I like to work on PCs, therefore I stored lots of ATX Power Supply’s over the years. I always thought that switching technology is something very complicated, but with some electric basic knowledge it is relatively easy to understand. Although this project is a bit complicated, and requires some basic knowledge, so this is not suitable like “first steps in electronic” project. This Mod will work with any ATX SMPS as long as it uses the standard TL494/KA7500/IR3M02 PWM Controller.


A standard AT/ATX Power Supply, the more power, the more current you can pull out at the end (available for cheap/free at your local junkyard)

An installation volt- and ampere-meter & current shunt (I use this one: http://s.aliexpress.com/QN3iaqYv)

A 470Ω resistor* (I harvested mine directly out of the Power Supply)

A 4,7kΩ stereo potentiometer & knob or a negative logarithmic potentiometer & knob*

A 1kΩ mono potentiometer & knob *

A 2kΩ trimmer*

Two 4mm laboratory/ banana plugs, one red & one blue/black

A 5V to 12V dc Booster or a 5V Fan (I use this one: http://r.ebay.com/F2n1Xo)

A round mains switch with the same diameter than the ATX-Wire outlet hole (I use this one: http://s.aliexpress.com/yYJVVJzM)

Some wire

4-rubber feed

a load/discharge resistor with about 50Ω and 10W*

Optional: A spindle trimmer with between 10kΩ and 100kΩ

Optional: A 1000µF 35V (if possible LowERS & 105°C rated type) Capacitor *

Optional: A 330µF 35V (if possible LowERS & 105°C rated type) Capacitor *

Optional: an ATiny45 or equivalent µC & IC socket

Optional: a pushbutton

Optional: an N-Chanel MosFET with a higher current rating than your PS 12V line & insulating washers

Optional: a piece of PCB & some solder pins

Optional: a few LEDs & resistors

Optional: a fuse holder

Optional: Heat-shrink tubing

Optional: a second 120mm fan grill*

All other materials, I have forgotten to write

Components marked with a * may have different values

Step 1: Understanding the Working Principe of a Switching Mode Power Supply

The idea behind the Switching technology is, that for a transformer running on a high frequency to transmit the same power a lot less material is needed, and it is much more efficient while doing it. ATX Power supply’s use some transistors to chop up the rectified and smoothed mains voltage to drive the HF Transformer. At the secondary side of the Transformer, there are three separated windings, one for each voltage. Everyone of these voltage rails has its own rectifier, smoothing capacitor and choke inductor. The three output voltages are add together for the control IC(TL494/KA7500/IR3M02), who switches the transistors and will increase/decrease the power going to the HF Transformer if a load is connected/ removed to keep the output voltage on a constant value. There is also a second smaller transformer, who is independent from the control IC. It is responsible for the 5V Standby and will run as soon as the power supply gets connected to main supply.

Step 2: From Theory to Practice

!DANGER! Inside SMPSs there are very high AC and DC voltages, who will be able to hurt/kill you! Before you start to working on them be sure you’ve unplugged it and the capacitors are completely discharged! Also important if you’re working on ESD sensible circuits is to bring you and your circuit to earth potential. You can do this by touching a radiator or grounded device or you can use an antistatic wristband like I do. After opening the four chassis screws, unscrewing the PCB, desoldering / clipping of the mains wire and plugging of the fan we can take it out. Next step is to clean out all the dust, which accumulates over the years there. The easiest way to do that is by using a toothbrush and suck the dust away using a vacuum cleaner.

Step 3: Time to Hack

Now we need to get rid of the original output cables. We cannot simply desolder them, because we won’t get the tin hot enough to melt, due there’s a huge amount of tin, and the wires will conduct the heat away. By using a pliers we shorten the wires down to about 1cm (about 3/8 inch) so we can grab them with a needlenose pliers and unsolder them very easily. Hang on the wires, you can make an ATX Extending wire out of them.

