I haven't done a circuit simulation, but I have built and tested the circuit and made measurements with both digital and analog scopes. With respect to circuit simulation, MicroCap is now free ( http://www.spectrum-soft.com/download/download.shtm ) so anyone so inclined can do simulations on all types of circuits. There are good tutorials on YouTube that describe how to use Microcap.
Below is a PDF of the schematic for my non-MCU version of an electronic selector magnet driver. Basically, when the TTY line goes from spacing to mark, both Q6 and Q4 become active (ON). At this point the effect of Q4 is insignificant and Q6 is doing the work. At this point the loop resistance consists of the ON resistance of Q6, the fuse resistance, the resistance of the selector magnets and the current sense resistor (R1, 10 ohms) - all of that totals about 80 ohms. The voltage source is 36 to 48 VDC. Because the current that flows through the magnets also flows through R1, the instantaneous voltage across R1 is directly proportional to the instantaneous current through the magnets and thus is directly proportional to the instantaneous magnetic flux produced by the selector magnets. This circuit, like all R-L circuits, will have an exponential current rise, and the final (asymptotic) current of this circuit will be about 350 mA. But the circuit will never allow the selector magnet current to get anywhere near 350 mA. Instead, U2A is set to trip when 600 mV is across R1, which is also when 60 mA of current is flowing through R1 (and thus 60mA through the magnets). This means that this circuit is operating on just the very first part of the exponential curve, and that part of the exponential curve is essentially a straight line. So when Q6 goes active, the current through the magnets rises in a straight-line linear way at a rate of about 17 mA per millisecond. At about 3.5 ms after the TTY line went from space to mark the instantaneous current through the selector magnets is about 60 mA and the voltage across R1 is about 600 mV. When U2A "sees" 600 mV across R1, it will clear the latch formed by U1C and U1D and will shut off Q6, but Q4 remains active. The magnets are still at 60mA but now the loop circuit consists of a 12V DC driving voltage and a loop resistance of about 400 ohms (330 from R8 plus 66 from the magnets) so the magnet current falls down to 30 mA and falls with a classic exponential curve because in this case the final (asymptotic) current of 30 mA is reached. Note that once the magnet was fully pulled in with 60 mA of magnet current, only about 20 mA is required to hold it active, and thus the 30 mA is more than adequate to hold the magnet.
When the TTY line goes from Mark to Space, Q4 remains active for a few more milliseconds, as determined by U2B, C2 and the 100K trimmer, and when the timeout has expired (~2.5ms), Q4 is cut off, and the magnet current goes (almost instantaneously) to zero. Normally, the trimmer is adjusted to extend the mark time about 2.5 ms, to compensate for the time on the front end when the magnet current was rising over a time period of about 3.5 ms. I'm assuming (guessing) that the magnet is effectively pulling in at around the 40 mA point (but it could be less). Note that when Q4 is cut off, and the magnet current is interrupted, a big voltage spike will be produced across the magnet coil by the collapsing magnetic field. The purpose of the 1nF across the magnets is to limit the instantaneous voltage spike so as not to produce (significant) clicking RFI, but the 1nF is not enough. It just stops the step function, but the voltage across the magnets would still go to a very large value (hundreds of volts) that could damage Q4 and/or Q6. The purpose of the series combination of 470 ohms and 100nF across the selector magnets is to limit the voltage excursion to something reasonable, like 150 volts. Because the R and C along with the L of the selector magnets forms an R-L-C circuit, that circuit does "ring" when current through the magnets is interrupted, but because of the series R the oscillation dies out in about one cycle (3 mS or so if I remember right). Note that these snubber values are right out of the ST-6000 schematic.
I built and tested this non-MCU version of the schematic on perfboard. But I now have PC boards back from the board shop for a version using a microcontroller and that allows me to remove a lot of the discrete circuit components. The MCU version is not tested yet (I've only tested the step up voltage converter) but I hope to get back to that project in a few weeks.
Paul - ad7i