[GreenKeys] Terminal Units

[jhhaynes at earthlink.net]

And I would be interested to hear what technologies have been used in
the ones you have designed.

I have a complete run of the original RTTY magazine if you need any
information from that.  That is RTTY as put out by Merrill Swan - I
dropped my subscription sometime after ownership of the magazine
passed to others.  Of course the complete run of all the RTTY journals
is available from HAL on CD ROM.

My first one was the W2PAT design, except I didn't have TV width
coils but was able to get some toroids as I recall.  Later on
somebody published a modification for that which was supposed to
allow copy on single tones.

Oh, but first let me point to the ancestor of them all, U.S. patents
1,705,211 and 2,012,407 by Larry Schmitt of Morkrum, which became
Teletype.  (www.uspto.gov, you can search for patents by number and
view or download)  So far as I have been able to learn, all the other
early RTTY work was done with make-and-break keying.  I believe Schmitt
pioneered FSK for the purpose.  (There was some earlier transmission
of FSK, involving arc transmitters where it was easier to shift the
frequency than to turn the carrier on and off.  But this was used for
Morse, and did not take advantage of FSK.)  Schmitt was working on
RTTY at the same time Armstrong was working on FM, but there was
apparently no mutual awareness.  Schmitt seemed to come to FSK by
analogy to polar DC transmission on landlines.

Then there was the pre-WW-II work of various people at Press Wireless.
I have a long list of their patents; and they published an article in
the November, 1944 issue of Electronics covering their TU design.
One of their patents that I don't understand is 2,459,485 by Bartlett.
Bart, W6OWP is still living and was one of the pioneer amateur RTTY
operators.

Amateurs were not allowed to use FSK on HF until February 1953, except
in the former 11-meter band.  They used AFSK on VHF, and some
make-and-break on HF.

I used to have the schematic of the W2BFD TU, but it has got away from
me.  This was a well-thought-out design that was duplicated by a lot of
hams in the New York area.  It occupied a big rack panel and was intended
to mount between the back legs of a Model 12 table.  The New Yorkers had
acquired a lot of Model 12 machines when the NYPD modernized.

The wartime AN/FGC-1 had the advantage of a Bell Labs design team and
cost (and weight) no object.  It's not all that good as a TU because
the filters were quite broad to accomodate transmitter and receiver
drift.  Indeed I think the reason for standardizing on 850 Hz shift in
the early days was because of drift and the need for wide filters to
accomodate it.

A lot of these early designs followed the detector with a threshold
corrector, so that the output signal mark/space threshold was not
upset by the signal drifting across the discriminator passband.

A little to the side of TU design, but in WW-II there was a device for
RTTY reception that supplied start and stop pulses from timers in
synchronism with signals received from a tape transmission.  The intent
was to decrease the terrible character error rate that results when
the STOP pulse is missed and the receiving shaft continues to turn.
This results in several errored characters before the receiving shaft
again gets into the proper phase to be stopped by a STOP pulse.  I've
often wondered, and have never learned, how effective this device was.

The early years of RTTY magazine are full of TU designs of all kinds.
There was a ham named Gates who designed a set of filters, so there
was a "Gates TU" made by a number of hams.  Merrill Swan and Herbert
Hoover Jr. were active designers.  There was a design involving beam
switching tubes that I built and was never able to get to work.

Bob Weitbrecht W6NRM was probably the most prominent TU designer before
the rise of Irv Hoff.  His first design employed choke coils with
the I laminations removed, putting two of them face-to-face to get
a bandpass coupled circuit.  Later everybody went to 88 mh toroids.

When people first started using 170 Hz shift there was some use of
a circuit called a "double doubler" to multiply the frequency by 4
and then 170 Hz shift could be received on a TU with 850 Hz filters,
if they were not too narrow.

There were some papers in the late 1950s by Prof. Baghdady of MIT
on the virtues of positive feedback around the limiter.  The idea
is that the FM capture effect improves the SNR, so if you feed some
of that improved SNR back around the limiter you get a further
improvement.  I did some experiments with this for an FSK TU and it
did seem to improve copy; but I never built a complete TU using it
nor published a design about it.

