[Laser] Re: PSK31 via light - up-convert to RF vs audio amp

[TWOSIG at aol.com]

In a message dated 8/1/2006 9:05:29 PM Central Standard Time,  
toasty256 at yahoo.com writes:
 

>I bet you wouldn't need any further amplification of a
>PGP  preamp output in the demo version at least. It
>seems to have plenty of  output to feed a soundcard
>directly. 
>The one transistor photo-RF  converter you described
>sounds like the light would modulate the  oscillator
>kind of weakly, although maybe it would still have  a
>usable signal. This would be an AM modulated signal to
>the radio  i'm guessing.

My intent was AM, though in a practical circuit the mixer might load  the 
oscillator and pull its frequency.

>Perhaps there is a way to add  an oscillator/mixer to
>the PGP RX without having to modify it too much.  So
>you could have a 3.58 mhz RF output in addition to the
>baseband  output. Then you could compare the two
>signals conveniently. 

I do not see much point in using using a PGP pre-amp to modulate the  3.58 
MHz carrier.  Any noise contributed by the pre-amp would be in the  modulated 
signal.  The intent of the suggestion to modulate directly  from the photo-diode 
would be to see if would contribute less noise (in the  desired signal) which 
could be amplified by a radio receiver (which again must  not contribute to 
noise in the signal).  I thought the suggestion would be  an RF amplifier whose 
gain would be controled by the photodiode, resulting in  modulation of the 
carrier.  Perhaps a FET would perform better, that is  produce a useable 
modulated RF signal with less noise.  If the overall  noise is less, then it is worth 
the complexity.  

This does suggest  that there might be a better chance of improvement by 
using the photo diode as  part of the frequency controling elements of the 
oscillator, then using a  receiver as an FM detector.  

At first blush adding complexity would  seem to add noise instead of reduce 
it.  In theory a crystal set followed  by a low noise audio amplifier should 
work better than a superhetrodyne  receiver.  And maybe when if you apply the 
concept to our light  communication experiments, then maybe the theory is right. 
 However,  superheterodyne receivers are better for the radio stations that I 
listen to  most of the time.  (  



>PSK seems to be a lot more  rugged than the ideal
>situation when it's a pure sinewave-ish tone. I  think
>i read the loss in signal gain or whathaveyou is only
>2 or 3  dB?? due to it not being perfectly formed (the
>raw digital output of that  OR gate) I've seen mildly
>over-modulated PSK signals on the ham bands  that were
>still readable just fine. Granted, they were  probably
>not raw squarewaves. But being an FM mode inherently
>at  least in part, i don't see why a PSK signal would
>suffer much damage from  being fully 'limited' or
>squared up. I assume that 2 or 3 dB is lost  forever,
>but on the receive end, the signal can be cleaned up  a
>little with a narrow filter before being sent to the
>computer's  sound card if needed. I'm not sure how much
>this does in restoring the  original PSK waveform, but
>it would do some good i bet.

A squared  PSK signal uses a lot more bandwidth, so it interfers with other 
users on  HF.  On a tight light beam you do not have other users to interfer  
with.  The difference in energy to transmit is area of the square wave  after 
you subtract the area of the sine wave.  Most of that energy is  wasted in the 
linear amplifier circuit anyway ( At least that was what I  understood, 
waaaaaaayyyyyyyyy back when I was trying to learn some of this  stuff.  Back then 
transistor radios were not yet available. )  Once  the signal is transmitted ( 
in the ether so to speak ) the ruggedness of PSK,  and I think it is just a 
signal to noise improvement, comes from the very narrow  frequency window needed 
to decode the signals.  The sound card is acting  like a receiver with a 
narrow IF bandwidth, which blocks much of the  noise.
 
Decoding a glitchless square wave that has phase shifting, either BPSK or  
QPSK,  is much easier than decoding a sine wave.  Any rising or  falling edge of 
the square wave can be used to decode the signal.  A sine  wave signal can 
only use the rising or falling edges that are above a minimum  threshold, and 
there is some ambiguity in the time domain where the edge should  be.  An 
alternate decoding scheme might use the carrier sine peaks, which  in theory fall 
half way between the rising and falling crossovers.  Still  there is an analog 
susceptablility to noise which can skew the time.
 
I would like to find someone with enough computer savvy to modify some PSK  
software to produce glitchless digital output instead of sound card audio, or  
maybe both simultaneously.  Then over a light channel it would be possible  to 
see if one has a significant benefit with different types of noise.   After 
that, it might be worth trying to tweak the receiver software to try to  use 
some of the harmonic content of the square wave to improve the  performance.
 
James
N5GUI