Change the atx psu voltage reference

[justin66]

Hey guys im stuck with this psu i need to modify it outputs 11.8v and i need somewhere close to 16v so is this possible with this psu? I tried several tips online like locating ka7500 voltage reference with no luck until i came across tl431 but the regulator in the corner changes 3.3v rail so can you please guys help me with this any help would be highly appreciated any info or knowledge you may have, thanks.
This is the pictures of the pcb
http://postimg.org/gallery/1dzwu1que/3434cc7a/

[Paul Price]

The problem here is that the 12V output of the ATX supply in question almost always have output filter caps that are rated no higher than 16V, you would be running near the edge of over voltage on the caps..but just the edge. It is always desired in design of equipment in he holy name of safety and reliability to choose components, especially electrolytic capacitors to run at working voltages some reasonable fraction of their max ratings.

It is certainly sane and cheap enough to find two  or three 25V caps to replace the 16V ones if you want to try this.

You must also be aware that the 3.3V output caps may also be rated at 4.3V and may need to be changed. The 5-V supply, on the other hand will have 10V or 16V caps so they needn't be bothered with, but it wouldn't harm to peek at their dresses to verify this.

The output voltage is set by use of an optocoupler (4 legs) and it can be found bridging the large cap between the output low-voltage section and the AC mains section. It is certainly possible, if not easy enough, by maybe just adding a 4.7 zener diode or else modifying the feedback resistor values  on the 12V side there to set the 12V output to 16V, but don't try going over that. Often there is a LM431 that sets the feedback voltage and drives the optocoupler and there are two resistors on this device that control the feedback voltage.

Some expert knowledge about electronics needed(a knowledge of ohms law, zeners, voltage dividers, how to use a potentiometer to change a voltage setting resistor's value,  knowledge about ooptocouplers and LM431 operation.)

But there are plenty of people who will discourage you on this forum to not to try doing this rather than offer advice on how to do it safely.

Wear safety glasses when testing. Capacitors can shoot off the PCB like little rockets when subjected to over voltage.

..batteries not included, some assembly required.

[bktemp]

The output voltage is set by use of an optocoupler (4 legs) and it can be found bridging the large cap between the output low-voltage section and the AC mains section. It is certainly possible, if not easy enough, by maybe just adding a 4.7 zener diode or else modifying the feedback resistor values  on the 12V side there to set the 12V output to 16V, but don't try going over that. Often there is a LM431 that sets the feedback voltage and drives the optocoupler and there are two resistors on this device that control the feedback voltage.

PC power supplies are a bit different than low power flyback converters.
There is usually no 431 used (except for the regulation for the 3.3V rail because it is generated using the 5V winding).
The KA7500 (=TL494) does all the regulation on the secondary side and has its feedback pins at 1,2 and 15,16. One of them is connected over a voltage divider to pin 14 (=5V reference) and another pin goes to the 5V and 12V rail.
You need to change one of the voltage dividers.
Most PC power supplies also have an additional overvoltage protection circuit shutting everything down if one of the rails gets too high or the rails get unbalanced. You also need to adjust this circuit.
Google should find many schematics. All of them are similar but not identical. But you can use them as a reference to get the basic idea how the power supply works.

[Paul Price]

There are many different designs for ATX P/S's, but having taking apart many and see that many of them are using a LM431 coupled to an optocoupler to the mains side for feedback, but this might be only for OV protection.

There might always be low-side controller chips that provide themselves with a must needed voltage reference pin and are monitoring and correcting against their own refererence. But I agree, the 12V setting could be accomplished by modifying the right voltage divider  resistors at the controller chip.

[justin66]

The problem here is that the 12V output of the ATX supply in question almost always have output filter caps that are rated no higher than 16V, you would be running near the edge of over voltage on the caps..but just the edge. It is always desired in design of equipment in he holy name of safety and reliability to choose components, especially electrolytic capacitors to run at working voltages some reasonable fraction of their max ratings.

It is certainly sane and cheap enough to find two  or three 25V caps to replace the 16V ones if you want to try this.

You must also be aware that the 3.3V output caps may also be rated at 4.3V and may need to be changed. The 5-V supply, on the other hand will have 10V or 16V caps so they needn't be bothered with, but it wouldn't harm to peek at their dresses to verify this.

The output voltage is set by use of an optocoupler (4 legs) and it can be found bridging the large cap between the output low-voltage section and the AC mains section. It is certainly possible, if not easy enough, by maybe just adding a 4.7 zener diode or else modifying the feedback resistor values  on the 12V side there to set the 12V output to 16V, but don't try going over that. Often there is a LM431 that sets the feedback voltage and drives the optocoupler and there are two resistors on this device that control the feedback voltage.

Some expert knowledge about electronics needed(a knowledge of ohms law, zeners, voltage dividers, how to use a potentiometer to change a voltage setting resistor's value,  knowledge about ooptocouplers and LM431 operation.)

But there are plenty of people who will discourage you on this forum to not to try doing this rather than offer advice on how to do it safely.

Wear safety glasses when testing. Capacitors can shoot off the PCB like little rockets when subjected to over voltage.

..batteries not included, some assembly required.

Thanks for all the safety precausions its always useful but why on the world would you suggest me not to do it i know what i want to do and i will do it anyways with or with out you i know you might think this newbie will kill himself but you know im not that dumb not to short input caps anyway can can if you know how to do it can you provide me some of your knowledge?

[justin66]

There are many different designs for ATX P/S's, but having taking apart many and see that many of them are using a LM431 coupled to an optocoupler to the mains side for feedback, but this might be only for OV protection.

There might always be low-side controller chips that provide themselves with a must needed voltage reference pin and are monitoring and correcting against their own refererence. But I agree, the 12V setting could be accomplished by modifying the right voltage divider  resistors at the controller chip.

Yea there is one part that looks like opto coupler but i cant find any info about it has this written on it
B0541
817b

[justin66]

Guys its pretty weird i tried to modify voltage ref from the datasheet but its quite odd that ka7500 has pin14 shorted with pin 13 you can see that on top pin row right side

[bktemp]

That's normal, because this enables the push-pull mode. Follow the connection from pin 14 until it ends at pin 15 or pin 2. Typically there is a voltage divider to set the reference voltage for the comparator to about 2.5V. You can change this divider, but it is better to change the divider from the 12V rail to pin 1 or 16.

[Paul Price]

Here's the datasheet for the 817 optocoupler.

[station240]

You could rewire the transformer taps to put the 12V and 3.3V taps in series, downside is it means disconnecting the transformer's common negative. Which is that big cable coming out the top, to find which strand serves the 12V rail. Then connect that to the 3.3V transformer tap that feeds the diode(s). Should output 15 to 16V.

[justin66]

That's normal, because this enables the push-pull mode. Follow the connection from pin 14 until it ends at pin 15 or pin 2. Typically there is a voltage divider to set the reference voltage for the comparator to about 2.5V. You can change this divider, but it is better to change the divider from the 12V rail to pin 1 or 16.

Man thanks so much i have no clue how can you guys be so bright anyway i just put pot between pin14,15 to pin 2 and the max output is 14.4 i wouldnt go ovet 15.5 but is it possible to increase the voltage just a little bit more? Oh and pin 1 isnt connected and pin 16 is going straight to ground before the small electrolitic cap.
Should i remove resistors in between pin14and 2? Btw i used voltmeter to match the resistance between those pins and for some reason psu was making coil whine up untill 13.5v. Since i dont have osciloscope im conserned

I noticed one thing when i power it up it starts with 16.5v and than it drops to about 14.5v what could be causing this?