Just one observation, the original circuit, back at the start of the thread, isn't actually a regulator. It's a pot with a current booster and a current limiter. Any variation of the input voltage comes through proportionately onto the output.
Mosfets can be reliable, but you can't take any shortcuts with them. Their destruction is fast, very fast. You can't abuse them for even a short time and think 'well, it's not for long'
A common misconception is that their gate-to-source resistance is immense, and so you can use very large drive impedances. You can, but only at pure DC. Power Mosfets have surprisingly large input capacitances into the nanofarads region and this gets forgotten when you have your DC goggles on. This can make reaction of regulator devices slow to transients and surges, and by the time the device reacts, something has gone outside its capabilities and has been damaged. If you want them to be responsive, you have to have fairly low impedance driver circuits.
Mosfet gate oxide is fragile, ratings are in the 8-20v range for power devices.
Bipolar transistors are also very fragile. Reverse bias of the base-emitter junction is the easiest way to kill one. Reverse b-e ratings are typically not much more than 5 volts. In this direction it's a reverse biased diode, so the resistance is very high, and the capacitance is very small compared to the mosfet. Th what extent the abused device lives depends on the current available when the junction avalanches. Low currents do damage, reduction in gain and Ft happens the transistor will never be the same again. Multiple events do cululative damage. Higher currents will do immediate and total destruction.
You can use the b-e junction reverse biased at low current avalanche as a noise generator. With an RF transistor the capacitance can be low and the output can go to UHF. It's a common trick in amateur radio circles for assessing the noise figure of receivers, but the transistors used this way are ruined for normal use and their noise level changes with use as well.
Back to Mosfets, the original circuit can be modified to be more robust. Firstly it needs a big enough device or bank of devices (with current sharing resistors) to handle the fault condition. The drive resistors and potentiometer resistance will need to come down in value. You need to also recognise that ordinary zeners are not fast. Quite sluggish and quite high capacitance. One trick is to use a properly fast diode to a point which is regulated by a zener, where the zener has a bit of current applied to it to set it up ready for action.
I have tens of thousands of mosfet transmitters out in the field doing pulsed operation approaching 500W dissipation during the pulse. I cannot for a picosecond let any voltage rating be exceeded, or any current rating. The devices have power rating specially for this application. Average dissipation is just a couple of watts. Very little heatsinking is needed, but the devices include enough thermal mass to handle the pulse. Should anything go wrong and the pulse last too long, the device fries. VERY expensively. But, with care, it can be done and millions of device-hours done so far without any design issues. Oh, and they have to survive induced lightning surge tests
The pulse modulator mosfet is chopping 50v at up to 12A in nanoseconds. The drive circuit has an impedance of only a few tens of ohms.
So rule number 1: Don't even think about short cuts or 'keeping it simple'
Rule number 2: Mosfet gates are high resistance items, they are NOT high impedance.
Rule number 3: Ordinary Zeners are too thick and slow as bouncers.
Rule number 4: Mosfets have infinite gain at true DC, but this comes crashing down when there are AC components and transients on the loose, so circuits behave quite differently to how we look at them with our DC goggles on.
David