Another ATX Power Supply variable voltage mod!

[spencerlinn]

Background:
I feel bad posting yet another topic on trying to get a variable voltage out of an ATX power supply.. I did plenty of searching and of course the issue is how many vasty different designs there are of these things, so for a n00b like myself it's hard to apply concepts from one design to another. I've been working on designing a regulated driver circuit for an e-cigarette, otherwise known as a "mod" when it comes to high-power vaping, so I need plenty of power out of my supply - upwards of 50 watts at minimum, and stability isn't critical (I've read these things get a bit unruly when you play with the voltage) since my circuit will have it's own on-board regulated rails.

Project:
What I need to do, is be able to simulate a li-po stack or lead-acid battery, and hope to get around 3.7-6 volts out of the thing. As far as my supply goes, I was able to find a place to start since one of my power supplies I had hanging around had a "431" voltage reference chip on it, a Bestec ATX-300-12Z, which another forum mentioned as being worth playing with to get some variable voltage. The post I read may have been for another "letter" of power supply, perhaps not a "Z", but mentioned that there will be two 431 chips, one for the 5v standby and one for the main 5v supply, and that the standby one powers a lot of circuitry in the supply itself so not to mess with it or cook the supply. I found the one that's on even with the green wire un-grounded (supply off) so I'd guess that's the standby chip, so I won't mess with that one. However, I found there are *two* other 431 chips… and putting a 20k pot on the vref pin of one of them does bupkis, and putting it on the other 431's vref does vary the voltage… about a volt down and half a volt up from 5v, but after that the whole thing shuts down and needs to be power cycled to get it back. I do have a 20w 12v bulb on the 5v rail to provide.. almost 10w of load for testing, maybe the load is too low? Should I forget this supply? Is the protection too smart for me? I've included a few photos of what I've been poking at in the thing - the arrows point to one side, the wiper, and the other side of my pot, from left to right.

Thanks for any insight you guys might have! And feel free to critique this harebrained scheme as well. 

[spencerlinn]

Ok, I know I'm bumping this, which normally would be fine, but I figured I'd let y'all know I done blowed it up.    So for now, probably won't be working on this particular supply. It still puts out voltage and sorta seems ok, but I definitely let the smoke out (with quite a bit of crackling along with the smoke too!) while fiddling with the pot I'd soldered onto the 431, and this was after I found that there was an over/under-voltage function kicking in on the control IC, and of course I just jammed that signal high with 470?. Oh well, I've got plenty supplies and this one had no particular use anyway so I'm just gonna have to either learn EE better (at all), or start with a supply that's closer in design to something someone else has modified. I guess if anyone has any suggestions for how I can approach a project like this without being so ham-fisted and blowing it up, I'd certainly listen, but otherwise figured I'd let everyone know this supply is probably cooked.

[IanB]

Let's just remind you that ATX power supplies are not designed to have their output voltages varied. Why, you may ask?

It's because of cost engineering. When a product is designed, the engineers work to a list of required features. If they add a feature that is not on the list it will bump the price up and the accountants will kill them. An ATX power supply has a required feature of providing a regulated 5 V supply. It has a required feature of providing a regulated 12 V supply. It doesn't have a required feature of allowing these output voltages to be variable. It doesn't have this feature because that feature is not in the specifications, and it would cost more money to design it in. If it costs extra money, and it is not asked for, it won't be there.

So, with a very high probability, you cannot get a variable voltage out of an ATX power supply. Any of them. Of any type or brand. The reason is not a technical reason, it is a cost reason. You are trying to find an extra cost feature that nobody asked for and nobody paid for.

[jwm_]

May I suggest as an initial project, rather than try to modify the supply too much, use it as the base for one of the many homebrew linear supply circuits out there. just replace the transformer taps and bridge rectifier with the output of your atx supply. a standard lm317 based constant current/voltage supply run off the 12v rail would be quite useful. You can be clever and relay switch it to the 5v rail when the output voltage is dialed down past 3.5v. and since you have the - 12v rail to drive the adjust pin of the lm317, you can dial it all the way down to zero volts. Something most beginning linear designs have trouble with.  Heck if you were to print up a board that I teg rated that with a proper atx plug, some banana jacks and a voltmeter and  I'd probably be in line to buy a couple.