Building a lab power supply from a computer

[bassplayer142]

I found this link and thought it may be of some use to build a power supply. I know that these power supplies can pack around 500W on the right power supplies. Is there anything I should know and be careful of when making this and when playing with smaller electronics?

http://www.wikihow.com/Convert-a-Computer-ATX-Power-Supply-to-a-Lab-Power-Supply

 

[berkeman]

Looks like a fun project, and it's nice that the UL safety stuff is still included in the PS (re)construction.

The main thing I'd caution about (I didn't see it mentioned, but could have missed it) is that PC type power supplies typically have a minimum output current requirement, at least on some of the rails. If the minimum output current is not drawn, then the rail goes out of regulation (typically high). So you will need to check that spec on the PS that you use, and ballast the rails up to their minimum current to ensure that they stay in regulation.

 

[bassplayer142]

Would I have to find information like that online? I'm looking at my power supply in my computer and it has nothing on the sticker about minimum current or anything. And what do you mean by rails and ballast? Are you talking about the different wires that are the same color being connected?

Thanks for the responseAlso, I just found this at the bottom.

"ATX power supplies are switched-mode power supplies (info at http://en.wikipedia.org/wiki/Switched_mode_power_supply); they must always have some load to operate properly. The power resistor is there to "waste" energy, which will give off heat; therefore it should be mounted on the metal wall for proper cooling (you can also pick up a heatsink to mount on your resistor, just make sure the heatsink doesn't short anything out). If you will always have something connected to the supply when it is on, you may leave out the power resistor. You can also consider using a lighted 12v switch, which will act as the load necessary to turn on the power supply."

Would this be implying that the 10 ohm resistor is there to hit the minimum current at all times to safely drive a circuit like you were saying?

 

[berkeman]

Yes, you found the part about ballasting. (Sorry I missed seeing that.)

Ballast in an electrical circuit means an extra parallel resistor on a rail (a voltage output) to sink some extra current from that rail. Switching power supply circuits can be made simpler/cheaper if you know that there will always be a minimum output current, like there is for PC power supplies (they are always connected to the motherboard, for example).

To find out the specifications for your PC power supply, you will need to find a datasheet for it. Hopefully you can use the manufacturer's part number to search for the datasheet. Worst case, you could do some experiments to figure out the minimum output current for stable regulation on each rail (5V, +12V, -12V, etc.). Note that there is some cross-effect in this... Ballasting on the +5V may be all that's required for stability of all the rails, or it may take ballasting on both the +5V and +12V rails.

You will also find that some PC power supplies have minimum output currents of 0. Those are typically the better ones, more expensive, and meant for PCs that can shut themselves mostly off & sleep to save AC Mains power.

 

[bassplayer142]

Thanks a lot for the help. Unfortunately I don't have a junk power supply or anything yet to get started but it is definitely on my list to do.

 

[Proton Soup]

so what's the noise like on switchers these days? from what i remember, there was a lot of high frequency noise that might make them problematic for some small-signal analog use.

 

[berkeman]

so what's the noise like on switchers these days? from what i remember, there was a lot of high frequency noise that might make them problematic for some small-signal analog use.

That's an excellent point. Might be worth adding a couple LDO linear post-regulators to this project to make something like quiet +/-10V rails for analog circuit use!

 

[dinges]

This link may interest you; it describes how someone else guided me through the process of modifying various PC powersupplies to variable voltage (2.5-24V) and variable current (0.1-11A) supplies. TRUE laboratory powersupplies, where you can actually adjust both voltage and set current limiting, as opposed to just slapping on power posts on a plain PC PSU.

http://www.fieldlines.com/story/2007/10/18/16953/116

The next link, esp. post 31 and further, show the schematic of the modifications:

http://www.fieldlines.com/story/2006/4/24/92132/1267

It takes a bit more effort but you end up with a *very* nice power supply that's actually worthy of the name 'laboratory supply'. I've made about half a dozen of such supplies now - proof that it actually works.

Peter.

 

[rolerbe]

One thing to perhaps be a little cautious about is whether the switching PS does or does not have an isolation transformer. If not, then you will have to be careful with insulation for the 'hot' side.

To get isolation, you can add a pair of isolation transformers back to back (e.g. 120 - 12, to 12 - 120).