Step 4: Continue Harvesting

Back to the PS, the next step is to remove the 5V, 3.3V, -12V and -5V power rails completely. I started with the 5V, removing capacitors, inductivities and resistors back to the diode bridges behind the HF Transformer. Repeat it with the 3.3V, -5V and -12V rails. In my case there was a connection between the GND Rail and the Case to ground the output. We want to have a galvanic isolation, otherwise we can’t simply connect it in series with a second PS, so this connection has to go. After locating the control IC, I followed the conductor tracks going away from pin1 to locate the voltage adding resistors and removed all of them too. In my case there was a second IC, a TPS3510P. After downloading the datasheet, I read, that it is a over& under voltage Protection IC, who will also manage the PS_ON and POWER-GOOD signal. I figured out, that I only have to remove the IC and bridge pin3 to GND (pin2) to bypass it and turn the PS on. If you haven’t got this IC you need to wire the PS_ON to GND instead. On the PCB there is also a circuit to observe the negative voltage lines. It has to be removed. In picture 5 and 6 you can see all harvested components.

Step 5: Start to Assemble

Now we need to replace the original capacitors from the 12V rail with some higher voltage rated one. In my case there was a 16V 1000µF, and a 16V 330µF Capacitor, so I replaced them with 35V Capacitors because I planned to go up to 24V (25V would be the next higher rating, but that gets a bit close). If you do not want to go over 13V, there is no need to replace those caps. In your PS there maybe are different values, so replace them with identical (or a bit larger) capacity and 35V Capacitors. Continue by soldering a 100nF Capacitor between pin1, who is our REF pin, of the TL494 IC and GND to terminate noise signals.

Step 6: First Test

To the spindle trimmer. Now we need to make a new voltage divider for the reference voltage, by soldering a wire from GND to one of the outer contact of the spindle trimmer, and a second one from the other outer contact over the 470Ω resistor to 12V. The middle contact is the slider connection. It gets connected to the REF pin. After verifying that the spindle trimmer is completely turned to 12V it is time to do a first test. For the first test it is recommend to power the SMPS over an isolation transformer, to avoid to become part of the electrical circuit. Connect the multimeter to the 12V rails, and power on the PS. You should get about 2,5V. By turning the trimmer, you can slowly increase the output voltage up to 24v. You will hear that the noise of the PS changes, what happens due the different power the HF Transformer transmits. If this first Test was successful, it is time to continue, otherwise check the circuit.

Step 7: ΜC

I want to make the output switchable, so I need a kind of Power Switch. At first I thought of using a relay, but after a few tests I decided to use a MosFET instead, because the Relay contacts used to stick together, when they switch High Current. To protect the Output It is very important that it is always off after the Power supply get turned on. My Voltmeter needs a few seconds to boot, so I also need a delay, who only let me turn the output on if the Voltmeter has completely booted up. I know, using a microcontroller for something easy like this is overkill, at first I also planed to make a TTL-grave using components like an NE555 and a JK FlipFlop, but I had problems with the bouncing of the button, so I finally decided to use a Microcontroller. I had an ATtiny45 laying around, so I use that on. I wrote a simple program for the ATtiny, and flashed it using my ISP programmer. After prototyping the circuit on a breadboard, I soldered it to the PCB, and mounted it over the MosFET to the heatsink using an insulated nut and a isolation washer.

Step 8: Panel

It’s time to layout the panel. Start by scanning the case and loading the picture to your favorite picture editing software. Now you can paint the existing holes, the voltmeter, the potentiometers, and all the other stuff to the picture and place them where you want to have them. Attention, verify if there is enough space behind the cover to install the component. After making the background white, you can print it, cut it out, and glue it with a glue stick to the metal case. Next step is to transfer the holes into the metal case. By using a rotary tool and an end mill bit you can simply mill out the hole for the voltmeter. !DANGER! Use safety googles, to protect your eyes while milling out the holes! Be careful to stay inside the outlines. The mil hasn’t to be perfectly strait, because we’re going to finish it afterword’s. After grinding the hole to the outlines using a file, we continue with the other openings. After finishing them too we need to get rid of the paper. The best way to remove it is by laying the panel into a bowl filed with some water. After a few minutes you can simply peal of the paper and clean away the rest of the glue.

