If I got the theory and maths of it right. As per my example above, I'm feeding 3.2V into a 2 ohm load. That's 1.6A and 5.12W. The FET is basically acting as a resistor creating a voltage divider, so in order to do that it would need to draw the same current from the supply; 1.6A. 1.6A at 12V is 19.2W meaning 14.08W needs to be dissipated? If that's correct then yes, it is a rather large amount of heat even if it is just for brief spurts.
I tried it at 2Hz and it just doesn't work. The heating coils used in e-cigs are tiny, I can visibly see the coil pulsing with excessive heat during the ON phase, that would cause burning and it still rattles, just slower, sounds like a clock ticking. Even at 30Hz there is some pulsing, the coil reacts quickly enough to cause physical movement at the frequency of the PWM. I can use something like 2kHz, and that stops the coil vibration, but the circuit itself emits a musical tone that is quite loud. Still doesn't cure the tank-o-meter problem, in that users need a smooth output voltage.
What cures one thing breaks something else eh? lol. I think I'll have to read up on Switch Mode supplies. That seems to be the way to go. Using a dedicated PWM controller and switch mode circuit.