Hacking an ATX power supply
by Antony Chazapis
Making an ATX power supply work at 13.8V has been thoroughly discussed. I found this implementation very interesting. Some weeks ago I got myself an old unit and decided to hack around and see what happens. The basic idea is to find the pin on the regulator that reads back the voltage from the output. Then, insert a trimmer before the regulator, in order to lower the feedback voltage, so the regulator is tricked into raising the output to get back the required reading. This power supply uses a regulator marked as 2003 – a mystery IC with no data sheet available. Luckily, I found the schematic online.
The regulator reads back the voltage from both the 5V and 12V rails at pin 14. Resistors R38, R39, R40, R60 and R62 divide the voltage so that the pin 14 is at 2.5V. I desoldered R62 and replaced R60 with a 100K trimmer. I powered it up and after some tries I managed to get 13.8V out. The problem is that the regulator has overvoltage protection and will shut down the supply if the output exceeds approximately 14V. So I had to find a trimmer setting that would allow the power supply to turn on, and then slowly adjust it to the required voltage.
Next, I turned over to the output rectifiers. There were three of them: a large one for 5V and two smaller ones for 3.3V and 12V. The power supply is rated at 40A for 5V and 17A for 12V. One method to get more power out the 12V (now 13.8V) rail is to switch the 5V and 12V rectifiers. The ultimate goal would be to completely get rid of all 5V output components, so I started by desoldering the corresponding rectifier. The power supply would power up and then immediately turn off, indicating that the regulator had some kind of 5V sense input. Or at least this is what I thought looking at pin 4 in the schematic. But ATX power supplies also provide 5V standby voltage from another source (here from U2). So I took the 5V standby output and connected it to the 5V output. It powered up, worked for some seconds and then made a short, high-pitch noise and shut down. I soldered back the rectifier and the R60 and R62 resistors with no success. Again, it would power up and promptly shut down. Close inspection did not reveal any damaged components, so I just decided to call it a day. I kept any parts that seemed useful.