[Elecraft] K3 roofing filter IMD

[Bill Tippett]

K6LL:

 >I was perusing the Sherwood Engineering receiver evaluation
data, 
<http://www.sherweng.com/table.html>http://www.sherweng.com/table.html 
, and I noticed
that the K3 2 KHz dynamic range is reported to be
significantly better with the 200 Hz 5-pole roofing filter
than with the 400 or 500 Hz filter.

 >Then I came across some IMD data on the K3 Wiki,
<http://www.zerobeat.net/mediawiki/index.php/K3_Roofing_Filters>http://www.zerobeat.net/mediawiki/index.php/K3_Roofing_Filters 
,
which does not show much difference between those filters,
but has an interesting footnote:
"It should be mentioned that in a published review of the
K3, G4AON observed degradation of close-spaced IMD
measurements with the 400 Hz, 8 pole filter: "These figures
are for a 400 Hz bandwidth with the 8 pole 400 Hz roofing
filter, the rather surprising discovery was the dynamic
range improved by almost 10 dB when the 2.8 KHz 8 pole
filter was selected." G4AON's findings haven't yet been
independently
confirmed however."

 >Any thoughts on resolving these apparent inconsistencies?

         My guess is G4AON had a measurement problem when
he originally published that.  The current version of Dave's
review has no such comment anywhere to be found:

"Two tone dynamic range testing was only possible on 14 MHz as I only 
have one signal generator, the other being a well buffered 14 MHz 
fixed crystal oscillator based on the design for dynamic range 
testing from the book "Solid state design for the radio amateur". 
Both these were combined in a hybrid coupler and fed via a variable 
attenuator to the K3. The factory figures give a 100 dB dynamic range 
at 5 KHz spacing and 95 dB for a 2 KHz spacing, both using a 400 Hz 
(8 pole) filter. My measurements give a two tone dynamic range at 2 
KHz signal spacing of 100dB with the pre-amp off. These figures are 
for a 400 Hz bandwidth with the 8 pole 400 Hz roofing filter, similar 
high dynamic range figures exceeding 100 dB at close signal spacing 
were also obtained by the ARRL (review in April 2008 QST), two other 
amateurs and also by Rob Sherwood the well known receiver tester, 
these tests were independent of each other and on different K3s. "

http://www.astromag.co.uk/k3/

         It should have been a clue that there was a measurement
problem when better performance was obtained with a wider
filter...that simply doesn't make sense.  ARRL has published
a few strange results (like better performance with preamp on
versus preamp off) for other products so I think everyone makes
a few measurement or data recording errors at times.  It's also
important to remember that everything we read on the Internet
is not always true!

         I believe the results from Sherwood, ARRL and Elecraft
have all been consistent so far (with the exception that Sherwood
uses a classical technique of measuring IMD which may result in
"phase noise limited" measurements).  ARRL and Elecraft use a
narrow bandwidth spectrum analyzer (which eliminates phase noise
effects) which makes their results look better that is actually
achievable in practice (i.e. phase noise masks IMD performance).

                                         73,  Bill  W4ZV