variable atx power supply

Thread Starter

lokeycmos

Joined Apr 3, 2009
431
i found a really cool video on youtube about a guy that built a variable atx power supply. he doesnt want to share the details. so i am on a mission to figure it out with your help. http://www.youtube.com/watch?v=3LBsFrbPzb8


this is obviously not your "typical atx conversion" with fixed terminals or an added on regulator. this guy actually went into the circuit and added on the pot internally. i would like to know where this pot gets connected up to. this project is using the built in regulating circuitry rather than all the "how to" videos where people add on a regulator such as a lm317. this is backed by the fact that an 18 amp regulator would need a beefy regulator and some beefy transisitors. im not concerned with the digital meter. only the variable part. TY
 

JingleJoe

Joined Jul 23, 2011
186
My first thought is that there may be some sort of voltage regulators inside the ATX PSU and I've seen set voltage regulators made variable by the addition of some simple circuitry ... but taking it that far, one may just want to build one's own PSU.

If It's as simple as replacting the right resistor inside with a pot that would be fantastic, but things are never that easy ... or are they? ;)

I'd love to know how he did this too.
 

tom66

Joined May 9, 2009
2,595
Here's my video:

http://www.youtube.com/watch?v=mK_rIs7uFlo

The modification is very simple.

First locate the TL431 IC. It could be KA431 or CD431 or any combination - but it looks like a small transistor with three legs and has "431" in the name. Most power supplies have one. Some do not; in those power supplies it is more difficult to mod them. There are sometimes two TL431's, one for the 5Vsb and one for the main rails. You want to avoid adjusting the 5Vsb as you could damage the controller IC.

The TL431 is a voltage reference. It has a REF pin with two resistors attached. The mod is to solder a potentiometer across all these resistors including the wiper connector of the pot to the middle of the two resistors which is connected to Vref. You can then change the reference voltage and thus change the output voltage. Usually the range is a little lower than the nominal voltage to about double the nominal voltage. You should probably consider changing the filter capacitors as they will fail under double output voltage. A 1k to 5k pot works best; higher values may not be able to "pull" the other resistors, lower values would dissipate a lot of power in the pot (which is directly across the 5V rail.)

Also consider current limiting may not be functional at higher voltages. You could draw 12V @ 15A normally but drawing 24V @ 15A would put double the load on the primary side components, and might lead to failure.

Please please cut off the motherboard connector so someone doesn't plug this into their computer after the mod is completed.
 

PackratKing

Joined Jul 13, 2008
847
It is totally workable...........my main bench power supply is wired in a similar manner, tho' I employ 4 317's in paralell on a rather large heatsink for better amperage.

Granted I'm rather severely bending some physical law to pull it off, and routinely draw 8 amps Max for short times, and it doesn't even break a sweat
 

Thread Starter

lokeycmos

Joined Apr 3, 2009
431
ok cool, i found the AZ431 transistor on the board. there is only 1. i posted a couple pics. could you please tell me where exactly i would solder on the pot? thank you very much! im excited to get this going!
 

Attachments

SgtWookie

Joined Jul 17, 2007
22,230
First, clean all of the dust off of that filthy thing! Use a can of compressed air aka "Dust Off" or the like.

The AX431 is a shunt voltage regulator; probably used as a voltage reference.
Here is a datasheet:
http://bbs.dianyuan.com/bbs/u/77/1749921243643062.pdf

These regulators are produced by a number of manufacturers with different prefixes and suffixes.

But, tell us what other ICs are on the board. I'm talking about DIP and/or SMT IC's.
 

Thread Starter

lokeycmos

Joined Apr 3, 2009
431
well i dont have compressed air so i tried wiping it off. here is a pic of the transistors and ICs on the board with labels. i took a bottom pic too. looking from the bottom the fuse is on the top left.
 

