Thanks for all the replies.
Post by Jeff Liebermann Post by Steve Pope
Suppose a power supply for a desktop PC has four 12 VDC outputs,
each rated at 20 amps, with an overall rating of 800 watts.
Can one safely gang together these four outputs to create
an ~66 amp power source (assuming sufficient connector ratings
and wire gauges)?
No, or rather not easily. The problem is that each power supply has
its own built in voltage regulation system. When you parallel the
outputs, you are putting the regulators in parallel. Since they
cannot be identically set, they will fight each other for control. I
con't predict what will happen, but it probably won't be anything
useful or desireable.
Yes, that was my suspicion.
Post by Jeff Liebermann
You could isolate the outputs with four Shottky diodes, one for each
power supply. They would need to be power diodes and properly heat
sinked. You would also need to carefully adjust each power supply so
that each provides its share of the total current. That's not easy
with commidity ATX power supplies. If you measure the current from
each power supply, I suspect the currents will be quite different,
especially if the power supplies are not identical. There are also
some response time issues caused by the diodes.
Incidentally, don't believe the nameplate ratings on most ATX power
supplies. Your 800 watt rated power supply will probably not be able
to produce 800 watts and certainly not with only a single tiny fan.
I've tested various lesser (about 350-350 watt) power supplies using a
resistor load rack and a bucket of water and found all of them lacking
near their maximum output. A few couldn't even produce 1/2 the
or possibly using an automobile battery for the 12V.
My goal here is to create an inexpensive supply/inverter combo
that will convert 240 VAC, 50 Hz to 120 VAC, 60 Hz, with
perhaps 600 watt capacity.
You'd think that would be an easy problem to solve, but not
at low cost.
For the inverter, the 12V input models are the most inexpensive
due to their being high volume since people run these of
car batteries (or marine etc.) Other things being equal,
I'd prefer 24V or 48V input but those seem to cost significantly
Whereas for the power supplies, the PC power supplies seem
to be the least expensive.
Julian states the four 12V outputs might be wired in series
so maybe finding a cheap-enough 48V-input inverter is
For paralleling the outputs, one idea I had for isolating them
is to insert 0.027 ohms of series resistance in each output.
At full power the resistor would dissapate under 10 watts,
and there would be up to 0.5V of voltage drop. The cost of
film resistors (Panasonic) that could do this is $3.60 (in
low volumes) but there would be non trivial packaging design/cost
(either a board or a wire harness). One may also want a bleed
resistor in there so that there is always some drop across the
I had not thought of the Shottkey diode approach, that seems
The approach of using a car battery.(and trickle charger) is
would do just fine for low-duty-cycle usage.
It seems the price of 12V DC supplies jumps up above 300W or so.
I'll keep looking, thanks everyone.