Post-HSMS QSO Analyses
by VE5UF

Date: Sun, 21 Sep 1997 17:23:32 -0600
To: hsms@tree.net
From: Doug, VE5UF 
Subject: [HSMS] VE5UF<>K0SM sked result and Questions


Just a note on this morning's 12:30-13:30Z sked with K0SM and a few
questions raised by the results...
Pings were more plentiful than our previous skeds, but were not strong.
On this end, got calls in the 13th minute and S2 from Andy in the 19th
minute.  But that's all we exchanged - no Rs.  Better luck next time, 
as they say.  No ping was stronger than the noise which today here was 
a combination of powerline QRN and general neighborhood crud.

To that end, I did a bit of post-sked playing with CoolEdit, applying
various filter bandwidths to the 8 ping files I saved during the session.
None of the pings is visible in the noise when the raw  WAV file is viewed
as a waveform.  A spectral analysis (by CoolEdit) reveals a small 
spike at 2760 Hz - the recovered audio ping frequency. (more about
this later)  I spec'd an audio filter with a passband of 2500-3000 Hz
and having an ideal square response, 15 dB of bandstop rejection, and
ran the ping files through the filter twice.   Major difference!!!!
Now the pings are readily visible in the waveform noise, and MS_DSP 
decodes them cleanly without skipping.  In fact, the pings are clean
enough to be visually decodable in waveform view if the file is 
stretched a bit so that you can zoom in close enough with the Zoom 
control in CoolEdit.

Which leads to question #1:
Is anyone using their 700Hz (or narrower) CW filter during HSMS work??

I don't know what the real response of this filter which CoolEdit
created might be, but from the spectrum analysis of the result, it
looks pretty good and the stopband rejection is there as promised.
Today's pings would have been quite useable with as little as a 3 or 4 dB
improvement in SNR and I would expect to see at least that much going
from the SSB filter to the CW filter in the IF receiver.

Thus question #2:
Is a 700Hz Rx passband simply too narrow to use when blindly presetting
the Rcvr dial considering the possibilities for Rx/Tx calibration
errors on both ends?

Two weeks ago, I carefully (I thought) measured the Rx dial accuracy of
my IC271 and concluded it was within 200 Hz of the display and thus not
worth worrying about.  Analysis of pings from K0GU last week shows them to
be at 2140 Hz, plenty close enough to expectations. After finding today's 
pings at 2760 Hz when I was expecting 2000 Hz, I got concerned and ran the 
dial calibration test again, and now it's off by 600 Hz.  FWIW, the 271 
is ON 24hrs/day, is in a fairly constant temp room, and seems relatively 
unaffected by line voltage variations.  No explanation for this - either I 
was wrong the first time, or it truly has drifted in two weeks.

This episode just underscores last week's discussion here about sked
frequencies - but let's not open that again, for now.  However, it's
obvious from the numbers above that a normally inconsequential 200 Hz
(or so) calibration error on K0SM's end means I hear *no* pings.

Any thoughts on filter bandwidths?  
I haven't tried it yet, but do noise blankers *always* eat pings?
Anyone ever tweaked the synthesizer in an older 271 (not an 'A' or 'H')??
My 271 manual is in Japanese.  :(

So many questions, so little time :)
--
TIA Doug VE5UF






 Return 



Comments:  Rein, W6/PA0ZN 



 Top Page