[GreenKeys] An advanced approach to driving a 60mA Teletype from a USB port

[John Nagle]

    Some time ago, I did a design for a Model 15 Teletype
driver that would be powered entirely from a USB port, eliminating
the need for a 120-130V power supply. I never built this, because
the parts required are surface mount only and I don't have the
tools to solder surface mount.  But I thought I'd share the approach.

    Here's the LTSPICE file:

http://www.aetherltd.com/public/flashsupply6.asc

    The free version of LTSPICE
("http://www.linear.com/designtools/software/ltspice.jsp") will read this,
and you can run the simulation.

     Graph the current through L3, the selector magnet.
(Point the cursor at a device on the schematic while the
simulation is running, and the icon turns to a clamp-around ammeter.
Click and you get the current added to a graph.  Pointing to a wire
gets you a voltage on the graph.) You'll see that it delivers about 60mA through
the whole MARK period.  Look at the voltage at the top of L3, and
you'll see that it goes up to 120V at the start of the MARK period,
then drops to 2 to 3V, still delivering 60MA.

     What's going on here is that the LT3484-1 IC, which is a photoflash
controller form of a switching power supply, charges C2 up to 120V
during SPACE periods.  L1/L2, which is a stock pulse transformer, is
doing the step-up.  On a transition to MARK,
S1 closes, dumping C2 into the selector magnet, providing the
necessary 120V to overcome the 4H inductance of the magnet coil.
LT3080 is just a linear regulator, providing 3.3V to sustain the
selector magnet once the initial pulse has pulled it in and
C2 has discharged.  You really need 120V only for the first 1ms
or so of each MARK period.  Once the current has stabilized,
the voltage needed is only 60mA * 55 ohms = 3.3V.

    The design needs a little work; the inductive kickback from
the selector magnet needs to be snubbed without stalling the
switching power supply.  A full system would need a USB to
serial converter chip, a keyboard input side, and motor control.

    The design is a clear demonstration that you could run a Model 15 Teletype
from power taken from a USB port. The circuit never draws more than
250MA at 4.75V, (the voltage you're guaranteed from a USB port) and that's only
half the current a powered USB port can deliver.   If you model the
standard "120V supply through a 10K resistor" approach to driving a Model
15, you get waveforms with current rise times no better than this provides.
So this approach should work.

    Note that this is designed for 45.45 baud.  At higher baud rates, there
may not be enough SPACE time to recharge the capacitor.  Also note that
this provides only enough energy for one 55 ohm 4 H selector magnet.  It
won't power multiple machines on a loop.

    This is how to approach the problem using modern technology.
The resulting unit would be tiny.  None of the parts are expensive or difficult 
to obtain.  Power consumption is around 1 watt.

                				John Nagle