TL783 - are there any newer high-voltage linear regulators?

The DMOS approach looks good. I tried the current limiter in LTSpice, and it works well:

I'm not clear on why R3 works at 16 ohms. Vgs for that DMOS part is -1.5V, and that's what the SPICE model has. Supposedly, you should set R to Vgs/I, where I is the desired current limit. Or so says "Current sources and voltage references", by Harrison, fig 7.9. So R = 1.5V / 0.060A = 25 ohms. But in Spice, that's too big; 16 ohms works.

Here's the entire cap charger in LTSpice.

This is a driver for 60mA current loop devices, powered from a USB port. There's a USB to serial controller, not shown. I've built and used this, but without the current limiter. Without it, it only works for 220 ohm 5.5H selector magnets. There are some other devices in the Teletype world that want 60mA but have different solenoids, so this thing needs current limiting to be more general.

John Nagle

The "Vgs" spec is the voltage at which there is *no* (or minimal) current.

Nothing is quite as abrupt as LTspice "idealisms" might imply >:-}

To see what I mean, "curve trace" the device: Sweep VDS with a small increment, say 10mV, while doing a STEP VGS in 100mV steps.

...Jim Thompson

Very long ago I found a AM radio by the side of the road in a pile of junk. It was a very cheap GE transformerless radio. From it I salvaged the audio output transistor, a TO-66 style case. This transistor was fed right from the 120 VAC rectified power line so it was capable of handling 160 volts. I used it to switch the selector magnets on my Model 15 TTY, I forget if I wired it up as 20 or 60 mA. Anyway, that transistor worked just fine for years. Oh and for current limiting I had a resistor capable of dropping about 120 volts from the 160 volt DC power supply, so it must have been around 2K or 6K or so. So anyway, you don't need a constant-current driver so much, a resistor should work just fine.