Mike Hanson
Trying to understand...
Well, that's a wonderful bit of insight! I thought perhaps I could put them in series like that, but this is the first time I've encountered the need, so I wasn't sure whether it was wise to do so. I've got a collection of Zeners on hand, so I may already have everything I need.Easy; use 5-6 , 12v zeners in series. or even 10 x 6.9Vz.
There's a significant noise advantage to doing this.*
Drive them fairly rich, 8-10mA, constant-current, for best effect.
NB that also means the dissipation is divided across teh string.
You'll probably only need 5 at that sort of bias current. Or trim with 5.6-10v zener in last position of 6 to suit.
* a statisical thing: because the zeners noise contributions are random, the addition is RMS... so you'll only have about SQRT(5) x the noise of a single 12v zener at that current; or SQRT(10) = 3.162 x the voltage noise of a 6.9V** zener , for a 70-72VDC voltage reference - and /not/ the full noise of a single c. 70v zener (which will be considerably worse, and need to dissipate up towards ~1W.)
** It's right around this value, maybe a little lower, that treu Zener action happens, not avalanche (higher voltages): and that's both best regulation, and - least noise. 10 zeners at 10mA = 70mW each, buy 20, cost about a buck or two .. about as good as you'll get.
Use a dab of RC following c 470-1K/ 47-100uF, to drive the base/gate of your pass device - and the whole thing is beyond contention. Broadband output noise/ripple at c -100dB, maybe a bit more - easily achieved with good layout.
I'll sketch the idea using CircuitLab, to confirm that I'm on the right track.