"Stacked" Power Supplies - any thoughts?

[Magnus Danielson]

>>>>> "S" == Stopp,Gene  <gene.stopp at telematics.com> writes:

 S> Hi DIY,

Hi Gene!

On the disclaimer side: My below comments is just me brainstorming, so
use it as some leads to investigate...

 S> I'm building another modular synthesizer. It is a slow process, a bit at
 S> a time. I'm incorporating a bunch of experiments that I've done over the
 S> past couple years behind panels and in a case, including some
 S> not-so-common modules like a Shepard Function generator, a quad VC
 S> bandpass filter bank, a VC commutating filter (the Electronotes
 S> "Quasi-Digital Bi-N-Tic" filter), a stereo 18-bit audio delay made from
 S> Crystal chips,  and a custom octal MIDI-CV converter made from a 486
 S> motherboard. There will be other more "normal" modules as well, plus an
 S> ASM-1. The cabinet is finished - it is about 38" wide by 26" tall by 12"
 S> deep. The panels are four large aluminum plates, with laser-printed
 S> graphics on laminated paper glued to the panels. Patching is done with
 S> banana jacks.

Hmm... interesting!

 S> Anyway, I'm looking into my power supply options. The PC motherboard
 S> will have its own supply. As for the analog side, I can always build one
 S> from scratch. However, I have a box full of Power-One +5/-5 supplies
 S> (1.5 amp each), plus a box full of Power-One +5 @ 3 amp supplies. Here's
 S> my idea: take three of the +5/-5 supplies and stack them, like this:

 S>           -5   Gnd   +5 --------- -5   Gnd   +5 ----------- -5   Gnd
 S> +5

 >> From the left, we get -15, -10, -5, Ground, +5, +10, and +15 volts.
 S> Notice that the center terminals (the "grounds") of the outer two
 S> supplies are used for the -10 and +10 rails. The two outer supplies will
 S> have floating chassis, the center one an earthed chassis. The primaries
 S> of all three mains transformers would be connected in parallel and
 S> driven by 110 VAC.

One thing here... now that you have cut loose the devils in the outer
supplies... as I have said before (several times I beleive) is that
the mains transformers (actually most transformers) will not only
transform the main supply through magnetic coupling but also act as an
capacitive divider between the main terminals with your ground as
middle terminal, this will (unless treated) introduce a nice little 60
Hz swing between chassi/secure ground and signal ground. I have seen
this hum being 50 Vrms on a 230V unit (an audio box) so it can be
large. Fortunatly is this a very high-impedance source of voltage, so
treatment doesn't need to be very low-impedance to do the trick.
When you stack your supplies would this hum be connected through the
723/pass-transistor and cause a deviating current (not large). If you
are worried about it you can treat it by applying a resistor or
possibly a capacitor between outer supply ground to inner supply
ground. An resistor in say 10k should certainly remove the effect.
I don't think this is a serious threat thought, but it is in the
non-obvious field.

 S> Since each supply output is rated at 1.5 amps, then I am assuming that I
 S> can provide 1.5 amps at -15v and 1.5 amps at +15v as long as no current
 S> is drawn from the intermediate power rails. I will be using +/-5v for
 S> some things (pot voltages, analog switch power, etc.) so I will keep the
 S> +/- 15v currents probably below an amp each.

 S> My question is about regulation stability - can this be an issue? Each
 S> one of these supplies is a dual 723/pass transistor design. I know that
 S> a single 723 has pretty good stability as far as maintaining a constant
 S> output voltage regulation under varying current conditions, but how
 S> about regulated-regulation, or even regulated-regulated-regulation? Will
 S> stacking these supplies cause any cumulative noise on the power rails,
 S> or any other effect?

If they are non-switching I would not expect too much to happend (just
a small puff of smoke:)

For switching supplies I think it is a diffrent ball-park, since then
you have high frequency components that may use the non-obvious track
to interconnect/intermodulate. There are a switching mode on the 723s,
so it is not totally unrelated. The normal caps should shunt much of
the problems, but I am not as sure with this case than a DC oriented
regulator.

 S> If I wanted to do the same thing with the +5 volt-only supplies, the
 S> same concept would apply, except the total current would be 3 amps and
 S> the number of cascaded supplies would be 6.

 S> I realize that the final, definitive answer to this question would be
 S> easy to achieve via strictly empirical methods (i.e. try it and see!)
 S> but I figured that it would make for some interesting DIY discussion.

I would actually connect up them, measure the voltage between the
ground of each unit relative each of the main leads. I would then
measure the voltage between the grounds of the supplies themselfs.
Then, I would measure the current that occurs when connecting -5V on
one to +5V on another. This could give some usefull info on the types
of aspects of the ground-to-ground introduced hum. If life treats you
good will these be swinging alongside each other, so no major AC
component will exist.

You might want to add +15V to Ground and Ground to -15V decoupling
caps near the supplies, it can't hurt... (ouch!)

Other than that is should be safe unless someone mayed a brain-damaged
solution or (quite possibly) I have gone brain-damaged.

Cheers,
Magnus