[GreenKeys] Early RTTY

[Jim Haynes]

I just read an article - August 22, 1924 issue of The Electrician, about
a system operated between Madras and Rangoon using the Baudot system.
This is about 1100 miles, and the circuit was operated full duplex using
25KW transmitters.  It replaced a Wheatstone Morse system running at 150
wpm, with a two-channel time division multiplex.  I wonder if they really
meant Baudot apparatus as of 1875, or if they meant something more
modern but using a five-level code.  The anonymous author believes
the system could be adapted to three or four channel multiplex if the
traffic required it.  Nothing is said about whether they used 
make-and-break keying or FSK, nor what frequencies were employed.

R. A. Heising reported tests done by Western Electric over a 25 mile
path in New Jersey.  The frequency was 666 Khz, and the transmitter
power is not clearly stated.  It used five "50-watt" tubes in parallel,
with 750 volts on the plates.  Make-and-break keying was used, and
operation was by four-channel multiplex at 180 wpm total.  The author
noted that most of the interfering signals in the neighborhood of
New York are from nearby spark stations, rather than from static.
This test operation began near the end of 1919.  [Journal of the Franklin
Institute, January 1922, "Printing telegraph by radio."  Gilbert Vernam
mentions a demonstration of his cipher system over the same radio test
setup in 1920. [Transactions of the A.I.E.E., vol. 45 p. 295, 1926]

There were two articles in the New York Times in 1922.  RCA tested
printing telegraphy between their station at Chatham, MA and R.M.S.
Majestic, reported in the November 14 issue.  The Navy tested printing
telegraphy between an airplane and a ground station, reported in the
August 10 issue.

E. H. Armstrong in 1928 published a paper suggesting the superiority of 
FSK - without a limiter - over make-and-break keying.  [Proc. I.R.E., vol. 
24 no. 5, 1936, p. 689, "Methods of Reducing the Effect of Atmospheric 
Disturbances."  His premise was that signal would be present in only the 
mark or space channel at any time, and that noise would appear in both 
channels equally and cancel out. His paper was criticized at the time by 
Carl R. Englund, and later in a paper by John Carson [Proc. I.R.E., July 
1928, p. 966, "The Reduction of Atmospheric Disturbances."] Although these 
critics showed mathematically that in some sense Armstrong was wrong, the 
ink-recorder traces included in his paper show clearly better signals when 
using FSK than when using make-and-break keying.  Perhaps the critics 
failed to take into account that Armstrong's FSK transmitted with about 
twice the power of make-and-break, considering that the transmitter is on 
all the time rather than being turned on and off.  Or perhaps the noise 
encountered in practice is not the kind that the mathematical critiques 
addressed.   As is well known, Armstrong went on to invent the use of
a limiter ahead of the discriminator for receiving, what we consider
true FM and which suppresses noise by a different mechanism.

About the same time Lawrence Schmitt of Teletype was working on a
two-tone or FSK scheme.  Presumably neither he nor Armstrong knew
of each other's work.  In 1924 he applied for a patent, granted in
1929, on his system. [U.S. pat. 1,705,211].  His concept was pretty
much the same as Armstrong's, that noise would cancel out in the
difference of the mark and space channels.  There are reports of
radioteletype experiments between the Teletype plant on Wrightwood
Avenue in Chicago and Sterling Morton's country estate, and also
between Wrightwood and Lake Geneva, WI.  There is a photograph of
Wrightwood sometime in the 1920s showing antenna poles erected on a
part of the lot that was then vacant.

In a paper presented in 1931, Bailey and McCann of AT&T discuss the
use of printing telegraphy on long-wave radio circuits and make more
cogent arguments in favor of FSK.  They focus on the fact that when
the transmitter is on the signal is more or less constant; when it is
off the noise being processed through the receiver is highly variable.
It is thus difficult to set a threshold for decision between mark and
space.  They give a very difficult-to-understand graph showing how
two-tone operation preserves printer margin over a much greater range
of received current than does single-tone operation.

They mention that, in addition to the rather short distance circuits
mentioned in previous Bell work, a printing telegraph system has
been in operation for three years on a long-wave (approx. 60 KHz)
telephone circuit between New York and London.  The teleprinter is
located in the same room as the telephone switchboard and is used
to exchange questions and notes about the operation of the service
that would be harder to deal with verbally.  This represents an early
instance of a printer which uses red and black printing to distinguish
between sent and received text.  [BSTJ, October 1931, p. 601 also
in Proceedings of the I.R.E., December 1931, p. 2177 "Application of
Printing Telegraph to Long-Wave Radio Circuits.]