A quality bench power supply is essential for electronics work. Nobody wants to go through the trouble of digging through their electronics bin just to find a wall wart with the right output. And, even if you were so inclined, it would be folly to assume that its output would actually be clean.
You could, of course, purchase a purpose-built bench power supply. But, this is Hackaday, and I’m sure many of you would rather build one yourself from an inexpensive PC power supply. Normally, you’d do this by separating out the different voltage lines into useful groups, such as 12V, 5V, and 3.3V. [Supercap2f] wanted to take this a step further, both to get a more useful unit and to practice his PCB-making.
His design uses a custom circuit design to fuse the circuits, and to provide some basic logic. Using the LCD display, you can see which lines are powered on. There is even a simple 3D printed cover to keep everything neat and tidy. [Supercap2f] has posted all of the design files, so you can build one of these yourself. We’ve seen similar builds in the past, but this is another nice one that anyone with the ability to etch PCBs can build.
51 thoughts on “Taking The Converted PC PSU Bench Supply A Step Further”
Uh, so using a 2 line character LCD, a micro, quite a bit of circuitry and a bunch of relays to – turn 4 rails on and off? LCD used only to indicate that the rails are ON?? Seriously??? WTF????
What’s wrong with just using 4 switches? Or, if one really wants to draw 20+A from the supply, using a switch to turn a relay on? Heck, even a LED could be added as an indicator that a rail is on if the the position of the switch is not enough – it would be also much more readable than trying to decipher the LCD from a meter or two distance.
This is a completely ridiculous mod. He should rather have added current limiting and over-current protection instead, that would have been a LOT more useful and would have saved him from having to constantly replace those glass fuses after every short.
I read it as if you were yelling loudly so that everyone could hear you.
Was there anything about this build that you liked?
Once again, nice rant you have.
I got the same impression. That first paragraph really set the tone for the post.
are you a little angry today? Might want to chill out, not everyone is an expert like you!
Guys, and what about something actually addressing the points I have raised? Or you just want to attack me because I have dared to point out that someone grabbed a soldering iron before actually engaging the brain?
I’m reading it as a training wheels project to make a PCB and use a PIC, the spinoff of having any marginal use was gonna be an afterthought.
There is always an opportunity to discuss ways to improve someone’s project. However you have to consider how you communicate your ideas – if you come across as angry, condescending, and rude no one will want to talk to you no matter what you are saying. How you talk to someone is just as important as what you have to say.
+1 many times over!
Like someone who might grab a keyboard before actually engaging the brain?
Someone built something that was usual for his needs and learned something in the process.
C´mon, when one can etch PCBs and program a micro, there are tons of better builds than this. This one is even dangerous, with contacts separated from 0.2″.
I really dunno what the “step further” of the title means. I would call that a fail of the week. One cannot even adjust voltage or limit current !!!
Dangerous? 0.2″ should be fine for 24V (the maximum voltage difference I could find), no?
Dangerous? He’s just wiring this board to the output of a PC ATX PSU. The only modification he did to the PSU was to chop off some unneeded wires like the peripheral plugs.
The “tone” may be a bit aggressive, but he has a point: this hack is overkill.
If pressured to use a micro and LCD “or else”, I would have added digitally adjusted output voltage and current limiting / monitoring, NOT pushbuttons to control a micro to control relays.
I would also question the need to 3D print everything, even if that mean waiting a long time to just get a rectangular box (easier and cheaper to build by another means)
At least should have IOTed it, so complete strangers can hack in and see which rails are on and off. :-D
despite the fact that I fully agree with your comments… I must say that this website isn’t about practical build. In fact to more nonsense the better. The problem that caused your disappointment is most likely the headline “Taking the Converted PC PSU Bench Supply a Step Further”, this creates hope and inspiration that is soon to be destroyed by the simple truth. Please don’t blame the guy who made the project, I’m sure he did his best and is pleased with the results or otherwise had great fun building it and learned a lot during the process. Instead, blame the author for the inappropriate headline.
