04-08-2015, 07:12 PM
That could work well. Similar versions using bipolar transistors were the mainstay of voltage regulators for decades 
I'm guessing the pot should be above the feedback sampling point rather than below? As it stands, the maximum output would be 2 x (12V + Vgs)
The resistors should all be suitable for working at high voltage. Personally, I'd go for metal oxide, but metal film might be OK. Putting two in series helps with the voltage stress. The one feeding the gate of the second MOS-FET could allow a small amount of ripple in, so if it does get split in two, the mid-point could be decoupled with a suitable cap - even 100n would help, thanks to the high values of resistance present. Similar comments apply to the 470k feeding the Zener... I'd probably reduce that a bit, just to get a bit further around the "knee" - perhaps 120k?
(I'm sure the ripple would be good enough for this application, but hopefully those comments are useful for anyone thinking of trying such a regulator elsewhere where noise is a concern?)
In circuits like these, it might be worth considering an old trick - you can drop quite a bit of voltage over a neon bulb, and that bulb provides a useful reminder that the circuit is powered and might "bite". Tek used to do that in their waveform monitor power supplies, for example. So that 470k feeding the zener could be replaced with a neon in series with something in the 100k region...
I would be tempted to add a protection zener across the gate of the error-amplifier MOS-FET, but it's probably not needed in practice.
So Nick, plenty of ideas to consider, and I'm sure there will be more to come. When the time comes, it'll be good to experiment (carefully!) - you'll learn a lot in the process
I'm guessing the pot should be above the feedback sampling point rather than below? As it stands, the maximum output would be 2 x (12V + Vgs)
The resistors should all be suitable for working at high voltage. Personally, I'd go for metal oxide, but metal film might be OK. Putting two in series helps with the voltage stress. The one feeding the gate of the second MOS-FET could allow a small amount of ripple in, so if it does get split in two, the mid-point could be decoupled with a suitable cap - even 100n would help, thanks to the high values of resistance present. Similar comments apply to the 470k feeding the Zener... I'd probably reduce that a bit, just to get a bit further around the "knee" - perhaps 120k?
(I'm sure the ripple would be good enough for this application, but hopefully those comments are useful for anyone thinking of trying such a regulator elsewhere where noise is a concern?)
In circuits like these, it might be worth considering an old trick - you can drop quite a bit of voltage over a neon bulb, and that bulb provides a useful reminder that the circuit is powered and might "bite". Tek used to do that in their waveform monitor power supplies, for example. So that 470k feeding the zener could be replaced with a neon in series with something in the 100k region...
I would be tempted to add a protection zener across the gate of the error-amplifier MOS-FET, but it's probably not needed in practice.
So Nick, plenty of ideas to consider, and I'm sure there will be more to come. When the time comes, it'll be good to experiment (carefully!) - you'll learn a lot in the process
