Date
1 - 2 of 2
jimlill:8088
Jim Lill
Some of you may have noticed that I stopped ranking on my http://jimlill.com:8088 page. I did it to allow easier/riskfree changes to the scripts that run that. With the proliferation of false decodes, I found that the data integrity had gotten bad enough so as to adversely impact the ranks of the top stations. As that system runs real time. It is hard for my small sbc to keep up with both call validation and ranking. However, progress on the revamp is proceeding. Until it is done the ranking will not be available.
At the same time, I am running tests to determine which values of -C and -o for the wsprd decoder yield the best results. This is being done by taking the wav files the 20m band and keeping them aside. Those files can then be tested with a variety of cli parameters and evaluated. So far -C 5000 -o4 looks most promising and is consistent with advice Steve Franke made a few years ago, -C 10000 yields more false decodes.
Those decodes obviously add to a reporters Spots count, albeit in a small percentage of the total. Less obvious is that every false decode generates a false unique, 1 for 1, and so it is much more impactful on the data integrity.
-Jim
At the same time, I am running tests to determine which values of -C and -o for the wsprd decoder yield the best results. This is being done by taking the wav files the 20m band and keeping them aside. Those files can then be tested with a variety of cli parameters and evaluated. So far -C 5000 -o4 looks most promising and is consistent with advice Steve Franke made a few years ago, -C 10000 yields more false decodes.
Those decodes obviously add to a reporters Spots count, albeit in a small percentage of the total. Less obvious is that every false decode generates a false unique, 1 for 1, and so it is much more impactful on the data integrity.
-Jim
Rob Robinett
Hi Jim,
Thanks for your efforts on this topic. I have already changed -C 10000 at KFS to -C 5000. Other sites I manage were already running at 5000 or less.
Rob
On Thu, Aug 11, 2022 at 1:21 PM Jim Lill <jim@...> wrote:
Some of you may have noticed that I stopped ranking on my
http://jimlill.com:8088 page. I did it to allow easier/riskfree changes
to the scripts that run that. With the proliferation of false decodes, I
found that the data integrity had gotten bad enough so as to adversely
impact the ranks of the top stations. As that system runs real time. It
is hard for my small sbc to keep up with both call validation and
ranking. However, progress on the revamp is proceeding. Until it is
done the ranking will not be available.
At the same time, I am running tests to determine which values of -C and
-o for the wsprd decoder yield the best results. This is being done by
taking the wav files the 20m band and keeping them aside. Those files
can then be tested with a variety of cli parameters and evaluated. So
far -C 5000 -o4 looks most promising and is consistent with advice Steve
Franke made a few years ago, -C 10000 yields more false decodes.
Those decodes obviously add to a reporters Spots count, albeit in a
small percentage of the total. Less obvious is that every false decode
generates a false unique, 1 for 1, and so it is much more impactful on
the data integrity.
-Jim
--