Topics

Continuing problem, WSJT-X fails to decode


lmeeny
 

Hello,

I've been off the air for 5 days now with a problem vexing many members of the group, WSJT-X stops decoding after working properly for months on end.

The following conditions are true:

My computer clock is properly synced every 40 minutes. I've used two different NTP time sync programs and both check fine against time.is. When working properly WSJT-X displays dt values varying +/- 0.2 sec around 0.0 sec.

Signals are displayed on the waterfall and amplitude on the main window vertical tape display varies reflecting waterfall signal strengths. Waterfall signals can be saved to .wav files and listened to but not decoded when opened by using the file menu. Stored examples can be decoded.

Measured latency values cluster around 1000  usec, (is this too long?) with occasional peaks well above 5000 usec. but the times between spikes are much longer than, for instance, the duration of a FT4 or FT8 received transmission. On FT8 these spikes I believe are responsible for the occasional "0 decode" count. Repeated cycling between STOP and MONITOR or Powering off and restarting / rebooting the PC have no affect.

Repeatedly closing and opening WSJT-X has no affect. CPU usage, according to Task Manager, varies around 5%.

USB ports are not put to sleep.

The cause or causes appears to be independent of computer OS, hardware or WSJT-X version.  That this failure happens to so many people seems significant.  Is there a single cause or many? What can we do collectively to help fix this! Can the WSJT-X programmers add temporary hooks in the software to flag where the failure(s) occurs?

73,

Ed W2GHD


Bill Somerville
 

On 17/10/2020 20:22, lmeeny wrote:
Hello,

I've been off the air for 5 days now with a problem vexing many members of the group, WSJT-X stops decoding after working properly for months on end.

The following conditions are true:

My computer clock is properly synced every 40 minutes. I've used two different NTP time sync programs and both check fine against time.is. When working properly WSJT-X displays dt values varying +/- 0.2 sec around 0.0 sec.

Signals are displayed on the waterfall and amplitude on the main window vertical tape display varies reflecting waterfall signal strengths. Waterfall signals can be saved to .wav files and listened to but not decoded when opened by using the file menu. Stored examples can be decoded.

Measured latency values cluster around 1000  usec, (is this too long?) with occasional peaks well above 5000 usec. but the times between spikes are much longer than, for instance, the duration of a FT4 or FT8 received transmission. On FT8 these spikes I believe are responsible for the occasional "0 decode" count. Repeated cycling between STOP and MONITOR or Powering off and restarting / rebooting the PC have no affect.

Repeatedly closing and opening WSJT-X has no affect. CPU usage, according to Task Manager, varies around 5%.

USB ports are not put to sleep.

The cause or causes appears to be independent of computer OS, hardware or WSJT-X version.  That this failure happens to so many people seems significant.  Is there a single cause or many? What can we do collectively to help fix this! Can the WSJT-X programmers add temporary hooks in the software to flag where the failure(s) occurs?

73,

Ed W2GHD
Hi Ed,

if you check the "Menu->Save->Save all" option then pick a good example of a non-decode period and send us the saved .WAV file for analysis please?

73
Bill
G4WJS.


lmeeny
 

Bill,

Thank you, files have been sent to your email address as shown on QRZ.

Ed W2GHD


Russ Ravella
 

Hi Bill,

I have often had this problem too and have reported it. In my case, sometimes restarting WSJT-X does fix it though often only for a few minutes. I also don’t believe it has anything to do with my computer h/w (it’s now happened with three totally different computers) and I don’t believe it’s my radio (I’ve had five during the period I’ve had the problem) or how its configured or connected (again, there have been many cases). I would like to also send you my saved.WAV file but I can’t find a file named “saved.WAV” anywhere on my computer after checking Menu>Save>Save all” from which to pick a good non-decode period. Could you please elaborate just slightly ?

Thanks,
Russ KR6W

On Oct 17, 2020, at 12:25 PM, Bill Somerville <@g4wjs> wrote:

On 17/10/2020 20:22, lmeeny wrote:
Hello,

I've been off the air for 5 days now with a problem vexing many members of the group, WSJT-X stops decoding after working properly for months on end.

