Bench Power Supply from old Compaq PSU

[Jump]

I have had a lookup about building a Bench supply from an old PC PSU. I read that for an ATX supply a load is usually required for the switched mode PSU to work, but this is not usually required for older AT supplies.

Anyway I have a PSU from an old Compaq Deskpro machine I would like to convert, it has a nice shaped case for a bench supply with a sort of stepped front where connectors can be placed. It appears to be an 'AT like' supply, with an external power switch, higher current/power +5V output than +12V and a -5V output, though the motherboard connectors are neither AT nor ATX standard, but that's Compaq for you. However the power switch isn't the standard AT version between supply and PSU but comes from the PSU circuit board with AC connected directly to the circuit board. I have connected 10Ω 10W resistors between both +12V and +5V and ground to give it a load in case it is needed however the PSU fails to spin up. The fan 'kicks' once and then no more.

The only wires currently connected are the +5V and +12V to Ground via the resistors.

Marked Output ratings, from the label, are :

OUTPUT:

244W MAX

+5V 32A +12V 6A -12V 0.9A -5V 0.2A

OR

+5V 15A +12V 6A -12V 0.9A -5V 0.2A +3.3V 25A

The wires on the circuit board that went to MB connectors are labeled as :

black ground
red +5v
orange +12v
blue -12v
yellow -5v ?
brown +3.3v
purple nic - led?
white/red hd - led?
white/orange pwr - led?
white fan sw
white/brown 3.3v S+
white/black 3.3v S-

As well as these there are the PSU fan, 110V/220V selector, AC power and the power switch conectors. Those marked 'led?' connect next to 3 small LEDs on the edge of the circuit board and look like they may be indicators as they poke through the side of the PSU case.

Does anyone have an idea what of the other wires may need connecting to where to get the PSU to fire up correctly ?

If there are no suggestions I'll just try a few and see what happens, since it's not of use now then there will be no loss if I blow it ...

[Bill_Marsden]

No way to be certain, but it sounds like you need a minimum load somewhere on one of the outputs. That or the PSU is bad.

[GetDeviceInfo]

See if this helps;

http://www.allpinouts.org/index.php/...N_Power_Supply

[kkazem]

Hi Jump,

I've got to agree with Bill in this case. I'm a professional power supply design engineer for over 35 years now designing switchers and linears. There are actually at least 3 possibilities. One, that one of the connector pins is for remote sensing lead and since the connector isn't plugged into the mother board, there is no voltage to remote sense. This would likely cause the outputs to overshoot and make the OVP (Over Voltage Protection) circuit kick-in to shut down the supply. This will be easy to see if you have access to a scope. If not, if you have access to a peak-hold DMM that might do as well. Two, lack of minimum load (usually on whatever is considered the main output, which may or may not be the +5V), causing too little duty cycle and the unit can't sustain it's own primary bias requirements and hence, shuts off. Three, you just may have a bad supply. If you still have the Compaq PC, plug it into that and use a DMM to check the outputs. Make sure to check all of them if you're going to go thru the trouble of doing that.

If the supply is good, you'll have to find the secondary-side main output voltage sense & feedback circuit, usually a TL-431 connected to an opto-coupler. The TL-431 is often in a TO-92, 3-pin small plastic transistor looking package and is a programmable shunt regulator (think of it as a programmable zener diode). The opto-couplers are nearly always a 6-pin DIP (dual-inline package) package. Instead of a TL-431, it could be driven by an op-amp circuit. You'll need to find the output voltage sense resistors going to the TL431 or op-amp, and change the voltage divider circuit to an adjustable one by replacing the bottom resistor with a pot of the same or similar value. You want to make sure that you're only messing with the isolated, secondary circuitry and not with the primary, off-line high-voltage side. Check this with an ohmmeter on the highest scale between one of the two prongs of the AC input and anywhere on what you think is the secondary circuit. If you read anything less than infinity on the 20 MEG OHM scale, you are on the dangerous side and subject to personal and property damage and maybe even death. The rectified, filtered and voltage-doubled input (AC) side of a modern switching power supply has a DC (yes, DC) voltage of up to 385 VDC, which can kill you without giving you a second chance! Anyway, continuing on, make sure you design the voltage divider with the new pot (panel mounted for a lab supply) that will not try to increase the output voltages any higher than they are already. In other words, you can safely turn-down the outputs, but do not turn them up any higher than they are or you could easily ruin your new lab supply. Also, since the outputs are almost always cross-regulated by multiple output windings from the same power output transformer, all outputs will sort of track each other when you move the pot. You cannot control them independently without adding lots of extra circuitry and if you're going to do that, I would highly recommend that you start the project from scratch rather than use a PC power supply as the starting point.

Good luck and feel free to contact me with any additional questions.

Regards,
Kamran Kazem
kkazem

[judge_jamin]

I have converted a circa '99 compaq psu for bench use, although dummy load resistors are connected +12v to 0v and +5v to 0v, the supply will not provide more than +5v on the +12v rail and all other rails remain dead, unless the white (on my psu) wire labled "rem" on the circuit board is connected to +3.3v (Brown on my psu) I have made this connection with a switch and all voltages are presented via banana posts for easy connection on the front of a custom enclosure.

Once the 3.3v to rem connection is made, all rails on my psu power up to their designated voltages. I can't comment on the safety of this design, my psu mods are best guess, experience and common sense driven. At the end of the day, my modified psu will never be left powered up unattended for extended periods, but it serves a (cheap) purpose.