 400 Hz filters.  If one really *must* have the
250/370 Hz filter, the next filter to have would be the Elecraft/Inrad
1000/1060 Hz, the custom UnPCBS <http://www.unpcbs.com/> 700/785 Hz
filter, or perhaps the Inrad 500/550? Hz 8 pole filter (although the
difference between the "250" and "500" is rather small but not as
small as between the 250 and 400 nominal filters).

Remember, the purpose of the first IF ("roofing") filter is to protect
the If and Analog to Digital converter (ADC) from strong, adjacent
signals, *not* to provide the final selectivity as in the case of a
conventional superhetrodyne receiver.

73,

    ... Joe, W4TV


On 2/27/2016 11:21 AM, Jeff Stai wrote:
> On Sat, Feb 27, 2016 at 7:56 AM, lstavenhagen <lstavenhagen at hotmail.com>
> wrote:
>
>> I'm 99.9986% CW, so that filter is adequate for me but if you're a PSK
>> hound, for example, I could see the 250hz filter. For RTTY, I think the
>> 250hz filter might be borderline, though...
>>
>
> For RTTY the sweet spot is 400. At 250 you might lose information your
> decoders need to work. 73 jeff wk6i
>
>