How to Convert ATX PSU to 24V output or +/-12V output at 10 Amps or more! Power Supply MOD

Well I don't know 100% sure if this is going to work but I have this sound theory how we can easily convert just about any ATX PSU to give us 24V or plus/minus 12V symmetrical output at high currents. It works on paper so let's see...
Amazing $1 Short Finder
• Convert Your MULTIMETE... Part 1
• The $1 DIY SHORT CIRCU... Part 2
• The Amazing $1 Short F... Upgraded Version
DIY Huntron Tracker / Octopus Curve Tracer
• How To Build the Amazi...
MilliOhm Meter
• Help!! My DIY Electron... Part 1
• Build A Cheap And Supr... Part 2
• Build A Low Cost Micro... Part 3
10KHz -255MHz RF Signal Generator
• Help!!! My Electronics... Part 1
• Build A Low Cost 10KHz... Part 2
300W Active Load
• Build A Low Cost DIY 3...
Amazing $1 ESR Meter
• Build This AMAZING $1 ...
Ring / Flyback Tester
• All You Need To Know A...
Amazing $1 Ring / Flyback Tester
• How To Find SHORTED TU...
Signal Tracer
• Build A Cheap DIY Sign...
Signal Injector In A Pen
• Build A Cheap DIY Sign... Part 1
• An Absolute Beginners ... Part 2
• An Absolute Beginners ... Part 3
Modified ATX PSU Analyzer
• This ATX PSU Tester An...
Single/Dual Op Amp Tester
• How To TEST OP AMPS Us...
Zener Diode & LED TEster
• How To Test Zener Diod...
TL494 Tester
• How to test TL494 / KA...
TL31 Tester
• How to test TL431 / KI...
Learn Electronics Repair is now on Discord! Come and join the fun, it's free.
/ discord
It takes a lot of time and effort to make KZbin videos. If you enjoy my videos or they helped you with your own repairs please consider subscribing to my channel or click LIKE
If you would like to help support this channel you could consider sending a Paypal donation or subscribing to my Patreon page. All monies received will go towards buying more items to repair and make interesting videos, or to improve my video recording equipment and will enable me to spend more time making better videos. If you subscribe to my Patreon you'll get my Whatsapp number and I'll give personal repair support
You can send donations via this link
www.paypal.com...
You can subscribe via this link
/ learnelectronicsrepair
Thank you for your support.
Richard

[@mikebond6328]

I don’t know why but I’ve never been able to understand what the dots on inductors means until hearing your explanation of it. Now I get it. Thank you!

[@andrebooysen6222]

HI, you explained in detail how to get 24v I didn't think it was possible. You put in theory (on paper) and it work. GOOD JOB!!! Can't wait for the next video.regards.

[@LearnElectronicsRepair]

Cheers 😁

[@johnmaulkin-ur8lm]

Thank you, you are amazing, no BS, no ego. It’s just fascinating to watch your work. Very best regards.

[@davidrobertson1980]

Well done Richard, your thinking is logical and obvious once you understand what you've done there.

[@4b5urd.]

Richard, very clever! I am fascinated by your work and this was quite a clever little experiment. I am currently working on a couple atx power supplies and this would be perfect so I may give it a go. I have a tremendous amount of respect for your knowledge and experience, really enjoy your content!

[@davidhollfelder9940]

Very cool project. In a past video in this subject, I mentioned the HP-DPS600pb server power supply. Upon discovering this PSU, I completely switched/abandoned modifying ATX PSUs. The HPs are better quality, cheaper, easier to mod, and have much more power. You ought to check those out.

[@chrishartley1210]

An alternative is HP's range of HSTNS server power supplies. The connectors are different and they don't have 5V output but they go from 460W right up to 1300W and possibly beyond. These are so over engineered they are almost bulletproof and are usually cheaper than the DPS600. Modification, if required, is very similar and I'm told 13.8V is possibly for radios.

[@luisledesma586]

great job, I am very happy to see it so well explained; I often thought about that but lack the confidence on my skills to try it. thanks.

[@will_doherty]

Good video, thanks for sharing. Great demonstration of your understanding of circuits, thinking "outside the box" and then putting what you have found to good use in making a very useful modification. I have been a little critical of some of your content on occasion - though I do appreciate your time and effort - so it's only right to give acknowledgement when due 🙂

[@LearnElectronicsRepair]

Thanks Will it's much appreciated 🙂

[@tarotsuperstars]

@@LearnElectronicsRepair You are using a Solder Fume Extractor aren't you Richard? 😀

[@Lightrunner.]

