1. Topics
  2. WD FST4W decoding does not affect WSPR decoding

WD FST4W decoding does not affect WSPR decoding

Rob Robinett
 

Thanks to Tom WA2TP I have been running an experiment which compares two sets of 20M signals from the same 5 Kiwis at his site.

WA2TP     is Tom's production i9 server running WD 3.0.3.1 configured to decode WSPR-2 and most FST4W modes:  "MERGE_K123456,40,W2:F2:F5:F15"
WA2TP-2  is Tom's test            i5 server running WD 3.0.3.1 configured to decode only WSPR-2:                                    "MERGE_K123456,20"

As you might be able to see from the fuzzy chart below, WA2TP recorded 4308 spots over that period while WA2TP-2 recorded 4156 spots. So adding the FST4W modes did not degrade WSPR-2 decoding.
Some of the additional WA2TP spots many be FST4W-300 spots being transmitted by N6GN, but most of the difference may be due to WA2TP decoding with WSPRD_CMD_FLAGS="-C 10000 -o 4 -d"
while I mistakenly left WA2TP-2 set to WSPRD_CMD_FLAGS="-C 500 -o 2 -d".
I have just changed those settings on WA2TP-2 to match WA2TP, and I expect the next 12 hours to show an even closer correlation in spot counts between them.
You can watch the progress of that test at:  http://wspr.rocks/head2head/

So those of you with enough CPU power and RAM to support FST4W decoding should feel confident that adding FST4W to your configurations will not degrade your WSPR-2 decoding performance.
Check out Arne's WSPR Mode Watch' Grafana page at https://wspr.live/gui/d/vhwRxD67z/wspr-mode-watch?orgId=1&refresh=1m to see your and other WD sites FST4W activity.

Ongoing studies by Glenn N6GN and Gwynn G3ZIL suggest that for reasons yet to be determined even the shortest FST4W-120 transmissions on HF bands may not be as effective as WSPR-2.
If our studies lead us to learn that the implementation of FST4W decoding, which differs slightly from that of WSJT-x, is the source of the impaired FST4W performance, then of course fixes will be put into WD.

73,

Rob
John
 

Well ! That is a disappointment for my station. I need to study the non FST4W hits.

John
TI4JWC

El viernes, 8 de julio de 2022, 01:51:13 p. m. CST, Rob Robinett <rob@...> escribió:


Thanks to Tom WA2TP I have been running an experiment which compares two sets of 20M signals from the same 5 Kiwis at his site.

WA2TP     is Tom's production i9 server running WD 3.0.3.1 configured to decode WSPR-2 and most FST4W modes:  "MERGE_K123456,40,W2:F2:F5:F15"
WA2TP-2  is Tom's test            i5 server running WD 3.0.3.1 configured to decode only WSPR-2:                                    "MERGE_K123456,20"

As you might be able to see from the fuzzy chart below, WA2TP recorded 4308 spots over that period while WA2TP-2 recorded 4156 spots. So adding the FST4W modes did not degrade WSPR-2 decoding.
Some of the additional WA2TP spots many be FST4W-300 spots being transmitted by N6GN, but most of the difference may be due to WA2TP decoding with WSPRD_CMD_FLAGS="-C 10000 -o 4 -d"
while I mistakenly left WA2TP-2 set to WSPRD_CMD_FLAGS="-C 500 -o 2 -d".
I have just changed those settings on WA2TP-2 to match WA2TP, and I expect the next 12 hours to show an even closer correlation in spot counts between them.
You can watch the progress of that test at:  http://wspr.rocks/head2head/

So those of you with enough CPU power and RAM to support FST4W decoding should feel confident that adding FST4W to your configurations will not degrade your WSPR-2 decoding performance.
Check out Arne's WSPR Mode Watch' Grafana page at https://wspr.live/gui/d/vhwRxD67z/wspr-mode-watch?orgId=1&refresh=1m to see your and other WD sites FST4W activity.

Ongoing studies by Glenn N6GN and Gwynn G3ZIL suggest that for reasons yet to be determined even the shortest FST4W-120 transmissions on HF bands may not be as effective as WSPR-2.
If our studies lead us to learn that the implementation of FST4W decoding, which differs slightly from that of WSJT-x, is the source of the impaired FST4W performance, then of course fixes will be put into WD.

73,

Rob
Glenn Elmore
 

I'm not prepared to issue an early assessment of WSPR v FST4W. I think that would be premature. There are simply WAY too many variables to make any kind of clear and definitive statement yet - even if we thought we knew.

Until this gets sorted out, I suggest not jumping to any kind of general conclusions.


Glenn n6gn


On 7/8/22 14:29, John via groups.io wrote:

Well ! That is a disappointment for my station. I need to study the non FST4W hits.

John
TI4JWC

El viernes, 8 de julio de 2022, 01:51:13 p. m. CST, Rob Robinett <rob@...> escribió:


Thanks to Tom WA2TP I have been running an experiment which compares two sets of 20M signals from the same 5 Kiwis at his site.

