old PC PSUs (ATX) as constant current source

[Mathias]

Dear ng,

I've seen people using old PC ATX PSUs as a 12v lab PSU.
Now I need a 20V 5A source and thought about using 2 ATX
PSUs in series (can't use a single one with combined
+12v and -12v since all units I know have >10A for +12V
but only <1A for -12V).
I don't exactly know how the switching power supply voltage
regulation works, so would you expect this setup to work?

Thanks!
Mathias

 

[Mathias]

nevermind the original subject, of course it should be
"constant voltage source"

 

[Tim Williams]

nevermind the original subject, of course it should be
"constant voltage source"

Oh, ok. It uses a voltage feedback circuit, as you might've guessed.

The problem you're going to have is breaking the ground on one of the power
supplies.

Also, since you just want 10V per (and no other voltages), I suggest tracing
the output circuit (everything past the power transformer) and removing all
unnecessary circuitry: namely, the 5V, 3.3V (if present) and negative supply
circuits.

But yeah, that ground is going to be a problem.

Tim

 

[William P.N. Smith]

I've seen people using old PC ATX PSUs as a 12v lab PSU.
Now I need a 20V 5A source and thought about using 2 ATX
PSUs in series

This will work with a couple of caveats.

First, the output voltage is referenced to ground, and the grounds are
tied together at the AC ground, so you'll need to lift at least one of
the safety grounds (and ensure that AC leakage currents don't cause
you problems.)

Second, you'll have to develop your own constant-current source from
the resulting 24V supply, as the current limiters in the PC supplies
are most likely foldback types.

Third, 12V voltage regulation usually depends on the 5V rail, so
unless the 5V load is some minimum value, the 12V output may not be in
regulation (and worst-case the supply may crowbar and shut down).

 

[Chris Jones]

This will work with a couple of caveats.

First, the output voltage is referenced to ground, and the grounds are
tied together at the AC ground, so you'll need to lift at least one of
the safety grounds (and ensure that AC leakage currents don't cause
you problems.)

Second, you'll have to develop your own constant-current source from
the resulting 24V supply, as the current limiters in the PC supplies
are most likely foldback types.

Third, 12V voltage regulation usually depends on the 5V rail, so
unless the 5V load is some minimum value, the 12V output may not be in
regulation (and worst-case the supply may crowbar and shut down).

I would suggest that you do NOT lift the ground to the metal box but instead
open it up and just disconnect the 0 Volts side of the DC output from the
casing. This way at least the box is still grounded and can't give you an
electric shock. The transformer insulation might still not be as safe as
it would need to be for a power supply where the output is intended to be
floating, I'm not sure about that.

It might be worth trying to get your output from just one ATX supply. It
might be possible to just reverse the rectifiers and capacitors of one of
the output rails, (and modify the feedback circuit) so that instead of +5V,
20A and +12V, 8A, you might be able to get +5V, 20A and -12V, 8A, giving
you 17V at 8A available between the two outputs. If you only load it to 5A
max then it should be possible to increase the output voltage a bit also,
though I think tweaking from 17V to 20V is a bit ambitious.

Chris