There was also an article by Lerner of MIT on bandpass filters with
linear phase.  An interesting feature was that most of the inductors
had the same value.  I never built any of these, but W6ZBV told me
he had built some and they worked.

Prof. Don Wiggins W4EHU wrote a number of articles briding the gap
between theory and practice for the ham RTTY community.

In the mid 1960s some of us got interested in the idea of a TU without
a limiter, detecting the two tones separately and combining the
detector outputs as in diversity reception.  Vic Poor and Keith
Petersen did some of the theoretical work and Irv Hoff did a lot
of building and testing, resulting in the TTL-2 TU.  I did some
commenting and cheering from the sidelines.  I believe this technology
showed up in the Dovetron as well as in the ST-6.  And there is a
TU design that some Navy MARS people did called the DT-600 that also
provides for switching the limiter on and off.  I have a couple of
these.  Limiterless detection turned out to be fairly unimportant,
because if the signal was so weak as to benefit from limiterless
detection it was usually too weak to copy anyway.

The two-tone (limiterless) TUs sometimes employed a "slideback" detector.
This attempted to measure the average tone level and use that to develop
a detection threshold for the tone that was going on and off.

There was a patented circuit which Irv valued highly, called a
Decision Threshold Computer, DTC.  This was a more modern take on
the threshold correctors of the 1940s and was intended to compensate
for the mark and space signals fading independently as well as for
drift.

Back in the 50s and again in the 60s people designed simple TUs in
which the discriminator is an analog frequency "counter".  The basis
of this design was a frequency meter of the period.  The input is
run though a limiter and then fires a pulse generator at each zero
crossing of the signal.  The pulses are integrated with an R-C circuit
so that the voltage becomes proportional to frequency.

Especially when suitable ICs became cheaply available, people built
some TUs using phase locked loops.  These too were simple, but unless
you had a good limiter-filter in front of them they didn't perform
very well.

There was also a technique I wanted to experiment with, but was too
lazy to do so, called frequency feedback.  The idea is to beat the
incoming frequency with the signal from a VCO, and put the result
through a discriminator that controls the VCO frequency so as to
keep the beat frequency nearly constant.  This lets you run the beat
frequency through a single narrow filter.  I'm sure there was some
commercial use of this, tho perhaps not for RTTY reception.  There
is at least one book on the subject: Phase-locked and frequency-feedback
systems by Klapper and Frankle.

There was some interest in phase-shift signaling, and perhaps some
experiments by hams for RTTY.  It never caught on.  The Robertshaw-
Fulton company had a system called phase multilock.  The idea was
to use a 3-phase detector with one of the phases forbidden, so that
it was possible to resolve the mark-space ambiguity.  W6ZBV wrote
this up in RTTY.

Finally there is the idea of using synchronous signaling and detection.
Some Collins people (Heald and Clabaugh) experimented with a "Predicted
Wave" system and then there was the Collins Kineplex multiplex system
which ran synchronously using integrators derived from mechanical filters
as detectors.  Also Teletype built a device called the Monoplex which
turned TTY signals into 6-unit synchronous code and worked with a
detector developed at Bell Labs.  Circa 1970 I built a circuit that
would turn TTY signals into 7.00 unit synchronous transmission.  I
had a collaborator at the time who was going to build a receiver that
would lock to the character stream.  I'm not sure whether he ever got
it to work.  Eventually the K6STI RITTY DSP software took advantage
of all this.  It turns out to be not all that helpful.  As with
limiterless detection, when signals are so bad that you get an
improvement with synchronous detection then they are usually so
bad that the synchronous detector can't do much with them either.

I played with a CV-89 TU in the late 1950s and was not impressed.  One
limitation of that kind of TU is that they had to receive various
shifts and fax as well as TTY signals, so they could not be well
optimized for RTTY.  I also have a TMC TU that suffers from more or
less the same problems.  A lot of the military TUs also suffer from
having a CRT tuning indicator that is a lot harder to use than the
kind amateurs were using regularly; some sort of flipping line that
you are supposed to adjust between two horizontal marks.

jhhaynes at earthlink dot net