>
getting +24 volts and +48 volts from the same supply
If you need a big 24V and a light 48V, and can possibly spring for a dual-secondary tranny (or two trannies), this is one sane way to do it:
Using one bridge as a center-tapped +/-35V, and a +24V and a -24V regulator, and ground moved to -24V, can "work", but is a real mind-bender to wire correctly and debug. Also the 48V sucks across both regulators, and as Jakob found a few months back, certain start-up conditions will pull one regulator "in reverse" and the whole thing locks-up.
Generating less than nominal 35V/70V to get regulated 24V and 48V is probably folly because power line variations and resistance losses could put you down at 26V in ripple-dips, and the regulator will go out of regulation, usually with a big buzz. See my post
"VAC=VDC" and comments.
Deriving a full high voltage, regulating to 48V for a small load, then tapping that to get a large 24V load, is very wasteful. There is about twice as much heat in the regulators as winds up in the 24V amp. The actual cash cost is small, and we can do "silly" things in DIY without bankrupting the company, but it still seems extravagant for no good reason. (It also needs a non-standard 24V regulator.)
If you just can't get a dual 24V winding: I suppose a 24-0-24 winding with one bridge will work. Ground the "-" pin of the regulator. Big cap (like 2,200uFd 50V) from there to the tranny "0" pin. Smaller (like 470uFd 100V) cap from ground to the bridge's "+" pin. The "0" pin feeds a 24V regulator, the "+" pin feeds a 48V regulator.
This would probably also work with a single 24V winding into the conventional voltage doubler. But in this case the 35V feeding the 24V regulator works half-wave, with high losses, high ripple, DC in the winding, and sometimes radiated field and acoustic buzz trouble.
And there is good reason to keep the Phantom power on a completely separate winding. Unlike other power supplies in good professional equipment, Phantom has power current flowing through a signal shield. And that pin 3 is likely to become shorted to the chassis (mike cables get wired funny), defeating any proper star-ground system. If the Phantom is totally floating, this is not necessarily bad. If it shares any conductor with the main power and signal ground, except at the XLR jacks, you've invented a ground loop. That's probably why commercial designs generally keep the Phantom apart from other power supply windings. In my sketch I've shown the +48V Ref pin returned to XLR pins 3 to reduce the problem, but I don't think that solves it. In fact it might be better, given a dual 24V tranny, to use one winding for the 35V raw 24V reg supply, and run the other into a voltage-doubler to get the Phantom. That way they are isolated and can't ground-loop through the XLR jacks.
