ATX supply to bench power
This is a very simple to build and very powerful power supply for basic electronics, but for a newbie can seem confusing if he tries to make the conversion on its own. The power supplies used in PC’s have useful output voltages like 5V, 12V, 3.3V and even some negative outputs -5,-12V. While the positive outputs can handle significant current, the negative rails tolerate just a few hundred milliamps.
I recently needed about 12 and 10A, since my bench supply can handle only 2A I used an old 250W PC source, on the label was 12V/15A but the voltage dropped to 10V at 5A load, so don’t be surprised if you encounter similar “effects”.
Don’t forget the dummy load on the 5V rail, to be honest I never used it and the supply worked just fine, but to avoid any problems load the output with 0.5A. In lack of power resistor use any 12V light bulb, it will do just fine.
ATX power to bench power: [link]
March 14th, 2009 at 10:59 am
The reason your 12V rail dropped so much when you loaded it was because you omitted the dummy load on 5V rail.
PC power supplies only really have one main regulator for all the outputs. When there is no load on the 5V rail it struggles to stop the voltage being too high, so it throttles back leaving the 12V rail, that is loaded, too low.
March 18th, 2009 at 2:03 pm
All the voltages are connected to the feedback circuit, by loading any positive rail, since there is only 1 transformer the supply will work the same.
Altough I will try with the dummy load connected on the 5V just to test your theory.
March 18th, 2009 at 3:25 pm
It is not “my theory” it is a well known feature of single-convertor multi-rail switchers, known as “poor cross regulation”.
The unloaded output voltages will be in the ratio of the transformer turns, but when an output is loaded the voltage drops relative to the others due to the resistance of its secondary winding, the slope resistance of its rectifier and the dc resistance of its filter choke.
I think there are clever tricks using shared magnetics in the filter chokes, which reduce this effect. When the rails have siginificant loads the cross requlation is reasonable, but when a rail capable of 20A is unloaded the effect is extreme. Its output loses a lot less voltage than it would normally because there are no losses. The converter has to throttle back to avoid it being too high, and that means the loaded 12V rail will be too low.
I have personally seen the effect of putting 0.5A load on the 5V rail increasing the 12V output by ~1.5V.
You may need to load the 3.3V rail as well.
March 18th, 2009 at 9:33 pm
Yes that is right, I loaded the 5V rail with 1A load, the 12V then drops about 0.8V. The main idea is that the PC supply labels are “slightly” pumped up.