[Laser] Sound card stability and DSP

[Andrew T. Flowers, K0SM]

Hi folks,

I'm writing a program  for troposcatter communication.  Does anyone 
know, or can anyone find out, what kind of frequency stability and 
frequency accuracy you get with a typical soundcard?  Right now the 
program uses the frequencies between 20 and 32 Hz and bins with a width 
of 122 mHz (that is millihertz, not MHz).  This sounds perfectly 
reasonable, but I imagine there is quite a bit of variance between 
cards.  I got my laptop and my home computer to talk to each other today 
through the audio cables--needless to say, S/N was not a problem!  It 
turns out that they didn't agree on frequency, which is what prompts my 
question.  Does anyone know if separate oscillators are used for 
recording and playback in some cards?  One of the machines didn't agree 
with itself.  I don't rule out programming errors, but with the exact 
same software on both machines this was my only explaination.  Can 
anyone help?

Since I'm sure there is going to be lots of questions about this, I'll 
try to prempt a few right now :-)  After reading through K3PGP's 
experiments and doing some of my own, I figured someone ought to go 
ahead and write  a piece of software that could be used for real 
communication.  It's clear that troposcatter is possible if we just make 
the right tool to do it.  This thing works pretty much like PUA43 at 
much lower frequencies.  Right now it uses an 8.2 second transmission 
length per character and incoherent averaging of the message as it loops 
around.  It seems reasonable to move this to 47.5 seconds (I think 2^19 
@ 11025Hz gets me there) for the really long stuff *if* a soundcard can 
deal with frequency bins that small.  It runs in full duplex, TX/RX in 
different ranges so your computer doesn't talk to itself through the 
poor isolation in the card.  Its written in Java (thats right, hardcore 
number crunching in Java), though some of the hardware stuff may be 
windoze specific at the moment.   That can be fixed.  I can answer 
questions beyond this, but understand that my knowledge of mathematics 
reaches back to a smattering of linear algebra 6 years ago :-)  I would 
be interested in hearing from anyone who knows something about signal 
averaging techniques, especially using the complex half of the FFT which 
is just wasting memory right now....

Andy K0SM/2