The following conditions are true:

My computer clock is properly synced every 40 minutes. I've used two different NTP time sync programs and both check fine against time.is. When working properly WSJT-X displays dt values varying +/- 0.2 sec around 0.0 sec.

Signals are displayed on the waterfall and amplitude on the main window vertical tape display varies reflecting waterfall signal strengths. Waterfall signals can be saved to .wav files and listened to but not decoded when opened by using the file menu. Stored examples can be decoded.

Measured latency values cluster around 1000 usec, (is this too long?) with occasional peaks well above 5000 usec. but the times between spikes are much longer than, for instance, the duration of a FT4 or FT8 received transmission. On FT8 these spikes I believe are responsible for the occasional "0 decode" count. Repeated cycling between STOP and MONITOR or Powering off and restarting / rebooting the PC have no affect.

Repeatedly closing and opening WSJT-X has no affect. CPU usage, according to Task Manager, varies around 5%.

USB ports are not put to sleep.

The cause or causes appears to be independent of computer OS, hardware or WSJT-X version. That this failure happens to so many people seems significant. Is there a single cause or many? What can we do collectively to help fix this! Can the WSJT-X programmers add temporary hooks in the software to flag where the failure(s) occurs?

73,

Ed W2GHD
Hi Ed,

if you check the "Menu->Save->Save all" option then pick a good example of a non-decode period and send us the saved .WAV file for analysis please?

73
Bill
G4WJS.




Bill Somerville
 

Hi Russ,

saved .WAV files go to the directory specified in "Settings->Audio". By default that is a directory called save under the WSJT-X log files directory ("Menu->File->Open log directory").

73
Bill
G4WJS.

On 17/10/2020 23:25, Russ Ravella via groups.io wrote:
Hi Bill,

I have often had this problem too and have reported it.  In my case, sometimes restarting WSJT-X does fix it though often only for a few minutes.  I also don’t believe it has anything to do with my computer h/w (it’s now happened with three totally different computers) and I don’t believe it’s my radio (I’ve had five during the period I’ve had the problem) or how its configured or connected (again, there have been many cases).  I would like to also send you my saved.WAV file but I can’t find a file named “saved.WAV” anywhere on my computer after checking Menu>Save>Save all” from which to pick a good non-decode period.  Could you please elaborate just slightly ?

Thanks,
Russ KR6W

On Oct 17, 2020, at 12:25 PM, Bill Somerville <g4wjs@...> wrote:

On 17/10/2020 20:22, lmeeny wrote:
Hello,

I've been off the air for 5 days now with a problem vexing many members of the group, WSJT-X stops decoding after working properly for months on end.

The following conditions are true:

My computer clock is properly synced every 40 minutes. I've used two different NTP time sync programs and both check fine against time.is. When working properly WSJT-X displays dt values varying ± 0.2 sec around 0.0 sec.

Signals are displayed on the waterfall and amplitude on the main window vertical tape display varies reflecting waterfall signal strengths. Waterfall signals can be saved to .wav files and listened to but not decoded when opened by using the file menu. Stored examples can be decoded.

Measured latency values cluster around 1000  usec, (is this too long?) with occasional peaks well above 5000 usec. but the times between spikes are much longer than, for instance, the duration of a FT4 or FT8 received transmission. On FT8 these spikes I believe are responsible for the occasional "0 decode" count. Repeated cycling between STOP and MONITOR or Powering off and restarting / rebooting the PC have no affect.

Repeatedly closing and opening WSJT-X has no affect. CPU usage, according to Task Manager, varies around 5%.

USB ports are not put to sleep.

The cause or causes appears to be independent of computer OS, hardware or WSJT-X version.  That this failure happens to so many people seems significant.  Is there a single cause or many? What can we do collectively to help fix this! Can the WSJT-X programmers add temporary hooks in the software to flag where the failure(s) occurs?