Attachments

SgtWookie

Joined Jul 17, 2007
22,230
I think that this is basically the same as your supply:



If that's the case, the 431 on your board is to sense the 5v standby. Leave it alone.
The STC945's are NPN transistors.
I believe the WT7514L is similar to the LPG-899.
Datasheet: http://www.rom.by/files/WT7514LV130_datasheet.pdf
Note VR1 in the example schematic on page 8.
Basically, it's the same as R38 & R40 on pin 16 of the LPG-899 in the schematic that I posted. R38 and R40 in parallel should be 30.82k Ohms. Decreasing that resistance from pin 16 to ground will increase the output voltage. You might consider replacing those two resistors with a 20k fixed resistor and 50k pot in series.
 

Thread Starter

lokeycmos

Joined Apr 3, 2009
431
ok well, i messed someting up. yes wookie ur schematic is correct and the two values of r38 and r40 are correct. i unsoldered them very easily(it was very easy to get to on the edge of the board) i soldered 2 wires in place of one resisitor(as they are in parallel) i used a 20k resistor and a 50k pot. i failed to measure the total resistance before i turned the supply on. turns on for 2 seconds and shuts off. i replaced the origional resistors and same thing....only runs for 2 seconds. not sure if something got fried or what. im open to suggestions and input. i dont give up easily, id like to perfect this, but it may take a few PSUs to get it right! one more question; which rail does this vary? ie, 12v, 5v, 3.3v? ty
 

SgtWookie

Joined Jul 17, 2007
22,230
Well, you might've caused a cap to short due to overvoltage. I suggested 30.82k as the starting value. Too bad you were in a hurry.

Don't know what to suggest offhand, except look very carefully to see if you lifted a trace, or shorted something.
 

tom66

Joined May 9, 2009
2,595
It's best you look out for a power supply with a TL431 instead of a dedicated IC as the 431's are far easier to modify. They'll usually use two TL431's, one for the standby and one for the main supply.
 

SgtWookie

Joined Jul 17, 2007
22,230
Well, his did have a '431 - but just for the standby supply.

I hadn't heard of either of the switching regulators/voltage monitor combinations that were posted; it took a bit of hunting to track down any info on them. I actually found the reference to the schematic on a Russian electronics site.
 

Jotto

Joined Apr 1, 2011
151
well i dont have compressed air so i tried wiping it off. here is a pic of the transistors and ICs on the board with labels. i took a bottom pic too. looking from the bottom the fuse is on the top left.
Run it through the dishwasher. I used to run all my boards for CRT units through a commerical dishwasher here at work. Works wonders, but of course you must allow the unit to completely dry.
 

SgtWookie

Joined Jul 17, 2007
22,230
Run it through the dishwasher.
They frequently do the same thing at electronics manufacturers; insert the completed boards into something that resembles a large commercial dishwasher and let the suds & hot water fly.

But, manufacturers also have ovens (and vacuum ovens) to dry out their assembled boards. Just to emphasize it more, you really need to make quite certain that the board is dry. One way to help that along is to place the board in a seal-able plastic bag that also contains a goodly amount (a cup or two) of raw uncooked rice. The rice will absorb moisture, acting like a dessicant.
 

Jotto

Joined Apr 1, 2011
151
I would always allow mine to dry over night. Might try the rice idea, it would be better then what they cook here.

I hate rock rice.
 

SgtWookie

Joined Jul 17, 2007
22,230
LOL!

Just to make sure, I'd leave it in dessicant for a couple of days minimum. No sense in hurrying things; if you aren't sure it's dry you may very well fry it.
 

atferrari

Joined Jan 6, 2004
4,764
Somewhere in the Web there was an article explaining, besides the actual washing, the complete process to dry the PCB and components. It was, as far as I recall, based in long time exposure to temperatures below 100ºC. Long time like more than 48 hours IIRC.
 

SgtWookie

Joined Jul 17, 2007
22,230
The TL494 is very popular.

Pin 3 is the feedback input on the TL494. The circuit you linked to is very different from a power supply; it is a current regulator for a motor.

I managed to find a schematiic for you last time, but you hurried through it and burned up your supply. You should be more careful and systematic this time. Get to know the circuit inside and out, and understand what the parts are for before you hack this one.
 
Top