Step 9: Continue Assembling

Now we need to install the potentiometers. In total we need two potentiometers, one for coarse, and one for fine. If you did the test with the spindle trimmer in step 6, you may have noticed that the voltage curve is not linear, but logarithmic. Therefore, we need to use a negative logarithmic potentiometer to compensate this effect and make the voltage curve more linear. Due negative logarithmic potentiometers are very hard to get I developed a circuit to make one using a standard linear stereo potentiometer. After installing them into the holes, and mounting on the knob, I wired them like shown in picture 3. Also the LED holder and the output button can be mounted now. Let’s swap to the PCB for wiring the shunt. For all connections around the main output I use 2,5mm² (AWG13) wires to keep the voltage drop as small as possible. I soldered on the wires, crimped on wiring terminals, and connected them to the current shunt as shown in the circuit on picture3. Don’t forget to putt some Heat-shrink over the solder connections. The dc booster has to be adjusted before we can wire it, so this is the next step. The nice thing about these dc converters is, that they do have a standard micro USB plug beside the input solder pads, and can be powered directly from a standard cellphone charger. By using a screwdriver, you can set the output on the blue spindle trimmer to about 12V. After adjusting it successfully and soldering on a flyback diode for the fan, we can wire it to the 5V SB rail on the PCB along with the supply connections of the voltmeter, and glue it in place with some hot glue.

Step 10: Fitting Everything Together

Now it comes to a bit tricky step: fitting everything inside the original case. I had to flip my circuit board for 180° to fit everything in, therefor I had to extend some wires, move and modify some components (the PFC coil and the power plug) and shorten down the heatsink a bit. You can do that, because the FAN is oversized, due when the power supply it is installed into a PC case it has to run with the pre-headet air from CPU and graphic card too. I also removed the input fuse from the PCB and mounded it into a fuse holder beside the power connector. We can continue by solder some wire to the LED and put it inside the holder.

Step 11: Wiring

Now were going to install the negative laboratory plug. Simply solder on a lead, fit it through the hole in the case and fix it there. Be sure that it isn’t making contact with anything. Solder the other end of the lead to the output pin of the µC board and seal it with some Heat-shrink. To the positive one. Add the lead, install it into its hole and solder the other end to the 12V output on the PCB. Its time to install the flyback diode, the output cap and the discharge resistor. The best way to do that is by soldering the capacitor and the resistor onto the diode like shown in picture 6 and connect that to the output plug. Verify that you connected all positive and all negative leads together. The Output switch and the LED can be wired to the µC board now. We need to install the mains switch. Wire the NO pins of the switch between the mains fuse and the PCB. The LED is to be connected to the output of the DC Booster. After finishing that I soldered the original FAN connector to some wire and connected them to the DC Booster too.

Step 12: Panel Meter

It’s time to install the voltmeter. Start by wiring the current sense leads to the shunt. The exact circuit for your Voltmeter maybe is different to mine, so please wire according to the schematic you get with your voltmeter. I use some Wiring Terminals, so I can simply unplug them from the shunt if I need to. The positive measurement lead is to be wired directly to the positive laboratory plug. Now it’s about to mount the voltmeter inside the panel. If you made the hole at the right dimensions, you should be able to snap it in easily. If it won’t fit please larger the hole a bit or take a part the voltmeter, and install the frame first, instead of pushing it in with force, otherwise you will distort the panel. Plug in the supply and measurement plugs, and you’re ready to continue.

Step 13: Potentiometers, Tests, and Calibration

After soldering the 470Ω resistor to the potentiometer like shown in picture1, we can wire the other lead to the 12V Laboratory plug. The slider connection has to be wired to the REF pin. To the ground connection. Connect one of the outer pins of the trimmer to the last remaining potentiometer contact, and wire the trimmer slider and the second contact directly to the MosFET Source pin. After double-checking all connections, especially the one around the potentiometers, and turning them both to minimum it’s about to do a test. Connect the power supply to mains current and turn it on. You should get around 2,5V on minimal voltage position. By slowly turning up the potentiometers, you can increase the output voltage up to the maximum of the potentiometers, or to 24V if the range should go higher. Regulate the voltage on the trimmer up and down using an isolated screwdriver until you have about 24V on the maximum position. Now the range should go from 2,5V on minimum to 24V at the maximum.

