Computer Power Supply 12v to 24v identify blown components

[natester]

Hi all, I tried to convert this power supply from 12 to 24v.

I identified the tl431 and associated voltage divider. Strangely the resistor values didn't match their measured resistance, but I went with the measured values and added a resistor R2 to bring the REG input to the tl431 down by 1/2

Thinking everything was good, plugged it in and got sparks an fire! Woohoo!

The problem is it looks like 2 caps are basically gone now (blown away), something that looks like a transistor and something else I don't recognize. Here are the photos:

Missing cap 1:


Missing cap 2:


Blown transistor-looking thing:


No idea what this is:


More photos of the whole device here: https://goo.gl/photos/xoWdZnFKWV6Agw2h8

So my question is, can this be fixed? Is it worth it? I already removed the resistor I added, and my plan is to give it another try if I can get the power supply running at 12v again first.

[Seekonk]

If it takes two or three power supplies to learn something, that is OK.  On the output of the supply there are electrolytic capacitors. It would be worth looking at the voltage they are rated for in relation to the voltage you tried to apply.  It is likely they shorted or went into an over current condition.

[natester]

Sweet, I'll check those. Do I need to replace the blown components as well?

[keymaster]

the unknown part must be an optocoupler . what optocoulper is another story .

Do I need to replace the blown components as well?

i am afraid ,that you do . Everything in there is essential for the operation of the psu.

[natester]

Could the transistor be an NPN 331 transistor? http://www.nteinc.com/specs/300to399/pdf/nte331.pdf

Also, does it matter what kind of caps I replace the missing ones with? Something that can handle more than the output voltage I assume? Or can handle in the input voltage?

[Armadillo]

Could the transistor be an NPN 331 transistor? http://www.nteinc.com/specs/300to399/pdf/nte331.pdf

Also, does it matter what kind of caps I replace the missing ones with? Something that can handle more than the output voltage I assume? Or can handle in the input voltage?

Also need to watch the primary side components rating if you intend to ratio up the secondary.

[jeroen79]

Perhaps instead of going for doubling the output on the first try you should better modify it so you can slowly crank up the output.
That way you can spot weak points before they break.

[james_s]

It's probably going to be easier to throw it out and get another one than to try to repair it. It's not too hard to get a few extra volts but trying to double the output is going to require more than just messing with the divider. Chances are you'd have to rewind the transformer secondary and upgrade most of the components in the output.

If you want 24V, pick up a pair of hot swap server PSUs, they're dirt cheap on ebay. Modify at least one of them to isolate the DC ground from earth ground and then put them in series. This is very commonly done by the RC airplane/multirotor guys to power big battery chargers so it's a pretty well documented process. These server PSUs are typically good for more than 40 amps too so they're likely more than adequate for whatever you want.

[ovnr]

If you want 24V, pick up a pair of hot swap server PSUs, they're dirt cheap on ebay. Modify at least one of them to isolate the DC ground from earth ground and then put them in series. This is very commonly done by the RC airplane/multirotor guys to power big battery chargers so it's a pretty well documented process. These server PSUs are typically good for more than 40 amps too so they're likely more than adequate for whatever you want.

They're also loud as fuck, and isolating them from ground is kinda iffy from a safety and EMI perspective.

There are plenty of 24V PSUs out there. Buy one.

[keymaster]

Could the transistor be an NPN 331 transistor? http://www.nteinc.com/specs/300to399/pdf/nte331.pdf

No. Your transistor is in a to-92 package . this is a to-220.
The 331 marking  most likely is the date of production.   

[Seekonk]

Perhaps instead of going for doubling the output on the first try you should better modify it so you can slowly crank up the output.
That way you can spot weak points before they break.

"Strangely the resistor values didn't match their measured resistance, but I went with the measured values and added a resistor R2 to bring the REG input to the tl431 down by 1/2"

A TL431 doesn't have that high an operating voltage.  Perhaps you went way over. In all my career I've never seen an opto blow up.

[james_s]

They're also loud as fuck, and isolating them from ground is kinda iffy from a safety and EMI perspective.

There are plenty of 24V PSUs out there. Buy one.

They're not all loud, I have a pair of IBM 675W server PSUs that I use for my charger that are almost silent under light loads, the fan ramps up as you load it but even at full rated load they are very quiet. Isolating the DC ground does not seem to cause any issues, it is very common practice with many hundreds of people using such a setup and I've yet to hear of a single issue. As long as the chassis remains earth grounded it is still reasonably safe. If one is concerned about the DC ground floating, a MOV between that and the chassis will provide reasonable protection. There are loads of power supplies that do not tie the DC ground to earth and I would certainly trust a well engineered server PSU over most of those.