[GreenKeys] Just about all you need to know about Selector Magnet Drivers

[John Nagle]

>From: Don Robert House <k9tty at dls.net>
>Date: Fri, Sep 28, 2012 at 3:54 PM
>Subject: Just about all you need to know about Selector Magnet Drivers*
>
> Attached please find Teletype Corporation 50042S
> This covers all information for the 177010 Selector Magnet Driver
> used with the Models 15, 19, 20, and 28 machines.

> http://www.navy-radio.com/manuals/tty/sel_mag-15-19-20-28.pdf

    That's interesting.  It's useful to have an actual Teletype
Corporation spec with waveforms for driving a 15 selector magnet.
In the past, I've looked at designing more efficient drive
circuits that wouldn't require such a big power supply and
ballast resistor.  You really only need 120VDC for the first
1 or 2 MS of a space to mark transition, to overcome the
inductance of the coils.  After that, the sustain voltage is
very low.  (0.60A * 220 ohms = 13.2V).  The circuit shown
does that.  They apply the full supply voltage through only
a 27 ohm resistor (R11) and Q5, so for a brief period the
selector magnet is almost directly connected to the power
supply.  This would burn out the selector magnet if that
state continued.  But once the selector magnet's inductance
has been overcome and the voltage at the selector magnet
has neared the 120 volt rail, Q5 cuts out and Q4 turns on.
Then there's a 700 ohm (R10) resistor in series with the 120V
supply, limiting the current to 60mA.  I would have expected
R10 to be 2K, so I'm missing something.

    It's common today to drive relays and solenoids with circuits
that supply much more voltage during turn-on, but this is the first
time I've seen it for older Teletypes.

    They do maintain 60mA through the entire MARK period,
so they're being conservative about holding current.  But they
accept a slower rise time than I'd thought was necessary.

    It's interesting that the circuit has an active load dump
(CR9 and Q2) to dump the energy in the coil on a MARK to SPACE
transition.  (This is the effect which causes arcing on keyboard
contacts in local loops.)  I'd been trying to design that entirely
with passive components, but their approach is better.

    Much of the complexity of that circuit comes from the limitations
of early transistors, which didn't handle either high voltages or
high currents well.  It's a very '60s circuit.  Everything is somewhere
in the middle between the power rail and ground, like tube circuits.
The selector magnet connections have active components on both sides.
Today, we tend to have active components connected to either
GND or Vcc and outputs relative to ground.

				John Nagle