73,

Ed W2GHD
Hi Ed,

if you check the "Menu->Save->Save all" option then pick a good example of a non-decode period and send us the saved .WAV file for analysis please?

73
Bill
G4WJS.



Bruce N7XGR
 

Russ,  Here it is located,  C:\Users\YourName\AppData\Local\WSJT-X\save

Bruce  N7XGR


On Sat, Oct 17, 2020 at 6:25 PM Russ Ravella via groups.io <rus5=me.com@groups.io> wrote:
Hi Bill,

I have often had this problem too and have reported it.  In my case, sometimes restarting WSJT-X does fix it though often only for a few minutes.  I also don’t believe it has anything to do with my computer h/w (it’s now happened with three totally different computers) and I don’t believe it’s my radio (I’ve had five during the period I’ve had the problem) or how its configured or connected (again, there have been many cases).  I would like to also send you my saved.WAV file but I can’t find a file named “saved.WAV” anywhere on my computer after checking Menu>Save>Save all” from which to pick a good non-decode period.  Could you please elaborate just slightly ?

Thanks,
Russ KR6W

> On Oct 17, 2020, at 12:25 PM, Bill Somerville <g4wjs@...> wrote:
>
> On 17/10/2020 20:22, lmeeny wrote:
>> Hello,
>>
>> I've been off the air for 5 days now with a problem vexing many members of the group, WSJT-X stops decoding after working properly for months on end.
>>
>> The following conditions are true:
>>
>> My computer clock is properly synced every 40 minutes. I've used two different NTP time sync programs and both check fine against time.is. When working properly WSJT-X displays dt values varying +/- 0.2 sec around 0.0 sec.
>>
>> Signals are displayed on the waterfall and amplitude on the main window vertical tape display varies reflecting waterfall signal strengths. Waterfall signals can be saved to .wav files and listened to but not decoded when opened by using the file menu. Stored examples can be decoded.
>>
>> Measured latency values cluster around 1000  usec, (is this too long?) with occasional peaks well above 5000 usec. but the times between spikes are much longer than, for instance, the duration of a FT4 or FT8 received transmission. On FT8 these spikes I believe are responsible for the occasional "0 decode" count. Repeated cycling between STOP and MONITOR or Powering off and restarting / rebooting the PC have no affect.
>>
>> Repeatedly closing and opening WSJT-X has no affect. CPU usage, according to Task Manager, varies around 5%.
>>
>> USB ports are not put to sleep.
>>
>> The cause or causes appears to be independent of computer OS, hardware or WSJT-X version.  That this failure happens to so many people seems significant.  Is there a single cause or many? What can we do collectively to help fix this! Can the WSJT-X programmers add temporary hooks in the software to flag where the failure(s) occurs?
>>
>> 73,
>>
>> Ed W2GHD
>
> Hi Ed,
>
> if you check the "Menu->Save->Save all" option then pick a good example of a non-decode period and send us the saved .WAV file for analysis please?
>
> 73
> Bill
> G4WJS.
>
>
>
>





Russ Ravella
 

Thanks Bill and Bruce,

I see it now.  microsoft apparently decided to re-hide my system files during the last update and “AppData” wasn’t showing or being included in searches.  Fixed now.

I happened to actually be in the no-decode state when I saw Ed’s post and your response.  The program controls are still responsive when this happens but the waterfall and input meter are frozen and of course no decodes come in.  I’m running a Flex 6700 (SDR 3.1.12, Win10 2004) and DAX is responsive while WSJT-X decodes are frozen.  Restarting DAX anyway doesn’t help.

So I set “Save all” for a few progress bar cycles and checked but out of a zillion files there were none covering that period.  Obviously the file doesn’t get generated when decodes are frozen and “Save” was off (somehow?) when it happened this time.  I’ve stopped and relaunched WSJT-X, it’s un-frozen again, and I’ve got “Save All” re-enabled.  So when it happens again I’ll grab the latest file and send it to you.