Step 14: Finish Assembling

After the successful calibration of the output voltage, it’s time to fit the current shunt into the case. After some trying, I found a position on the inside of the cover to be the perfect spot. By laying it in place, painting down the spot of the holes and drilling them I was able to fix it there using some bolts and nuts. After some unsuccessful attempts to fit the fan back inside the case, I finally finished at installing it on the outside, fit the wire inside, and let it blow in the air. I use a second fan grill on the inside of the PS, not only to prevent wires from going inside the fan blades, but also to shield out disturbances and noise signals. I mounted the original grill to the external side of the fan to protect the blades from the outside too. After reconnecting al wires to the shunt and the fan, and zip tie all wires together, we can close up the case, and screw down al screws.

Step 15: Load Test

Now it’s getting interesting. Does the power supply work like expected? To test that we need an electronic load. If you haven’t got one you can also use some High power Load resistors, like I do, instead. Simply connect the load to the power supply and turn it on. If you increase and decrease the load you should hear a change on the noise, and the output voltage should always keep on the same value. In picture 3 and picture 4 I made a test to show, how the voltage behaves between load and no-load operation. For a more detailed description please view the comments on picture 4. I also did a long-term test over a whole hour with 10A at 24V, but the blown out air did not get warm.

Step 16: Finishing Touches

The next to last Step is to stick some rubber feed under that piece of art to give it a better stand, and prevent the surface it lays on from scratches. The very last step is to add some labels. I used my Dymo to print them, and stuck one to every operational element.

Step 17: Using It

To the easiest, but also most important step. Turn the SMPS on on the main switch and you will see the lights light up. You won’t be able turn on the output right now. After some seconds, the voltmeter will start to read out the voltage. Now you can start to set the output voltage using the two potentiometers. Once the voltage is set correctly, you can switch on the output at the button on the front. By pressing the button again you can turn it back off. I add some pictures and a short video while using it. Attention, the Power supply isn’t shortcut and overload protected anymore, so if you make a shortcut it will turn the power to 100%, and perhaps it will blow up your whole circuit! Maybe I start to build a improved version who includes a current regulation, spindle potentiometers, and a more accurate and lower output voltage. I hope you liked my instructable, if you have any questions or suggestions please feel free to leave a comment.