WA2TP     is Tom's production i9 server running WD 3.0.3.1 configured to decode WSPR-2 and most FST4W modes:  "MERGE_K123456,40,W2:F2:F5:F15"
WA2TP-2  is Tom's test            i5 server running WD 3.0.3.1 configured to decode only WSPR-2:                                    "MERGE_K123456,20"

As you might be able to see from the fuzzy chart below, WA2TP recorded 4308 spots over that period while WA2TP-2 recorded 4156 spots. So adding the FST4W modes did not degrade WSPR-2 decoding.
Some of the additional WA2TP spots many be FST4W-300 spots being transmitted by N6GN, but most of the difference may be due to WA2TP decoding with WSPRD_CMD_FLAGS="-C 10000 -o 4 -d"
while I mistakenly left WA2TP-2 set to WSPRD_CMD_FLAGS="-C 500 -o 2 -d".
I have just changed those settings on WA2TP-2 to match WA2TP, and I expect the next 12 hours to show an even closer correlation in spot counts between them.
You can watch the progress of that test at:  http://wspr.rocks/head2head/

So those of you with enough CPU power and RAM to support FST4W decoding should feel confident that adding FST4W to your configurations will not degrade your WSPR-2 decoding performance.
Check out Arne's WSPR Mode Watch' Grafana page at https://wspr.live/gui/d/vhwRxD67z/wspr-mode-watch?orgId=1&refresh=1m to see your and other WD sites FST4W activity.

Ongoing studies by Glenn N6GN and Gwynn G3ZIL suggest that for reasons yet to be determined even the shortest FST4W-120 transmissions on HF bands may not be as effective as WSPR-2.
If our studies lead us to learn that the implementation of FST4W decoding, which differs slightly from that of WSJT-x, is the source of the impaired FST4W performance, then of course fixes will be put into WD.

73,

Rob
WA2TP - Tom
 

I agree. I am seeing a very significant degradation in my 20m spotting today and I have changed nothing other than the WD changes made yesterday in adding the test system.

I am unfortunately away from the system to evaluate if there is a new RFI source. 

I won’t be able to asses until I return next Wednesday.

On Jul 8, 2022, at 5:24 PM, Glenn Elmore <n6gn@...> wrote:



I'm not prepared to issue an early assessment of WSPR v FST4W. I think that would be premature. There are simply WAY too many variables to make any kind of clear and definitive statement yet - even if we thought we knew.

Until this gets sorted out, I suggest not jumping to any kind of general conclusions.


Glenn n6gn


On 7/8/22 14:29, John via groups.io wrote:
Well ! That is a disappointment for my station. I need to study the non FST4W hits.

John
TI4JWC

El viernes, 8 de julio de 2022, 01:51:13 p. m. CST, Rob Robinett <rob@...> escribió:


Thanks to Tom WA2TP I have been running an experiment which compares two sets of 20M signals from the same 5 Kiwis at his site.

WA2TP     is Tom's production i9 server running WD 3.0.3.1 configured to decode WSPR-2 and most FST4W modes:  "MERGE_K123456,40,W2:F2:F5:F15"
WA2TP-2  is Tom's test            i5 server running WD 3.0.3.1 configured to decode only WSPR-2:                                    "MERGE_K123456,20"

As you might be able to see from the fuzzy chart below, WA2TP recorded 4308 spots over that period while WA2TP-2 recorded 4156 spots. So adding the FST4W modes did not degrade WSPR-2 decoding.
Some of the additional WA2TP spots many be FST4W-300 spots being transmitted by N6GN, but most of the difference may be due to WA2TP decoding with WSPRD_CMD_FLAGS="-C 10000 -o 4 -d"
while I mistakenly left WA2TP-2 set to WSPRD_CMD_FLAGS="-C 500 -o 2 -d".
I have just changed those settings on WA2TP-2 to match WA2TP, and I expect the next 12 hours to show an even closer correlation in spot counts between them.
You can watch the progress of that test at:  http://wspr.rocks/head2head/

So those of you with enough CPU power and RAM to support FST4W decoding should feel confident that adding FST4W to your configurations will not degrade your WSPR-2 decoding performance.
Check out Arne's WSPR Mode Watch' Grafana page at https://wspr.live/gui/d/vhwRxD67z/wspr-mode-watch?orgId=1&refresh=1m to see your and other WD sites FST4W activity.

Ongoing studies by Glenn N6GN and Gwynn G3ZIL suggest that for reasons yet to be determined even the shortest FST4W-120 transmissions on HF bands may not be as effective as WSPR-2.
If our studies lead us to learn that the implementation of FST4W decoding, which differs slightly from that of WSJT-x, is the source of the impaired FST4W performance, then of course fixes will be put into WD.

73,

Rob
1 - 4 of 4