Force Computer ATX Power Supply to turn on Without Motherboard

[teroenza]

I'm trying to re-purpose an old Compaq (and another HP) ATX computer power supply to use as a lab PS. I've seen tons of articles on line about how to do this, for example.

I can't get it to even start up though. I have the power-on pin from the ATX connector that goes to the motherboard shorted to a ground wire, and have tried various loads ranging from 10 ohm power resistors to just plugging in an old CD drive to one of the molex connectors.

Does anyone have an idea of what I might be missing?

 

[Windadct]

There is a "PS on" lead that needs to be jumpered to common ( Gnd) - I think 14 to 15 on the larger ATX plug

 

[teroenza]

I have that one shorted to common already. Are you talking about the PS_OK wire at pin 8?

ATX Pinout

 

[meBigGuy]

+5VSB should be at 5 volts even before you ground PSON#. It is the standby power used by the motherboard to power the circuits that ground PSON#.
If it is not at 5V then something is seriously wrong.

I don't know if this is happening, but some switching power supplies will not regulate without a minimum load. It may the crowbar the supply (shut down). If you see the voltage spike and then shut off, that could be happening.

I'd only suspect that once I was sure that the connections were correct for it to be turned on and 5VSB was at 5V before you grounded PSON#.

 

[teroenza]

One of them has a voltage spike, and the fan does a 1/4th turn before stopping every time I switch it on. I am getting the +5V to the motherboard from both of them. It must be the load. Apparently a DVD drive is not enough of one.

I've tried various loads across one of the SATA power outputs, and nothing has worked. Perhaps it is looking for a spesific load on one of the motherboard pins.

 

[meBigGuy]

It could be a bad supply. Was 5VSB at +5V before you grounded PSON#.

BTW, The minimum loads can be quite substantial.

For example: http://www.antec.com/pdf/article/info_DIYArticle1.html (I had to select the tables to see the text)
Antec tests its power supplies at a minimum load of 10%. For a 420 watt supply that comes out to 3.5A at 5V, 2A at 12V, 3A at 3.3V and 2A at +5VSB. 1A at -12V and 0.5A at -5V.

I would suspect that adequately loading the 5V and 12V would get the supply to work (but, take note ---- I'm really just guessing)

One comment from the article you posted a link to originally was:
PaulG14 -> RijeeshP
" Sound like you have no load. ATX power supplyes need to have a constent load on them, somthing like a 10 ohms 10watt power resistor. This in normaly placed between the 5V rail and ground. "

You should google around and see if there is any consensus on this.

 

[Baluncore]

With the earlier AT supplies only the +5V was sensed to regulate the output voltages. Because other voltages were set by turn ratios of the +5V rail, it was necessary to draw a current from the +5V sufficient to maintain the other loaded outputs.

The fan running on the +12V output would only do a part turn if there was insufficient +5V current being drawn.

If you do not need +5V but want +12V, then consider pulling out the +5V rectifiers and deriving a fake +5V sense from the +12V output by using a resistive divider.

 

[teroenza]

I've tried resistors up to 100k. I might try loading the ATX motherboard pins themselves instead of the SATA/ATA ones to see if it looks for a motherboard too.

 

[Baluncore]

The brown “+3.3 V sense” wire suggests that at least the +3.3V rail must be loaded and connected to the sense line.
With ATX you will possibly need to load both the +5.0V and the +3.3V rails.

Resistance in ohms is voltage divided by current in amps. R = V / I
Resistor power in watts is voltage multiplied by current in amps.W = V * I

To load the +5V rail with 1 amp, use 5 ohms rated at 5 watt.
To load the +5V rail with 2 amp, use 2.5 ohms rated at 10 watt.

To load the +3.3V rail with 1 amp, use 3.3 ohms rated at 3.3 watt.
To load the +3.3V rail with 2 amp, use two 3.3 ohm resistors in parallel rated at 3.3 or 5 watts each.

100k is way too much resistance resulting in way too little current.
Get the idea ?

 

[teroenza]

I had tried the 100k because I had a pot on hand and didn't care if it burned out. I'll try to get some power resistors and load both the 5V and 3.3 rails.

 

[meBigGuy]

the 3.3V sense line is optional.

I expect that 10 ohms 10 watts connected to +5 will work since that's what someone suggested in the comments of the original article you posted a link to.

That's probably the only resistor you need, but it wouldn't hurt to have some others on hand. Especially if you may be doing this again someday.