It is alive and work 👌👍🖖 Nice explanation Richard .

[@LearnElectronicsRepair]

SPOILER WARNING for the impatient viewers Revealed - does it work or doesn't it.... *For those who commented that they don't want to watch unless it works - Yes the theory was good. It Works*

[@daz41262010]

nice job Richard :) so glad you worked out that problem :) well done sir :)

[@mik310s]

I rewound the transformer on one oh these years ago to give 5, 12, 24 and 36v. I disabled all the OV protection circuits and used some of the unused pins on the controller for extra feedback, it worked for several years then exploded :D

[@gordslater]

I wonder if it was my comment you were referring to about 12V+5V = 17 - I typed a long rambling comment when you were trying to get that variable PSU running. I just checked and I did state you'd need 2x ATX PSUs to get 17V this way: "SUGGESTION - temporarily combine 12V from one ATX PSU in series with the 5V rail from the second one" (the second ATX PSU's PCB needs to be floated from the mains earth of course) Missed the last couple of your vids, sorry, or I would have pointed that out earlier, but hey, all this ATX PSU rail messing is all fascinating and very educational too. A few years ago I saw some good mods that ham radio guys had done on ATX PSUs to jack them up to 13.8V as well as more advanced mods but I can't find it in my bookmarks unfortunately. A 100W HF transciever needs a beefy "12V" supply (actually, they are designed for 13.8V for car mobile use) and a sagging PSU rail can cause non-linearity and be a bad thing, If ever you need a big load resistor for testing ATX PSUs buy some galvanised steel fence wire and wind it spiral fashion around an old wooden chair or stool turned upside down. Tap into it with some ceramic "chocolate block" screw terminals that are sold for heavy duty earth bonding - plastic ones can melt because they clamp onto the hot steel wire. Just be careful you don't set the chair legs on fire haha.

[@LearnElectronicsRepair]

Yeah it was your comment. Maybe I misunderstood what you meant about 12V and 5V in series but it doesn't matter as it made for some very interesting experimentation and learned a lot along the way, as did many others I expect 🙂

[@gordslater]

@@LearnElectronicsRepair yeah I'm abolutely loving this ATX stuff. I spent a few years on custom, ad-hoc and modified 0 to 48 volts variable PSU in the 800 to 1200 watt region with the 723+3055 design being my typical custom-build solution Modifying surplus ATX is sooo cheap compared to what I used to do but being a analogue-focused guy I limit my tinkering to basic repairs on them and occasionally placing them in series for 24 and 48V with a chassis ground lift. My next job will be to get 12, 19 and 48V from ATXs with battery fail-over to provide my aging parents with a house-wide PoE UPS now that they are soon to get full-fibre internet. No landline 48V battery voltage from the exchange means the customer needs to come up with a power-cut solution in the UK) We found that in Storm Arwen the mobile masts lasted 5 minutes or less (in some cases they failed immediately with the power cut!) and when they did come back on it was low-power-only for several weeks.

[@gordslater]

and that 24 volt single PSU mod you just did is pure genius haha

[@alanwhite4003]

Excellent (again). I love when you push the limits 😂😂😂

[@filenotfound__3871]

In order to get a stable 24V Dc you had to use the grup coupling inductor on the board, and make a new loop on it identical to the +12V rail one.

[@mariushmedias]

DiodesGoneWild has a series of videos where he goes in detail about various atx power supply designs and how changing the voltage or modifying the transformer affects the components and how the voltages change at various points in the circuit, those videos are really worth watching. (edit: interesting approach, didn't consider doing it like this and wrote before watching... end edit ) Converting this to 24v may work but you have to be careful about all the components, changing the resistors on the voltage reference, changing capacitors to ones rated for 35v, changing the resistors in the over voltage / under voltage protection circuit, and you may also have to change the rectifier diodes on the secondary heatsink, because you have much higher voltages and the diodes may not handle them.

[@CliveChamberlain946]

Yes, I recall that. I was impressed enough to post on the EEVBlog forum. Somewhere, there's a part showing that increasing secondary output must be followed by snubber adjustment to protect the primary switch. See *How Does a Switching Power Supply Work 1 (schematic, explanation, example, modifications)* for the effects of how changing the output divider changes ratios and snubber ringing, etc.. on the primary side.

[@mikepanchaud1]

The capacitors still only have 12 volts across each of them... He reduced the max current so heat should not be a problem in theory.