Yes, blame me :) But I think this does take it a step further, though maybe not a leap. It’s a good way to learn some new skills along the way to a more complex build.
So it’s one giant leap for [Supercap2f] one tiny step for hackerkind :-D
This build was mostly about Supercap2f learning some new skills. I think it’s a good next step for someone looking to upgrade from a basic conversion.
If I remember correctly, PC PSUs expect to supply power mainly on the 12v rail, and will behave weirdly if you just tap the 5v or 3.3v rails with no 12v load.
Also, isn’t one of the reasons to use a lab supply to be able to limit the current? Preventing Magic Smoke and stuff?
Not all of them expect a load on the 12 volt supply. The one I usually use for experimenting with needs a draw on the 5 volt, so I connected an LED circuit. Now the whole thing lights up internally.
You need to check – some supplies have a built-in load resistor, either on 5V or 12V rail that ensures the minimal load. Some don’t have that. More modern supplies will work even without it, but the older ones will either not start or not regulate the output voltage correctly without some minimal load on one of the rails – which one that depends very much on the design of the supply in question and it varies.
Yeah, it’s an older one from a 386! ;)
Notarealmail is right, some power supplies need the 5v rail to be slightly loaded to be stable instead of the 12v rail(s).
I have seen some supplies that have a load resistor in them so they don’t need an external one for stability.
I have also seen some designs that are able to switch that internal load off when an external load is connected to raise efficiency a little.
I’ve got an old 40 Megabyte hard drive kicking around, what’s sole remaining purpose in it’s old age is to be a load to insure random AT or ATX (Or whatever else*.) PSUs fire up at my command.
* Few weirdies kicking round the basement somewhere, like a halfassed halfway crossbreed of an AT and ATX, some custom odd shaped ones.
Dammit grammar nazis, ENSURE, I meant ENSURE.
I used a “half working” H4 auto-bulb (12V) as base load on the 5V of an old AT PSU for this purpose years ago.
I’ve seen plenty that require a load on the 5V rail. I’ve read about ones that require a load on the 12V rail. What really surprised me though was when I recenlty discovered a supply that requires a load on the 3.3V rail. Granted, this was out of a network switch, not a PC. I guess there are all kinds out there.
So far no one in the comments seem to like this project. Lets get some positive comments in here to way out these life sucking dick-weeds.
I personally like what he did, it inspires people like me to build something similar and possibly improve on his design.
If you want to expand on your work, think about adding a part for variable voltage.
I’ve been really happy with a board you can find on eBay for pretty cheap based on the LTC3780.
They are usually fused for 15A, and can buck and boost, so you would be able to hook it up to the 12V line and get a large range of voltages, from just under 1V up to roughly 32V I believe. They easily output 10A, and you get current control.
How dare you interrupt the whinging with an actual constructive suggestion! ;-)
Seriously, that is the kind of thing I was hoping this project would be about.
And Nixies for displaying the voltage and current… :D
I cant really understand the hype for this – in my opinion – outdated technology. I prefer the look of LED digits. Or – if it really has to be glass – VFDs
Exactly! I think this a good start and learning experience. There is no reason the project can’t be expanded further, but starting with basics is usually a good idea when learning.
Matt – some of those boards on eBay seem to have multiple inputs and outputs. Do you get the same voltage out on all 3? Other boards also seem to have 3 separate potentiometers. Why 3? Is that for individual control of separate outputs? Can you post the link for the board that you have had good experience with? Thx.
I have a few of the ones that have blue pcbs, with only one input and output and 3 potentiometers. The ones that have 3 potentiometers are controls for voltage, current, and a undervoltage cutoff point. I don’t mess with that last one because I think it’s used for battery-powered applications. This is one version I have:
Another one has 2 little heatsinks on top instead of one large one on bottom, no idea which way is better.