Thanks for the help !
Russ KR6W


On Oct 17, 2020, at 3:31 PM, Bruce N7XGR <n7xgr1@...> wrote:

Russ,  Here it is located,  C:\Users\YourName\AppData\Local\WSJT-X\save

Bruce  N7XGR

On Sat, Oct 17, 2020 at 6:25 PM Russ Ravella via groups.io <rus5=me.com@groups.io> wrote:
Hi Bill,

I have often had this problem too and have reported it.  In my case, sometimes restarting WSJT-X does fix it though often only for a few minutes.  I also don’t believe it has anything to do with my computer h/w (it’s now happened with three totally different computers) and I don’t believe it’s my radio (I’ve had five during the period I’ve had the problem) or how its configured or connected (again, there have been many cases).  I would like to also send you my saved.WAV file but I can’t find a file named “saved.WAV” anywhere on my computer after checking Menu>Save>Save all” from which to pick a good non-decode period.  Could you please elaborate just slightly ?

Thanks,
Russ KR6W

> On Oct 17, 2020, at 12:25 PM, Bill Somerville <g4wjs@...> wrote:
> 
> On 17/10/2020 20:22, lmeeny wrote:
>> Hello,
>> 
>> I've been off the air for 5 days now with a problem vexing many members of the group, WSJT-X stops decoding after working properly for months on end.
>> 
>> The following conditions are true:
>> 
>> My computer clock is properly synced every 40 minutes. I've used two different NTP time sync programs and both check fine against time.is. When working properly WSJT-X displays dt values varying +/- 0.2 sec around 0.0 sec.
>> 
>> Signals are displayed on the waterfall and amplitude on the main window vertical tape display varies reflecting waterfall signal strengths. Waterfall signals can be saved to .wav files and listened to but not decoded when opened by using the file menu. Stored examples can be decoded.
>> 
>> Measured latency values cluster around 1000  usec, (is this too long?) with occasional peaks well above 5000 usec. but the times between spikes are much longer than, for instance, the duration of a FT4 or FT8 received transmission. On FT8 these spikes I believe are responsible for the occasional "0 decode" count. Repeated cycling between STOP and MONITOR or Powering off and restarting / rebooting the PC have no affect.
>> 
>> Repeatedly closing and opening WSJT-X has no affect. CPU usage, according to Task Manager, varies around 5%.
>> 
>> USB ports are not put to sleep.
>> 
>> The cause or causes appears to be independent of computer OS, hardware or WSJT-X version.  That this failure happens to so many people seems significant.  Is there a single cause or many? What can we do collectively to help fix this! Can the WSJT-X programmers add temporary hooks in the software to flag where the failure(s) occurs?
>> 
>> 73,
>> 
>> Ed W2GHD
> 
> Hi Ed,
> 
> if you check the "Menu->Save->Save all" option then pick a good example of a non-decode period and send us the saved .WAV file for analysis please?
> 
> 73
> Bill
> G4WJS.
> 
> 
> 
> 







Bill Somerville
 

On 17/10/2020 22:02, lmeeny wrote:
Bill,

Thank you, files have been sent to your email address as shown on QRZ.

Ed W2GHD
Hi Ed,

thanks for the files. I have tried to reply to your direct email but you ISP or mail server is refusing connections from my mail server. I will try again using a gmail account.

Executive summary is: it looks like your files are LSB recordings, WSJT-X requires audio from a USB Rx.

73
Bill
G4WJS.


lmeeny
 

Thanks to Bill Somervilles’, G4WJS,  insightful  and detailed analysis of the .wav files I sent him I’m back on the air. For reasons yet unknown the audio seen by the decoder was LSB where the various digital modes require USB.

How this happened is undetermined. My transceiver indicated it was in USB mode as I had been for a very long time never having been turned off. Selecting LSB mode the audio signal on the waterfall disappeared as is expected. Selecting USB mode the waterfall again was active … and decoding began, a puzzle.

I forgot the most basic troubleshooting technique, turn it off then on again. :-)

Thank you Bill for your patience.

Ed W2GHD