[rec.audio.high-end] Regulators

jas@proteon.com (John A. Shriver) (03/05/90)

Regulators are prving to be very important in current high-end audio
design.  For example, Stan Warren (Superphon) argues that his preamps
are better than Paul McGowan's (PS Audio) because Stan uses shunt
regulators, and Paul uses series regualtors.  The choice of regulation
topology is being advanced as a sonic determinant.  The presence of
regulation is by now a given.

One can also look at the evolution of Audio Research's regulators.  At
least in the era when their schematics were being published in Audio
Amateur, one could see the regulation improve with each generation.
They went from a lousy zener/pass transistor regulator that was so
unreliable that they had to put in series resistance to keep blowing
up, to a whole chassis just for regulation (SP-10).  I replaced the
zener/pass transistor regulator in my scratchbuilt SP-3/6 with the
"POOGE" design from Audio Amateur that was in the Marantz Model 7
article.  It sounded better, and it didn't explode every six months.
(That sickening "crack" on power up.  I left out the de-regulating
resistors, which increased the risk.)

Joe Curcio is also a strict adherent to super-regulation.  No question
that the effects are audible, the Daniel pre-amp that replaced the
SP-3/6 showed the audible improvement most clearly.  I presume his
Stereo 70 modification (Glass Audio) also shows the results.  (Has
anyone built this?  Any comments?)

Regulators do two things.  One, they attempt to allow your circuits to
actually operate from a nearly-true voltage source.  The stage does
not encounter distortion from modulating its power supply.  Two, they
keep the stages from modulating each other through the power supply.
This eliminates unexpected feeback and feedforward loops, and cleans
up distortions that have a time factor (inter-stage delays are very
small, but non-zero).

As people continue to learn how we hear, distortions in the time
domain are being seen as very important, the ear has a large
time-domain component.  Power supply regulation helps a lot in the
time domain, but not much in the "frequency response is everything"
universe. 

Good old three terminal regulators (LM317) are a damn sight better
than what is on most equipment, and make a great starting
modification.  (They are also smaller than lots of capacitance.)
However, there are regulators with lower output impedance, and output
impedance that is flatter in the frequency domain.  These are often
servo based, using a three-terminal regulator soley as a voltage
reference.  Also, for regulated tube supplies, you can't get a
regulator on a chip, and have to build from parts.