What are good levels of PSRR before you consider a regulator?
I've been working on PSRR using an Aikido-style approach to reduce 50Hz hum and noise for a complicated design - cascoded dual differential with tube ccs front end, drivers and push pull paralleled tubes OTL. Class A. Fully modelled in LTSpice.
The power supply consists of two twin-phase rectifiers to produce ±320V and ±200V rails. Without a choke and approximately 1500-2000uF on each rail results in about 1Vpp for both. The total output noise on the channel is 200mVpp currently (although there's a strong harmonic spread).
The fun is that each rail isn't a inverse of the other rail, hence I've been looking at injecting each rail's noise into certain points to result rejection.
A simple regulator on each power supply rail to give the final smooth but it doesn't hurt to get the amp noise rejection as good as possible first. As always that follows the law of diminishing returns.
I still have to circle back and tune for signal again. I would expect that adding a regulator would then see a return to psrr and then signal tuning again as the final step.
I've been working on PSRR using an Aikido-style approach to reduce 50Hz hum and noise for a complicated design - cascoded dual differential with tube ccs front end, drivers and push pull paralleled tubes OTL. Class A. Fully modelled in LTSpice.
The power supply consists of two twin-phase rectifiers to produce ±320V and ±200V rails. Without a choke and approximately 1500-2000uF on each rail results in about 1Vpp for both. The total output noise on the channel is 200mVpp currently (although there's a strong harmonic spread).
The fun is that each rail isn't a inverse of the other rail, hence I've been looking at injecting each rail's noise into certain points to result rejection.
A simple regulator on each power supply rail to give the final smooth but it doesn't hurt to get the amp noise rejection as good as possible first. As always that follows the law of diminishing returns.
I still have to circle back and tune for signal again. I would expect that adding a regulator would then see a return to psrr and then signal tuning again as the final step.