[GreenKeys] 85 Hz shift

[Jim Haynes]

Yeah, I've been kinda wishing to try out 85 Hz shift in ham RTTY, and also
wishing someone would implement that MD-522 diversity scheme in a software
demodulator so we can see how well it works without having to have two
guys with MD-522s set them up.  But I nag W1HKJ for enough things as it
is; hate to request yet another useless feature for fldigi.

It's my belief that the choice of 850Hz shift in the 1940s was mainly to
accomodate drift in the radio equipment of the day.  You can see in the
AN/FGC-1 documentation that it does not have a linear discriminator, but
rather mark and space filters that are fairly flat over a wide frequency
range.  So this would let you accomodate a fair amount of drift without
having to retune nor have distortion increase.  Later equipment goes more
toward linear discriminators, but with AC coupling and DC restoration
after the discriminator, so that being off frequency does not bias the
signal. (Until it gets severely off frequency)

Amateurs no doubt adopted 850 because that's what the commercial and 
military operators were already using.  And amateur equipment was more
likely than commercial/military to have drift in those days.  The
difference was that the commercials wanted a radio circuit to be good,
untouched, over a period of hours, while the amateur conversations were
shorted and conducted with a hand on the receiver tuning knob.  A
significant problem for hams was that we had to cohabit a frequency space
with CW operators; and the CW operators resented us taking up about a
KHz of spectrum for a conversation they could have handled in 100 Hz
or less.

It's doubtful there was ever a correctly-designed amateur demodulator
for 850 Hz shift.  It was not until 1968 that Don Wiggins W4EHU, an
E.E. professor, published an article in RTTY explaining why the
discriminator needed to be linear far beyond the frequency range
occupied by the shift, and that it was very important to have bandwidth
as narrow as possible ahead of the limiter.  By then amateur radio was
well along the road to abandoning limiter-discriminator demodulation
and pursuing two-tone, which treats the mark and space signals as a
frequency diversity pair.  The ordinary amateur demodulator, even if of
the discriminator type, tended to tune the peaks of the tune circuits to
the actual mark and space frequencies.  This made it convenient to use
the discriminator as a tuning indicator but worsened its tolerance to
interfering signals.

Especially with the onrushing popularity of SSB phone amateur equipment
had to become much more frequency stable, and this made it feasible to
employ much narrower shift than 850 Hz.  There was a lot of discussion
and trial before the community settled on 170 Hz as the new amateur
standard.  For quite a while we were building demodulators that could
accept either 850 or 170 Hz shift.  Eventually 850 became practically
extinct in amateur use.  Irv Hoff noted that competition from CW operators
became much less a problem when 170 Hz shift was being used.  However
170 Hz throws away some of the frequency diversity advantage of two-tone
detection that we get with 850 Hz.

Frequency stability of amateur equipment has further improved since
170 Hz was being adopted.  Maybe 85 Hz would be a viable option today.