Hi Steve,
Thanks for finding and fixing the mysterious half-symbol offset!
-- Joe
On 11/26/2015 8:53 AM, Steven Franke wrote:
> Hi Joe,
>
> Sleep is a wonderful thing. I found the problem right away this morning. The
> first pulse generated by jt65sim (which is a sync pulse) *was* a runt. It was
> half as long as it should have been — essentially, the wave form was starting
> promptly at 1.0s into the file, but it was starting half-way through the
> first pulse. This explains all of the observations that I reported last
> night. I committed a fix for jt65sim and also a very minor tweak to fchisq65.
>
> There is still a residual 4 millisecond offset caused by the delay inserted
> by fil6521. I have not yet done anything to remove that because I’m not clear
> on how it would affect signal subtraction. It’s not at all important for
> display purposes, as we don’t display enough digits to reveal that small
> offset (though it is responsible for the perpetual “-“ sign in front of the
> displayed “0.0” when we process simulated files.
>
> Steve k9an
>
>> On Nov 25, 2015, at 7:40 PM, Steven Franke<[email protected]> wrote:
>>
>> Hi Joe,
>> OK, now I’m scratching my head.
>>
>> I was ready to just put in a half-symbol offset and call it a day, until I
>> made the mistake of looking into it in more detail…
>>
>> 1. the files generated by the windows JTSim program (the ones that we used
>> for early testing of sfrsd) give DT=0.0 (see the first decode, below)
>> whereas the files generated by the latest jt65sim are giving the 0.2s offset
>> that you pointed out.
>> <dts.tiff>
>>
>> 2. I used jt65sim to generate a file with +30dB snr and looked at it with
>> baudline and also did some plots. No doubt - it does start at 1.0s into the
>> file.
>>
>> But check out the attached pdf which is a plot of the coherently integrated
>> sync sequence (abs(z) from fchisq65) for the +30dB snr file. The overlayed
>> plots are from the multiple calls to fchisq65 as the algorithm peaks up f0
>> and dt.
>>
>> Note that the first (coherently integrated) sync pulse is a runt compared to
>> the rest. In particular, note that the isolated bits around x-index 275, and
>> 305 produce sharp peaks, but not the first isolated pulse. I told jt65sim to
>> generate multiple files, and they all show this phenomenon.
>>
>> Does this make any sense to you???
>>
>> Confused,
>> Steve k9an
>> <coherent_integration.pdf>------------------------------------------------------------------------------
>> Go from Idea to Many App Stores Faster with Intel(R) XDK
>> Give your users amazing mobile app experiences with Intel(R) XDK.
>> Use one codebase in this all-in-one HTML5 development environment.
>> Design, debug& build mobile apps& 2D/3D high-impact games for multiple OSs.
>> http://pubads.g.doubleclick.net/gampad/clk?id=254741551&iu=/4140_______________________________________________
>> wsjt-devel mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/wsjt-devel
>
>
> ------------------------------------------------------------------------------
> Go from Idea to Many App Stores Faster with Intel(R) XDK
> Give your users amazing mobile app experiences with Intel(R) XDK.
> Use one codebase in this all-in-one HTML5 development environment.
> Design, debug& build mobile apps& 2D/3D high-impact games for multiple OSs.
> http://pubads.g.doubleclick.net/gampad/clk?id=254741551&iu=/4140
> _______________________________________________
> wsjt-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/wsjt-devel
------------------------------------------------------------------------------
Go from Idea to Many App Stores Faster with Intel(R) XDK
Give your users amazing mobile app experiences with Intel(R) XDK.
Use one codebase in this all-in-one HTML5 development environment.
Design, debug & build mobile apps & 2D/3D high-impact games for multiple OSs.
http://pubads.g.doubleclick.net/gampad/clk?id=254741551&iu=/4140
_______________________________________________
wsjt-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