Best Greetings, Gabs’e

<p>I just did something similar to a PSU i've been using to power a small CNC machine.. The stepper drivers run on 12v-35v, but i kept having issues where the PSU would completely power off due to fluctuations in the power because the steppers were causing the 12v rail to dip if all three were running for an extended period of time.. so like clockwork, at exactly 30min's of run time, the PSU would fault protect, and power off. <br><br>After removing the TPS3510P and bridging pin 2 and 3, the PSU no longer powers off at that point in the job, also, a strange side effect, the PSU actually runs cooler, not by much, only about 15C, but that tells me the fault protection IC might have been faulty to begin with.. </p>
<p>This is excellent but you clever guys HOW CAN I GET RID OF THE RADIO INTERFERENCE these toys generate?</p>
<p>Krokenoster; I do not have one of these built, but am considering it after seeing this project. From the amateur radio side of my brain, if you installed ferrite beads on the wires just prior to where they connect with the external banana jacks that should resolve RFI on the line prior to feeding your instrument, toy, radio, etc.... You could also install some on the leads after you are out of the box leading to your appliance to ensure any surrious emmisions around the box are not transerfered as well. Just my thoughts with respect to RFI.</p>
<p>Hello KROKKENOSTER,</p><p>I understand your worries, but due the whole PCB is packed up inside a grounded metal case I don&rsquo;t think that it generates more radio noises than any other SMPS powered device.</p>
<p>Hey super Projekt!! H&auml;ttest Du diese Anleitung vielleicht auch auf Deutsch?</p><p>Viele Gr&uuml;&szlig;e, weiter so!</p>
<p>Hallo Dwargh,</p><p>Es freut mich, dass dir mein Projekt gef&auml;llt! Die Anleitung habe ich leider nur auf Englisch. </p>
Schade :(
<p>I always wonder why do you people start from taking Mount Everest rather than smaller mountain. Switchable supply is not even a bit as simply as you thinking it is.</p><p>Did you removed input filtering and protection circuit ? I suppose no so you must accept that case might be connected to mains voltage (and even higher) so its not save to use.</p><p>Of course it can be lab supply but user must be experienced of using switchable supply - why ? Because you even forget to filtering input and it might switch to very high voltages in first phase of run. It can simply burn circuit connected to.</p><p>In lab we did use commercial smps supply which time to time burning us projects. We did no idea why (to be honest we had idea but no equipment). When one of us bring 1 GHz scope we noticed there are very quickly disappearing pulsation with amplitude of 10 max output voltage (we were working at 5V when max was more than 100 V).</p><p>As I said it was confirmed supply... in university lab.</p>
<p>Hi w_h_y</p><p>Okay, you are right, switching technology may is a bit more complicated as you may can see from the text above, but for this mod it is not necessary to know exactly how to calculate the PWM Frequency, or selecting a storage choke. If you would like to build one from scratch instead, you were right.</p><p>I did no modifications on the circuit of the &quot;HOT&quot; side, The only thing i did was installing the new switch and moving some components to get more space inside the case.</p><p>Although the output of the PS is switched on a few seconds after the power-up, so it's not possible that the output voltage will rise to high on the star phase, I will add a 28V Supressordiode in version 2.</p><p>At work we use commercial SMPS too, who also burn through sometimes, but mostly they just stop to work or end up in magic smoke. I never watched the behavior described by you.</p>
<p>Why did you remove the components for the other outputs?</p>
<p>Hi asfi235,</p><br><br><p class="MsoNormal">It was necessary<br>to remove these rails, because they aren&rsquo;t regulated anymore, so there voltage<br>will rise and may damage some of the components, especially when the main rail<br>is under Load.</p>
<p>Welcome to the club: Just a note to let you know I have added this instructable to the collection: Encyclopedia of ATX to Bench Power Supply Conversion</p><p>&gt;&gt; <a href="https://www.instructables.com/id/Encyclopedia-of-ATX-to-Bench-Power-Supply-Conversi/" style=""> https://www.instructables.com/id/Encyclopedia-of-A...</a></p><p>Take a look at about 70 different approaches to this project. This topic is one of the more popular of all instructables</p>
<p>Hello,</p><p>A very good work Gabs'e .</p><p>i have two questions :</p><p>1- Is this idea applicable for all ATX Power Supply ?</p><p>2- How do you calculate the capacitors to increase voltage from 12v to 24v?</p>
<p>Hello bilux,</p><br><br><p class="MsoNormal">To 1: Like I<br>wrote above it works with every ATX-Powersupply if it uses a TL494, a KA7500 or<br>a IR3M02 as PWM Controller.</p><br><br><p class="MsoNormal">To 2: Witch<br>capacitor do you mean? If you mean the 1000&micro;F and the 330&micro;F on the PCB, I<br>haven&rsquo;t calculate anything, the only thing I did was replacing them with models<br>who have the identical capacity but a higher voltage rating to avoid that they<br>explode when I go over 16V<span style=""> outputvoltage.</span></p>
