[RTTY] BPSK125

[Kok Chen]

On Feb 1, 2009, at 2/1    8:34 AM, Ed wrote:

> As does fldigi....your choice Linux-Mac-MS-BSD.

BPSK125 is actually almost a no brainer to implement.

As background; PSK31 is really "PSK31.25"  The 31.25 (baud rate) is  
simply rounded down.

When Skip KH6TY experimented with higher speed PSK, he simply doubled  
the sampling rate of everything (in both receive and transmit  
directions) to get "PSK62.5".  Skip rounded the "62.5" up to "63" and  
the "new" mode became known as PSK63.  Just like PSK125, there was a  
sudden spike of interest and people then settled back to PSK31  
because 31 uses less spectrum and gets through when PSK63 could not.

That was similar to the spike of interest for 75 baud RTTY (remember  
the HS RTTY Contest?) and 23 Hz shift RTTY  (MSK, actually, not FSK)  
-- we all go through those phases... 170 Hz shift is a good match for  
HF selective fading (23 Hz shift is too narrow to take advantage of  
having two FSK tones during selective fades).

There is really almost no new code to write to implement PSK63 except  
to shove the data samples into the demodulator at twice the speed as  
before,  If the demodulator was preceded by a decimation stage, all  
that is needed is to change the decimation ratio by a factor of two  
and you have PSK63 demodulation.  At least this is how I did it in  
cocoaModem.  You also need to change the bandwidth of the filter that  
precedes the demodulator, if there is one.

The bit and tone frequencies in cocoaModem's transmission ("AFSK")  
are generated by using what are called DDA (digital differential  
analyzer -- the digital version of Vannervar Bush's differential  
analyzer that he invented during WWII to guide projectiles)

http://en.wikipedia.org/wiki/Vannevar_Bush
http://en.wikipedia.org/wiki/Differential_analyser

To change the transmit tones and bit rates, all cocoaModem had to do  
was to change the dx/dt parameter of the differential analyzer.   
Pretty much a one line of code change to go from generating PSK31 to  
generating PSK63.

Other programs may not use decimation or DDAs and will have different  
mechanisms to demodulate and generate PSK63.

PSK125 is basically the same thing -- bump the rate of everything in  
PSK31 by a factor of 4.

There is lots more computing in PSK125 than PSK31.  Four times more.

If you use the same filter profiles and all, PSK125 also use 4 times  
more bandwidth than PSK31.  The "rails" are 125 Hz apart and visually  
looks almost as wide as 170 Hz shift RTTY on a waterfall.  And the  
audio sound is truly ugly sounding -- much worse than PSK63, and  
pretty much sound as bad as HF Packet.  ("HF Racket.")

As such, PSK125 also has 4 times (6 dB) more noise coming to the  
demodulator than PSK31.  If you look at the typical shape of the Bit  
Error vs Eb/No curve (or character error vs SNR curve, etc), you will  
see that PSK31 could be printing virtually 100% while you get no  
print from PSK125 -- it just depends on where your signal is along  
the SNR axis.  But in general, you fail printing earlier than with  
PSK31.

To make up for the loss of SNR, the trick is to use lower case in  
PSK125, which would reduce the number of bits per character in the  
Varicode encoding.  Something that I did not see people do when I  
came across some signals yesterday evening.  Lower case would also  
make an exchange faster in addition to reducing print errors.  The  
PSK125 contesters still need to hone their skills.  It took a long  
while of constant cajoling for RTTY contesters to stop delimiting  
exchange fields that are alphabetic with a dash, although I still see  
some OR-OR exchanges, which not only slows you down in RTTY but adds  
extra time slots for error hits, especially when an error hit on a  
FIGS or LTRS is not especially pleasing.

PSK125 will also eat up more processor cycles; the PSK125 demodulator  
will use up 4 times more processing than a PSK31 demodulator.

(Just as I use 9 demodulators for RTTY in cocoaModem, I run 3  
demodulators for BPSK and perform a vote between their outputs -- a  
standard DPSK demodulator, an Okunev demodulator, and one that I had  
extended the standard Okunev 3-slice model to a 4-slice model (even  
slower than an Okunev, HI HI).   So PSK125 is definitely not for  
older computers if you use cocoaModem.  You can find out more about  
Okunev demodulators from his book "Phase and Phase-Difference  
Modulation in Digital Communications, Yuri Okunev, Artech House, ISBN  
0-89006-937-9.  A fascinating book.)

73
Chen, W7AY