I'm restoring a capacitor microphone PSU. Schematic is as follows
I have an EE question about the existing circuit. [Whatever the answer, I'm going to replace the upper 4-5V heater supply with a 7805 (and the lower part with a TL783 jacked up to 120V output), i.e. this is not a repair question.]
It is arranged via the current-sense resistor, the 9V1 zener, and the OC26 (germanium) pass-transistor as a constant-current source. Considering there is a valve in the microphone and considering the low signal levels prevailing, it surprises me that it isn't a constant-voltage source. Would that be for example because of techology limitations at the time (1964)? say even just crude parts count? Or is there a technical reason concerning valve heaters that I'm not aware of?
The schematic is quite interesting. Note the huge and bootstrapped 180M grid resistor on the valve, to match the high-impedance output of the microphone capsule (only a few pF I believe).
For reference this is a Phillips EL 6154/01, aka AKG N60A.
Best Answer
The OC16 based current source is good because the valve gets DC for the heater and inrush current issues stated in the comments are covered. Sure it's not the worlds greatest current source but mains variations aren't that great so it will be fine. Remember that in 1964 transistors were much more expensive so "better" but more complicated current source circuits stayed within Academia.
Your voltage source proposal is actually better than an ideal current source as long as you limit inrush current by whatever means using the previous comments as a guide. Steady state tube operation is best on a voltage source because the increasing resistance with temp characteristic of the filament assists thermal stability. Remember that temperature is the more important parameter than current or voltage.
Now you want to regulate your HT .... Fine why not? Remember that in 1964 high voltage transistors were unobtainium so the RC filtering was the cheap reliable way to get the ripple and hence hum down. If you regulate the HT you achieve lots of ripple rejection if the reg is worth its salt. Ripple rejection is much more important than V out accuracy. In fact why don't you just do a simple pass transistor and not have to worry about jacking up anything?