[linrad] RE: Newbie's first impressions

[Leif Asbrink <leif.asbrink@xxxxxxxxxxxxxxxxxxxx]

Hello Joe,

Glad to see you here:-)

(first mail)
> I have a Delta 44 sound card, 
> but have not yet installed it.  I need to pay for the permanent 
> OSS license.
For narrowband you do not need OSS, the builtin driver
of RedHat 9 is usually working fine. Just run sndconfig.
If you are lucky the soundcard you have is supported:-)

As far as I know you need OSS to use the Delta44 at 96 kHz.

> One goal (which does not even require the wideband receiver) is 
> to build a JT65 decoder onto the back end of Linrad.
Interesting!

(second mail) 
> Should be pretty straightforward, I believe.  I would be strongly 
> inclined to keep the two programs largely independent.
That means you would have to add a second screen for the JT65
user interface. Linrad uses svgalib. You can move the windows to 
make room for a JT65 user interface, but I do not think you can
write on the free part of the screen from another program.
It would be trivial to send data to another program on a different
terminal on the same computer or on another computer (I think)
 
> For various reasons -- some of them good reasons, others dictated by my 
> vast ignorance of Windows internals --  WSJT operates in a sort of 
> "batch mode".  Its decoders expect to have available the digitized audio 
> for a full RX interval, all at once.  (These intervals are typically 30 
> seconds for FSK441, one minute for JT65.) I would probably have Linrad 
> write its audio output stream to a shared file, perhaps organized as a 
> ring buffer, with appropriate time tags on the data. 
There are several formats within Linrad. The audio output is not the
one to use. Not that there is anything wrong with it, but it
is computed by fractional resampling from a typically much lower 
sampling rate, it is then frequency shifted by the BFO frequency.
I would suggest the complex baseband signal I and Q. You could take
it from the output of the baseband filter, but since JT65 has its own
filters you might equally well take it from the input side where the
bandwidth is what you select with the "first mixer bandwidth reduction"

Linrad is designed to process many channels in parallel so you can
have several JT65 detectors running simultaneously. This is a feature
that is only halfway implemented because it is meaningless with 
loudspeaker output. It is intended for use with the CW decoding
routines AND with all other kinds of digital decoding that lead to
ascii on the screen. Right click for loudspeaker and ascii, left 
click for ascii only on a different frequency:-)


> The WSJT decoders would read the desired quantity of data from 
> the shared file and proceed pretty much as usual. 
Yes. And if they are fast, they can process many channels...

> Code writing will be made easy by the architecture of the Windows 
> version of WSJT.  The program's user interface and "real time" stuff 
> (such as T/R switching, audio input and output, etc.) is done in a front 
> end written in Visual Basic.
I think this section could be the difficult one. 
Visual Basic + svgalib = ????   (maybe #%$&)

> All of the significant computation 
> (waveform generation, spectral shaping, birdie excision, ping detection, 
> AFC, decoding, etc.) is done in a separate DLL coded mostly in Fortran, 
> plus a smattering of C.  The Fortran and C code is developed and tested 
> under Linux, anyway.  It should be quite easy to graft it onto the back 
> in of Linrad.
I totally agree. There is no reason to expect any problems at all on
"the essentials"



73

Leif / SM5BSZ




LINRADDARNIL