[GreenKeys] Loop supplies, or how to drive a selector magnet.
John Nagle
nagle at animats.com
Wed May 18 18:23:56 EDT 2016
Here's an explanation of what's going in driving a selector
Model 14/15 60mA selector magnet, and what a "loop supply" really
does.
This is a simulation with LTSpice IV, a free version of Spice
from Linear Technologies. This simulation is of a classic
120V loop supply with a 2K resistor in series switched by an
optoisolator. The graph shows voltage across the selector
magnet, and current through it. The selector magnet has an
inductance of 4 henries and a DC resistance of 55 ohms, which
is standard for a Model 15 Teletype.
http://www.aetherltd.com/public/misc/techdesigns/selectorcurrentsimple.png
At turn-on (SPACE to MARK), the voltage immediately climbs to
almost 120V. Because of the huge inductance of the selector
magnet, the current takes almost 10ms to rise to near 60mA.
In the steady state (MARK), current levels out near 60mA, but
the voltage across the selector magnet is only about 3.5V.
Most of the power is now going into heating up R1, the
2K resistor.
What this tells us is that the efficient way to run a
selector magnet is to charge up a capacitor to 120V, and on SPACE
to MARK, dump the energy in the capacitor into the selector
magnet to pull it in. After that, only 3.5V is needed to
keep the selector magnet pulled in.
So here's a circuit to do that, powered entirely from a
USB port.
http://www.aetherltd.com/public/misc/techdesigns/selectorcurrentcompare.png
(This won't actually work; the transformer specs are
wrong and the parts choices aren't that good.) There's
a switching power supply (the LT3484-1 and the transformer)
to step up 4.5V to 120V. This charges up 1uF capacitor C2,
then stops charging. (That IC is intended for camera flashes.)
On SPACE to MARK, opto S1 turns on and dumps the capacitor into
the selector magnet.
In parallel with this a simple linear 3.3V power supply,
the LT3080. D3 and D4 allow both the high voltage supply
and the low-voltage supply to feed the optoisolator. So
after the capacitor is discharged, the selector magnet still
gets 3.3V to provide 60mA, keeping the selector magnet engaged.
Look at the graphs. Note the blue lines, which indicate current
through the selector magnet. Both ramp up quickly to 50-60mA.
Both take the voltage up to 120V, then taper off in under 5ms.
Power consumption of the switching supply is far lower. There's
no huge ballast resistor. This switching supply needs only
200mA at 4.5V, so it could be driven from a USB port
(500mA at 4.5V).
This is just a proof of concept and needs more work. The
Model 33 Teletype uses a selector magnet driver board
which does something similar. The insight here is that
if you can charge up a 1uF capacitor to 120V, that
stores about the right amount of energy to pull the
selector magnet in fast. Then it's just 3.5V to hold it.
John Nagle
More information about the GreenKeys
mailing list