This circuit uses the LM431 as an alternative to strings of high voltage zeners. High voltage zeners are subject to drift with temperature.
The LM431 is used in an active feedback circuit.
R49 R50 determine output voltage. Q7 is current limit R43 determines max current. At least 1ma should pass through R44 for LM431 to work correctly. R47 reduces power dissipation in Q6 and is not needed for lower voltages. C1 is for stability although there was no oscillation on the prototype even with it not fitted. D2 prevents the output HT from being stolen when the main HT collapses.
The MOSFET needs a zener G-S if there is no internal one. Hope this is useful.
R49 R50 determine output voltage. Q7 is current limit R43 determines max current. At least 1ma should pass through R44 for LM431 to work correctly. R47 reduces power dissipation in Q6 and is not needed for lower voltages. C1 is for stability although there was no oscillation on the prototype even with it not fitted. D2 prevents the output HT from being stolen when the main HT collapses.
The MOSFET needs a zener G-S if there is no internal one. Hope this is useful.
If you want to compare the difference:
At 100V zener drifts by 100mV/degC so over 50deg (not unreasonable in a valve amp) thats 5V.
The LM431 is 50ppm/degC so thats .25V for 100V output.
So at least 10-20x better.
At 100V zener drifts by 100mV/degC so over 50deg (not unreasonable in a valve amp) thats 5V.
The LM431 is 50ppm/degC so thats .25V for 100V output.
So at least 10-20x better.
Last edited:
Have you considered a Maida Regulator?
Another point to consider is whether you need low drift. 5% variation in B+ is not likely to result in major shifts in performance of a tube circuit. That said, precision DC is fun so have at it if eliminating the drift is your thing.
Tom
Another point to consider is whether you need low drift. 5% variation in B+ is not likely to result in major shifts in performance of a tube circuit. That said, precision DC is fun so have at it if eliminating the drift is your thing.
Tom
The Maida works extremely well and is a very nice solution. Moderate to heavy heatsinking and overspeccing parts will have it live a long and happy life.
For whatever reason I've not had good luck with LM431, they seem to be somewhat fragile devices. Randomly failed open even at low dissipation.
For whatever reason I've not had good luck with LM431, they seem to be somewhat fragile devices. Randomly failed open even at low dissipation.
I use a TL783 in mine - then I don't need the MOSFET. The Zener is just a protect for the TL783 and it will also work with LM317 (just change the zener to a 30V part and adjust R1 to take the extra voltage). R3 should be 5W. TL783 needs a heatsink.
It works remarkably well. Silent phono stage (seriously, I HATE hum).
It works remarkably well. Silent phono stage (seriously, I HATE hum).