[@CliveChamberlain946]

Also DanyK still has my favorite MOT experiment. Look for *INSANE high voltage supply / ŠÍLENÝ zdroj VN*

[@LearnElectronicsRepair]

The regulation is on the 3.3V rail, this mod does not affect that, nor the voltage reference or anything else. The voltage across all capacitors is the same as before, I just made use of the redundant half cycles. The only way this changed the performance of the PSU is that you should only draw half the maximum current rating at 24V. Therefore mounting the additional double-diode on the same heat sink should be perfectly fine as the total maximum wattage remains the same. If you feel I am wrong in these assumptions let's talk about it.

[@LearnElectronicsRepair]

@@CliveChamberlain946 I'm not changing any of that, I'm just making use of the redundant half cycles to generate the -12V. As far as the primary circuit snubber and all are concerned, nothing has changed

[@ThomasEnder_pt]

Amazing Richard. Learned a new one. Thank you.. 🙂

[@servisiprosound820]

Hi Richard, this is a great idea, I could not wait to test it but two of ATX PSU I am trying it I am getting -24V out of the common anode, even after filters and caps connected. I'm scratching my head why but may be this is happening to someone else as well. Measuring the transformer output I find around 9V AC and in the output of the diodes I have +12V and -24V. Strange..... Thank You for all your Lessons.

[@BjornV78]

It is maybe already suggested by others, but what about converting a (old) high wattage audio amplifier to a bench powersupply ? Most of the older amplifiers have a decent transformer inside, which can easy put out +60V AC, then a DIY PCB with voltage and current regulation, and the amplifier housing can be repurposed by adding a blanc sheet of alumium as the frontpanel, a panelmeter and some banana jacks, and you are good to go. Grtz

[@LearnElectronicsRepair]

I'm already working on that one as shown a couple of videos ago when I failed to get 17V from an ATX PSU using 12V and 5V in series

[@bobbysparky4299]

Impressive young man. Will you make a functional +/- 12V 10 amp benchtop power supply as a follow up that uses the most excellent work you have done here? By the way, XRayTonyB recommended your channel to his viewers and I am extremely happy I followed his advice and watched your channel. I am certainly going to recommend your channel at every opportunity.

[@LearnElectronicsRepair]

Yeah something like that is the plan

[@kokodin5895]

one more thing to add by using this method you can make 17V power suply using full 12v amp ratings if your 5v is twice that for instance with 10A 12v and 20 A 5V rails, you can buildnegative 5v rail, don't use positive rail and get pseudo 17V at rated 10 A from an old pentium 3 era power suply and the load capabilities should scale up the only problem with those circuits is you better insolate dc ground fromcha sis ground if your device powered from it uses negative rail as ground

[@GapRecordingsNamibia]

Well done Richard.... I am sorry, but most of these comments are just asinine. What are these people on about, changing resistor dividers and capacitor ratings and adding snubber circuits? All you did was utilize a redundent half cycle? Am I missing something? Where was the regulation or any of that changed? People need to stop making assumptions before they finish watching or have only skipped through the video and not actually seen what was done!

[@igortcanas4268]

I appreciate your help. I'm wondering if this would be compatible with 110v ATX power supplies?

[@justme-n-gracie]

Great presentation. Thank you

[@elsaarmstrong-zp6ng]

Have you an isolating transformer Richard? I once had a bad short scoping on a SMPS! I won’t forget that in a hurry!

[@fixitfaster]

Great , great video.Thanks for sharing your expertise. I learned a lot here. I tried it and get the inverse voltage alright but any load whatsoever makes the power supply unit shut down. Any ideas appreciated

[@CTCTraining1]

Well done Richard, I’m very impressed but not at all tempted to try to emulate this myself. Maybe you could design a board which connects to an unmodified ATX psu and gives the bench controls needed? At least in that way the variable implementations inside the psu are shielded from the project and, if one fails, another can be swapped-in. Just a thought. Keep up the great work 👍😀

[@king_james_official]

very elegant

[@George-nt8uw]

I think that the inductor you used was a power transformer and not a filter inductor. Additionally, there should be a second capacitor after the filter inductor. One before and one after the filter inductor.

