[GreenKeys] Polar Relay

[Jim Haynes]

On Tue, 7 Feb 2023, Harold Hallikainen via GreenKeys wrote:
>
> I have heard that Teletype liked mechanical stuff and wanted to minimize
> electrical/electronic circuitry. They used a mechanical UART (though
> electronic ones did not exist at the time) and directly drove the selector
> magnet instead of including an electronic driver. They could have made it

Howard Krum considered it quite a triumph when the mechanical receiving
selector replaced the receiving distributor sending to multiple 
electromagnets.  We have to remember than in the 1920s wire insulation
was not all that great and contact materials were a problem.  So the
earlier electrical sending and receiving distributors were the source
of a lot of trouble.  And multi-magnet receiving machines continued
to be used for time-division multiplex systems, which were mostly popular
with Western Union.

Then there was the force of company history, in that Teletype engineers
were so clever with designing and manufacturing complicated mechanisms
that it became a sort of achievement challenge to develop ways to do
things mechanically and avoid electrical alternatives.  And of course
stamping parts out on puch presses made them cheaper than if parts had
to be wired up.

I was told that the cost to manufacture a Model 33 keyboard was about
$50.  We would never have believed at the time that the day would come
when you could go into Walmart and buy a keyboard with over 100 keys
for about $10.

The UART was an exciting development, but apparently originated with
Digital Equipment Corp. rather than with Teletype.

The whole thing played out with Model 37.  Engineers worked long and
hard to design a mechanical selector that would work at 150 baud.
Which considerably delayed the product getting into manufacture.
I would say the last successful product of the all-mechanical genre
was the stock ticker, which used Model 37 technology but did not have
to run at 150 baud.  And the Model 37 keyboard was based on Model 32/33
technology.  Never should have been done, considering that the competing
product was the IBM 2741 based on the Selectric typewriter with its
superb keyboard touch.  (But was highly mechanical)

And some old lessons had to be learned over.  In the early days electrical
contacts were a major problem area.  Contact materials, contact erosion
from sparking, contacts getting dirty from operating in an oily 
environment.  Good reasons for keeping things mechanical.

In Model 35 signals generated by the Model 28 style of keyoard were
less clean than desired - bit boundaries tended to shift around from
one character to the next.  A solution was developed, a circuit using
an SCR and a timing contact that would sense the natural keyboard
signal and retime it with the pulses from the timing contact.  This
was a plug-in option for the machine.  Well a day came circa 1965 when
AT&T decreed that all future shipments of Model 35 for the Bell System
would be equipped with the circuit (which was called a regenerative
repeater, even though it was not what one usually things of under that
name).  Well it turned out that the circuit didn't work very well.
The aisles in the factory started filling up with machines that were
completed but could not be shipped.  Lots of intense study revealed
that the problem was the old one of oil films being formed on the
contacts.  The contacts would work fine at higher voltages and currents,
but electronic signals were too weak to burn the oil films off.  I
don't remember now what solution was chosen, probably increased voltage
and current to clean the contacts.

Which reminds me of another funny story about the Model 35 tape reader,
but I'll save that for later.