Thank you very much.<br>
<p>HURREY,, Gee this is just I allso aimed for.. having some AT-powersupplys just collecting dust.</p><p>Only 1 advice please, I stlughtered the one thus having this ON/OFF main detection, kind of a TRIAC with a gate controlled by logical levels of the controller. Would it require kind of a zero-crossing detection to activate? </p>
<p>Hi KISELIN,</p><br><br><p class="MsoNormal">I&rsquo;m sorry<br>but I&rsquo;m not that familiar with AT-Power supply&rsquo;s. In theory it should work too<br>if it uses a 494/7500 PWM-Controller, but due they normally don&rsquo;t have a 5V<br>standby rail, and also the output current normaly isn&rsquo;t that high like it is on<br>newer ATX PS I would recommend of using a ATX supply.</p>
<p>This is what I would called an &quot;INSTRUCTABLE&quot;. A weird term that defines a very complete step-by-step guide to achieve something.</p><p>Congratulations! This IS an exemplary &quot;INSTRUCTABLE&quot;! Added to my all-time favorites.</p><p>Now, what I would like to see next is some readings about performance, ripple noise, and so on. But that's a detail. </p>
<p>Thanks,</p><p>I'll try to make some measurements in the weekend.</p>
<p>Nicely done!<br>Thanks for sharing!</p>
<p class="MsoNormal">Thank you dieferman</p>
<p>Great use of an old ATX power supply, but I believe there is a small error in your final schematic,... the rectifying diodes from the transformer are the wrong way round...</p>
<p>Hi MarkH43,</p><br><br><p class="MsoNormal">Yes you're<br>right, I messed up the polarity of that diode-bridge in the schematic. Just<br>corrected.</p>
<p>Holy crap, nice to see someone who actually knows what he's doing with electronics. Awesome. </p>
<p>Thanks fstedie!</p><font class="Apple-style-span"> </font>
<p>Great instructable !</p><p>I'll try to make a current limited one too, it shouldn't be too hard to add to the supply</p><p>In your photo of the flyback diode, I don't completely understand why you put the capacitor accross the diode, wouldn't you want to put it accross the outpout to ground ? </p><p>thanks !</p>
<p>Thanks Mihemine,</p><br><br><p class="MsoNormal">The capacitor<br>is just fine there, It is used as output capacitor to terminate overshoots on<br>the output if load is added or removed. </p>
Neat as hell...but,what IS it? What does it do?<br>As a 'non computer nerd', I am kinda at a loss....<br><br>LOOKS neat as hell tho...
<p>It&rsquo;s a regulateable Laboratory Power supply for the workbench with enough power to run more-consuming-circuits.</p>
<p>it's a &quot;Adjustable High Current Lab SMPS Out of a Standard ATX PS&quot;</p><p>It gives you POWAH to run stuff</p>
<p>Enjoyed your build and learned a few things. Might have done a few things differently but then, everyone is different - makes ir difficult to be always correct or wrong.</p><p>I am fortunate to have some auto transformers 8A to 50A (500cm diameter) so I just dial a volt. AC or DC. The DC is full wave rectified with a pi filter arrangement. I would like to find an isolation transformer but it works fine as is.</p>
<p>Hello, it's a very nice instructable. I will make one in the summer! I am very interested in seeing a solution for current regulation!!! maybe you can do it as a part 2 of this projekt?</p><p>Greetings from Austria! </p>
<p>Thanks echterWiener.</p><p>Yes, I'll definitely make a part2 including the said features as the constant current mode if time allows.</p>
<p>I don't understand the need for the DC booster as you already have a 12V output available and, anyway, fans exist in 5V. Would you please enlighten me ?</p>
<p>Hello rafununu,</p><p>As I&rsquo;ve already written in the text the DC-Booster is needed to generate the 12V for the fan (and the LED) out of the independent 5V standby voltage, because the 12V output is the one who gets regulated from 2,5 to 24V. Sure you could also go and get a 5V fan, but that gets a lot more expensive, and a 12V model powered with only 5V won&rsquo;t blow that much air.</p>
<p>OK. And using the standby 5V you're sure the fan will work what could be the PSU state.</p>
Well done! I always thought about using an old atx ps as a basis for a home built lab ps but I never did. Here it is! Greetings from Germany p.s. Will you add a controlled/constant current feature?
<p class="MsoNormal">Hi<br>tapirdreams,</p><br><br><p class="MsoNormal">I'm glad<br>you like my instructable. I&rsquo;ll definitely add a current regulation using the<br>second error amplifier of the TL494 in a second moment.</p>

About This Instructable




Bio: I am an 20 years old mechatronics and I&rsquo;m living in the north of Italy. My interests are working on electronics, playing computer and ... More »
More by Gabse:858D SMD Hot Air Reflow Station Hack Adjustable High Current Lab SMPS Out of a Standard ATX PS Laptoprecycling-Touchpad 
Add instructable to: