I've had Skimmer running for a little over a month but last
night was the first time I've tried it in serious contest conditions.
My setup is a Beverage > Elecraft K3 > LP-PAN beta > M-Audio
Audiophile 2496 > 1.1 Ghz P3 > Skimmer 1.1. This setup is definitely
CPU-limited but works OK up to ~100 decoders running at a 48 kHz
sample rate (48 kHz bandwidth).
From 0145-0215 I decided to try an SWL test on 160. The band
wasn't too noisy and I used one 1000' Beverage toward EU with no
preamp for the entire test. During this period Skimmer positively
decoded (in bold) 4 calls (US & VE only) plus a few decoding errors
(e.g. both E3OSZ and VE3OSZ).. Tuning the band over the same period,
I copied those 4 plus 15 others (13 EU plus 3V8BB). Skimmer did list
a couple of the latter but not in bold. I believe that means it
copied local CQ-ers answering DX callers. All of the DX signals were
near the noise floor and the locals were well above it, but some DX
stations were quite readable (e.g. 3V8BB was the strongest and was
flitting around answering CQs). CPU utilization was well below 100%
the entire time due to the few active decoders.
Although I was disappointed with the number of calls actually
decoded, Skimmer's waterfall did a nice job of locating nearly every
weak DX signal. When Skimmer's decoders turned yellow, I could click
on them (sending the K3 to that frequency) and then copy the calls by
ear. This was a very efficient way to scan the band...probably
similar to what VHF contesters do with SDR waterfalls. My conclusion
is that while Skimmer will be a very useful tool on bands where
signals are a little stronger, it still cannot decode weak signals
near the noise floor. I'm sure Alex will be improving his algorithms
over time and of course we'll have more powerful computers. But even
as it now stands, Skimmer's waterfall can be quite useful when
scanning the band for weak signals. I suppose if you could put
Skimmer into a waterfall-only mode (i.e. no decoding) it would not
qualify as "assistance" any more than a band scope or SDR waterfall.
One more point especially important for 160...Skimmer has problems in
QSB. For example we would probably all recognize G*SXW as G3SXW, but
Skimmer cannot do this (yet). I suspect as Alex adds SCP capability
we may see it decode the above as G3SXW? in the future.
I also listened briefly on 40m but Skimmer had my CPU in heavy
overload due to many signals spaced about every 250 Hz. If there was
ever a case for the K3's 200 Hz roofing filter this was it! Even with
my CPU in overload, many calls were decoded rather quickly. However I
was surprised in listening to E77DX, who had a very good signal, that
Skimmer never decoded his call. Perhaps that could be due to the
current software not recognizing E7 as a valid prefix.
This was a fun experiment and hopefully of interest to some.
73, Bill W4ZV
_______________________________________________
CQ-Contest mailing list
CQ-Contest@contesting.com
http://lists.contesting.com/mailman/listinfo/cq-contest
|