I want to do more than the usual mods to make this a bench power supply.
This is for a workshop for OlyMEGA, a maker group in Olympia, WA. We have a bunch of 300W ATX power supplies, the majority are Sparkle Power Int'l Ltd FSP300-60ATV.
What I'd like to do: Retune it to put out 5V from the 3.3V output, this will cause the 12V output to go up to about 18V. Swap the diodes on the erstwhile 5V and erstwhile 12V, because the shottky diodes on the 5V line are rated for more current.
In this way the power supply will still have a regulated 5V line, and we'll add an adjustable regulator connected to the 18V, using the -12V (now -18V) to bias the regulator so that it can go from 0 to 14V.
Of course the stock adjustment only goes up a few tenths of a volt. Someone on BadCaps.net put up an unfinished schematic he'd reverse engineered, it appears to combine the 3.3V, 5V, and 12V outputs as part of the feedback. I can hack that. What I'm concerned about is any output overvoltage protection. An attachment about 2/3 of the page down:
https://www.badcaps.net/forum/showth...t=FSP300-60ATV
It uses a proprietary chip (at least I cannot find any information on it), FSP 3528. I did find some generic schematics on a few Russian websites. From those and the partial reverse engineered schematic, I see that 3.3, 5, and 12V lines all go back into the feedback. I should just be able to disconnect all but the 3.3V, but recalculated to regulate it to 5V. The 12V should go up to 18V since it'll be proportional.
But there are two more trimpots on the PCB. One is connected to a 3.3V Sense line, which goes to several transistors which may be part of the overvoltage protection on 3.3V.
I thought I'd swap the 3.3V output to where the 5V is now connected to the FSP 3528, and put a resistor and 3.3V zener from there to the 3.3V over voltage protect on the FSP 3528. If the 3.3V Sense line mentioned above requires some voltage on it, I might connect that 3.3V zener to it, too.
About halfway down the page:
http://www.rom.by/forum/BP_FSP_ATX-4...vyhoda_resheno
More schematics:
http://www.rom.by/files/plata-ShIM-3528.gif
http://www.overclockers.ru/images/la...sp3528_big.png
This one seems to be telling how to mod the power supply to put out 12V and 40V, plus and minus, possibly for a CNC mill or an audio amp? Looking at the picture at the bottom of the page, it looks just like my PC supply.
http://www.overclockers.ru/lab/44201...y_chast_2.html
Maybe all I really need to do is cut the 5V secondary out of circuit, cross jumper the 3.3V secondary to it, swap the 3.3V rectifier to 12V circuit, and add the 3.3V zener from the 5V lines just to run voltage to the chip. Or better, an LM317 adjusted to 3.3V out.
Then it will up the PWM so 5V comes out of the 3.3V winding, which is about (5-3.3)/3.3 = 51% greater voltage, 12V x 1.51 = 18V. Then everything expecting a particular voltage will still get it.
The supply should then be able to output over 20A on 5V, at least an amp on 3.3V, and depending on the linear regulator, 0-14V at up to 20A.
Steve Greenfield AE7HD
This is for a workshop for OlyMEGA, a maker group in Olympia, WA. We have a bunch of 300W ATX power supplies, the majority are Sparkle Power Int'l Ltd FSP300-60ATV.
What I'd like to do: Retune it to put out 5V from the 3.3V output, this will cause the 12V output to go up to about 18V. Swap the diodes on the erstwhile 5V and erstwhile 12V, because the shottky diodes on the 5V line are rated for more current.
In this way the power supply will still have a regulated 5V line, and we'll add an adjustable regulator connected to the 18V, using the -12V (now -18V) to bias the regulator so that it can go from 0 to 14V.
Of course the stock adjustment only goes up a few tenths of a volt. Someone on BadCaps.net put up an unfinished schematic he'd reverse engineered, it appears to combine the 3.3V, 5V, and 12V outputs as part of the feedback. I can hack that. What I'm concerned about is any output overvoltage protection. An attachment about 2/3 of the page down:
https://www.badcaps.net/forum/showth...t=FSP300-60ATV
It uses a proprietary chip (at least I cannot find any information on it), FSP 3528. I did find some generic schematics on a few Russian websites. From those and the partial reverse engineered schematic, I see that 3.3, 5, and 12V lines all go back into the feedback. I should just be able to disconnect all but the 3.3V, but recalculated to regulate it to 5V. The 12V should go up to 18V since it'll be proportional.
But there are two more trimpots on the PCB. One is connected to a 3.3V Sense line, which goes to several transistors which may be part of the overvoltage protection on 3.3V.
I thought I'd swap the 3.3V output to where the 5V is now connected to the FSP 3528, and put a resistor and 3.3V zener from there to the 3.3V over voltage protect on the FSP 3528. If the 3.3V Sense line mentioned above requires some voltage on it, I might connect that 3.3V zener to it, too.
About halfway down the page:
http://www.rom.by/forum/BP_FSP_ATX-4...vyhoda_resheno
More schematics:
http://www.rom.by/files/plata-ShIM-3528.gif
http://www.overclockers.ru/images/la...sp3528_big.png
This one seems to be telling how to mod the power supply to put out 12V and 40V, plus and minus, possibly for a CNC mill or an audio amp? Looking at the picture at the bottom of the page, it looks just like my PC supply.
http://www.overclockers.ru/lab/44201...y_chast_2.html
Maybe all I really need to do is cut the 5V secondary out of circuit, cross jumper the 3.3V secondary to it, swap the 3.3V rectifier to 12V circuit, and add the 3.3V zener from the 5V lines just to run voltage to the chip. Or better, an LM317 adjusted to 3.3V out.
Then it will up the PWM so 5V comes out of the 3.3V winding, which is about (5-3.3)/3.3 = 51% greater voltage, 12V x 1.51 = 18V. Then everything expecting a particular voltage will still get it.
The supply should then be able to output over 20A on 5V, at least an amp on 3.3V, and depending on the linear regulator, 0-14V at up to 20A.
Steve Greenfield AE7HD
Comment