Fantastic. Thank you for the link. Much appreciated.
you mean one of this “automatic lifting pressure constant voltage current” ? ;-)
Yup, got that term to use for searching ebay from julian ilett’s youtube video on them.
Based on the kinds of reply’s I see to this project, I would NEVER want to post something I was proud of to this forum. Why and to what end – So I could get beat up? Great way to encourage people to participate.
Unfortunately, you are right. Luckily, the comments around here have gotten better. But there are still a lot of people who are negative.
In my opinion, it’s good for us here at Hackaday to post about projects from a variety of skill levels. Nobody gets into this stuff already being an expert, and it’s important to make that point.
It’s not going to win a beauty pageant anytime soon but if it does what it is designed to do then it is a success. To all the people complaining about it: there are many people in the world with different levels of skill Hackaday is no different while you might think this could have been better, This could be the pinnacle of achievement for others. So don’t complain unless you are being constructive. Well done Supercap2f
Fine build… esp. the controller part
I need 13.8V output for a Car Stereo. Has anyone modified a ATX PSU for that?
When searching the web i only found the modification for a old AT PSU, http://www.qrp4u.de/docs/de/powersupply/index.htm
but i only have a few ATX PSUs at hand.
One of the ATX i have stuffed in an 19″ Rack with all outputs on the frontpanel.
I will now use your controller board as inspiration for my build. Since I have a few 18F23k20 in my parts bin, i will use this controller.
For higher voltages (up to 36V 1Amp) i have an additional step up module built in.
No mod required, just run it off 12…
Okay, if you’re one of those real sticklers… open up the case and twiddle the trim pot on the 12V line until you get 13.8 under load…. but you won’t be able to do that if you’re a stickler, because of the warning labels threatening bodily injury and death from opening the case…. yeah, there is wild electricity caged in there, do not unleash it into your body.
I am a bit appalled at the rants. It’s a reasonable build, converts a bunch of stuff from the junk box into something useful. And it should be easy to add variable outputs, current limiting, etc. Are there other solutions? Of course, but often what you need is a warm up project to make sure that a more complex project succeeds. I’ve been watching for a cheap modular PSU for a similar build. I plan to use switches, LEDs, pots and LM317s, but that’s just my preference. I already have enough projects that need an MCU.
There was an article in “Fine Woodworking” by a guy who strongly advocated making dovetail joints out of wood scraps purely for the practice. He would make one whenever he started working in the shop and then toss it in the trash.
I’m a big fan of incremental development. Build the general framework and then add features one at a time. Doing this made a big difference in time to completion for projects at work. It is also a good way to avoid getting bogged down in feature creep. I’d just tell people I’d add that later. Often it never happened because no one really needed the feature. But by soliciting such requests I was forewarned to allow for it in the design.
We feature so many amazing builds around here that a lot of people forget that you have to start somewhere. No one is born an expert in anything. Supercap2f specifically says in his article that this was a learning experience for him (mostly an excuse to practice his PCB making skills). Unfortunately, a lot of people will only appreciate a project that would be a challenge for them personally.
Folks, puh-leeze. Someone has mentioned already above – if the author of the build can actually program a micro and make an LCD work, they are probably capable of googling around to see what actually makes a PSU useful.
The “have to start somewhere” argument is just nonsense – you do realize that using 4 plain mechanical switches replacing that entire LCD/MCU/relay pointless boondoggle would be actually *simpler*? I guess not.
I don’t mind people building stuff for learning, I do that myself, but I do hate poor engineering – people building stuff without thinking about how it is going to be used and building complex rube-goldbergesque contraptions where simple solutions would have worked much better.
I am sorry if my tone came across as condescending, but this PSU design is seriously facepalm worthy.
Don’t feed the trolls! [Supercap2f] don’t let random ppl on the internet discourage you. Sure there might be room for improvement (rev 2.0?) But as long as you learned something and enjoyed the project then that’s what matters.
I like it as a learning project.
Why is the pcb cut in half?
Please be kind and respectful to help make the comments section excellent. (Comment Policy)