[@djfernando16]

I'm just thinking, you said in the title 10 amps or more. However, one of the 12v supply is rated at 10 amps, the other one at 2 amps. 24v at 1amp should be fine, but once you go over 1amp, won't it overload the lesser circuit? I'm asking because I want to use an atx for a cnc to power 3 motors (rated at 24-80v), drivers, control board, fans, etc

[@POLJgaming]

I'll try this one

[@snakezdewiggle6084]

Did we get confirmation on the current symmetry ? My last attempt, transformer went short, sparks, smoke, etc. at just over 3 amps. Have to remember these are mass-produced by the lowest bidder... I wind my own toroids now. ;)

[@daviddicker-w8f]

New Gere but I have a question about my 100 amp switching power supply. It has two adjustment screws one is for voltage and I can't find anywhere what the other is for . I ask because I moved both adjustments and now don't know what I've done. Help plz

[@MiniDave-d5q]

Great. Looks like you have got a 0-12v or 0-24v PSU that would work great in the Variable Bench PSU you were building. How about it?

[@suryapatnaik5720]

What if we make a separate circuit with diodes in bridge connection taken from the two 12v points. Two inductors and pair of capacitors, just like 12v output, but we get 24v without connecting the ground. For 20 amp which number diodes can we use?

[@mahlapropyzm9180]

Could you replace the common cathode diode pair with a bridge rectifier and tap off +12V and -12V from the outputs? If not, why not?

[@gheorghepalade3882]

This diodes used in atx psu are ultra fast diodes. Bridge rectifiers are low speed, used for 50Hz so you cannot used instead of ultrafast one.

[@GregOnSummit]

The capacitors to ground at 8:53 and 9:04 ... how does this work? Does it smooth the output? What stops a direct short to ground.

[@justme-n-gracie]

I don't understand why you think the caps would be a direct short to ground...? What comes out of each diode is a DC pulse that is accumulated in the filter capacitor. The cap-choke-cap just provides more filtering and a bit of static regulation.

[@jjsprtn627]

I have a problem, the output voltage is -19v. Is it okay if I keep it? While I just change the capacitor to 25v. Note: I used 2 diodes from my broken ATX PSU. Because I don't have common anode diode.

[@f1remandg]

Hi I did send a comment last evening, however this was in the comments section in the vlog on your new adventure live Q&A. I tuned into watch last night at the designated time, but nothing and nothing in your KZbin site, could you update me on what happened and have missed or mixed up something. Regards David

[@tenlittleindians]

Your half way to the bench supply I don't need but want and that would be a 50v supply that's capable of 10 amps. Two of these Frankenstein power supplies in series?

[@LearnElectronicsRepair]

Yeah that should do it. Or a modded one plus another standard one for an additional 12V input when needed 🙂

[@kokodin5895]

small update on my questions i diged out old pentium 4 power suply, that is it was made in year 2001 and got a sticker inadequate for the device i have aded negative 5v rail to it acording to the video, but it did not play nice since as any old power suply the only regulated power rail is +5v by ka7500b chip so whenever i loaded 12v or from 12 to minus 5 the voltage saged dramatically so i cut feedback from 5v to the regulator ic sensing pin and conected it to voltage divider potentiometer located on 12v rail , 10k potentiometer to ground and fed output to pin 1 on the regulator ic now 12v is rock solid down regulated to 11 so i could get 15 under load from between 12 and -5 overcurrent protection likely don't exists in this one thanks for the idea

[@f1remandg]

I was going to ask what is the easiest way to check a motherboard with appropriate meters.

[@hilariohernandezandaverde3077]

puede decir que ic-pwm tiene esa fuente atx, I did that but it does not give more than 800 mA why, but is ok it does give 24 vdc, whats the ic-pwm sorry if my englis is not good I hope you will you understand.

[@anthonydenn4345]

Very interesting mod. Genius really, never seen that before. In the video you mentioned putting 2 atx psu's in series to get 24 volts. If you convert 2 atx psu's like this, I suppose you could then get 48 volts in series using your mod. Would make a useful cheap psu if you could do that and have the 48 volts adjustable. I'm guessing this mod could be implemented on any atx psu's that outputs 12 volts in this way.

[@kokodin5895]

now here's funny qyestion how would overcurent protection of the power suply play around if you load just the negative rail? and how wuld it work with loads connected to 24v? because as far as i know if you didn't disarm the current monitoring, which is most likely on the output side, sensing the voltage and compensating on off timing of the primary, you would still have a brain that thinks it can give the same current as it did before the mod if the ic is smart and monitoring more than one rail it may go crazy and trip before you draw that half power because other voltage rails might go up too much, or the primary transformer might get too hot if it don't what would happen if you forcfully lower the 12v rail to 8-10 can you get the amps back or would the other rails be a problem

[@nilolee2426]

Is it true in a atx power supply , if you connect all 12 volts wires together will give you the max amps listed , let say it has total 4 wires will give you the total amps? Example 12-16 amps

[@justme-n-gracie]

the only effect the wires have on the current output is the current capacity of each of the wires. Paralleling the wires gives the effective current carrying capacity of a larger conductor with less resistive loss . Example: a 1200 watt power supply will supply 1200w or 12v @ 100a. The ourput conductor/wire(s) should be sized for that current to be delivered to the load. Paralleling smaller conductors is the same as one larger conductor. When you attempt to pull a large amount of current through a conductor that is sized to small for the current needed, the resistance of the wire causes voltage drop and heat in the conductor until the conductor melts. LOL short and simple answer is yes you need to use all (and an equal number of each) of the black and yellow wires if you want to obtain the maximum current from the power supply without problems aka failure, smoke and fire.

[@keithcitizen4855]

Not to appear dum but what you sort of created is a centre tap transformer?

[@andreasproteus1465]

My thoughts exactly. Effectively he connected the two transformer windings in series giving 24 V. The centre tap is the two grounded ends of the of the windings and the two double diodes form a bridge rectifier.

[@LearnElectronicsRepair]

@@andreasproteus1465 No not really. The two transformer windings are connected exactly as they were before to the +12V dual rectifier diode. In fact no changes were made to the original ATX circuit whatsoever. I simply used the existing and unused negative half cycles to generate a -12V supply rail. The connect between +12V and -12V outputs to get 24V. Basically I guess you could say each 12V winding was half wave rectified, and I just added two reverse direction diodes to they are now full wave rectified

[@GapRecordingsNamibia]

No. Not even close. He made use of an unused halfwave. Did you not see the scope traces or hear the explanation of how the inductor is wired or works?

[@GapRecordingsNamibia]

Wrong! He did not connect anything together. Warch the whole video before commenting!

[@therealspixycat]

Can you get 24v out of an atx with active pfc?

[@FlavioCasimiro.dancer1983]

Do you have 320v monofase in the canary islands? Is it not 230v like in Europe? I heard about 120v and 230v monofase and 340v trifase. Never heard of 320v...

[@GapRecordingsNamibia]

The "320V" Richard was refering to, is 230Vac that is smoothed dc, that still lies ahaed of what he drew, he did not include that here, it was not relevant to what he was explaining in this video.

[@mevk1]

Takes a few minutes to bolt diode to heat sink/fan. Why not stress test it at say 9A for a day? A successful test would make most of the doubting thomas's scramble for cover.

[@Scoobydcs]

you sound like a fellow stokie

[@LearnElectronicsRepair]

Yeah I was born in Heron Cross, grew up in Blurton and lived in Normacot when I married. We left in 2016 due to a certain referendum result and moved to Maspalomas, Gran Canaria. Guess you would call us political refugees LOL 😅

[@Scoobydcs]

@@LearnElectronicsRepair wish I could've gone with you!!! But that's a whole other can of worms

[@josecito-moncep-acosta7328]

tienes -14 volts, porque esa nueva linea negativa ,NO PASA POR EL C I CONTROLADOR !!!!! ....

[@juniora5732]

If I do this the protection circuit still works someone who can confirm for me

[@101markharris]

camera from above view filmed in the seventies

[@LearnElectronicsRepair]

?

[@GapRecordingsNamibia]

So? Leave if you don't like it, there are enough of us that want to see it and could not care less about whatever the angle is as long as we can see it. We came here for the electronics and knowladge..... Not for glamor.....

[@justme-n-gracie]

Do you know why there is no "save" function for this video?

[@lattehour]

hahaha he is allways coocking something

[@mitkopetrovik259]

Already have minus 12v in the factory ATX power supply, the wire with blue color... you just need to replace the built-in diodes with others for higher amperage, I would install x4 identical ones for +/- Recommendation for an identical capacitor and to run the same number and cross-section of wires through the same inductor... stop polluting youtube space with nonsense, no need to invent hot water!

[@toseltreps1101]

The title sounds like you figured it out but in the description it sounds like you didn’t, that's stupid and makes me not want to watch

[@CliveChamberlain946]

It's hard work to get >100k subs in electronics. Sounds like judging his humility makes someone feel better? We're here to learn and the work he does helps us all *think* better. Not look for "how to" handouts.. 😉

[@LearnElectronicsRepair]

@@CliveChamberlain946 True, and anyway it worked out good this time

[@LearnElectronicsRepair]

Well actually you should watch... probably 😉