Using a Computer PSU

A remarkably cheap high-capacity power supply

Computer power supply units (PSU) output large currents at +12V, +5V, +3.3V and smaller currents at
bias voltages -12V and, formerly, at -5V. They are small, readily available, and inexpensive compared to
laboratory supplies. Obsolete versions are still very useful as power supplies, and can be obtained at very
low cost. The one I adapted as a laboratory supply here cost only $10, but it supplies +5V at 32A, +3.3V
at 24A, and +12V at 6A, in addition to -5V and -12V at 300 mA.

Computer power supplies were designed for that purpose, not as a laboratory supply, so if we use them
for this purpose, we must be sure to provide a proper environment. They are switching supplies, which
always demand a minimum current output. In a computer, this load is always connected, but in a
laboratory supply it is not, and we must provide it. Some switching power supplies (especially those
intended for laboratory use) have an internal load. A computer PSU cannot be assumed to look into an
open circuit, unless you have the correct specifications, and they say so. A modern ATX supply has
minimum loads of about 1A on the +12 supples, 0.3A on the +5V, and 0.5A on the +3.3V. The -12V and
the +5V standby ouput have no minimum current. These figures can be used as a general guide if the
exact specifications are unknown.

I used 12 , 3W sand power resistors in parallel as a load for +5V and +3.3V, which gives currents of
0.83A and 0.55A. These currents will not be missed. For +12V, I used a 20 , 10W sand power resistor,
which gives a current of 0.60A. These resistors were soldered to 7-terminal tie point strips (also available
from All). The power dissipation is less than the ratings, but the heat will be present, so the resistors
should be supported in the air and not touching anything.

My PSU is a Teapo TP-511B-4, from All Electronics, made in Sep. 1994 (before the ATX specifications).
The power supply came with a main output connected to a 16-pin plug, a drive ouput to two large (HDD
and CD) 4-pin connectors, and a drive output to one large, and one small (floppy) connectors. The wire
colors were similar to those used in ATX supplies, but not completely. The five black wires were ground
(earth), and the four red wires were assumed to be +5V. Three wires of a dark color, perhaps grey,
indicated a high-current output, since the current maximum on one wire is generally about 8A. This could
only be the +3.3V. The black wires, red wires and grey wires showed zero resistance between each wire
of a group, so we can assume that each color of wires is equivalent. A red wire and a grey wire were used
to connect with the load resistors, as well as black wires for earths. This left only four wires. The yellow
wire was probably +12V, as yellow is generally used for this voltage. It was connected to one end of a
20 bleeder resistor, with a black wire on the other terminal. The blue and white wires were then the bias
supplies; one was -5V, the other -12V. We can wait until the supply is tested to find out which. The
remaining wire, orange, could well be a Power Good signal. This color is used in the ATX standard for
3.3V, but apparently not here. Three red wires, two black wires, two grey wires, and the four remaining
wires, were brought to a 12-position "Eurostyle" terminal strip. I used the smallest strip (7 Amp), but if I
had it to do over, I would use the next larger size, 20 Amp, as more suitable for the large wires.

Everything was mounted on a 14" length of good-quality 1 x 10 board. This makes an excellent
breadboard, and can be finished attractively if you prefer. The PSU rested inside 6 small corner braces,
the ones with one hole in each leg. 4 braces at the corners located the PSU, and a brace on each end kept
it from moving sideways. I did not dare to drill and tap holes in the PSU for attaching it to 4 braces, not
knowing what I would hit inside. The PSU is free to be lifted out. Wire over the top could be used to hold
it down, but I did not think this was necessary.

AC power is supplied through an IEC (SPH-386) detachable power cord. Be sure the voltage selector is
set to 115V (or whatever your voltage is!). A push-on, push-off switch at the end of a cable coming out of
the supply turns it on and off. When the PSU was turned on, there was no smoke or noise. The DMM
verified the voltages on the several wires, which were as expected. The blue wire was -12V, the white
wire -5V. The orange wire showed 3.3V, so it is probably the PGood signal. The actual voltages were
5.17, 3.31, 12.19, -4.99 and -11.98, and they were steady. I did not do a load test. The supply draws a
maximum of 8A from the 115V line. Adding up all the maximum power outputs, one gets 319.5W,
though the supply is rated at 222W. In laboratory use, however, the load will never approach these values.

Modern ATX supplies have a 24-pin main output connector (previously 20-pin). Pin 16, a green wire, is
grounded to turn the supply on. Pin 8, a grey wire, is the PWR_OK signal. The 3.3V is on orange wires.
Pin 13, an orange wire, is used to sense the 3.3V at the point of use so it can be regulated to the proper
value. The purple wire at Pin 9 supplies the standby +5, which is always on. No load resistor is necessary
for it. The tendency in ATX power supplies is to increase the output at +12V (there are even two
independent sources, each of which requires a minimum load) and decrease the load at +5V. The -5V bias
supply is also disappearing, since newer chips do not require it.

References

For excellent information on ATX power supples, including pinouts and use, see
[Link] Links to power supply specifications
(form factors) are included. There is even a section on the subject of this page.

My PSU was obtained from All Electronics, [Link]. Phone orders at 1-800-826-5432.
Cat. # PS-511, $10. All components used here were also available from All.