ATX Based Lab Power Supply - 10th Year Anniversary Edition

I had published an ATX lab supply conversion about 10 years ago. It inspired a lot of folks to take their first steps in hobby electronics through that instructable. And now there are many many examples of people doing the ATX conversion on youtube, instructables, etc. My original ATX conversion got packaged more professionally with an added ampmeter which was really useful to have.

In the 10 years since access to cheap electronics modules on ebay, Amazon ... has improved by leaps and bounds. I had used one of these variable power supply modules in making a nice little variable power supply but this only supplied 25W. I was toying with the idea of adding higher voltage to my old ATX conversion, at least to 40V. I ordered two of these power boost power supply modules through ebay at approx $10 each with specs of settable constant current and voltage to 60V with a low volt input. At the same time I also ordered a low voltage variable supply module and some of those incredible littlle volt-amp meters.

By conincidence someone at freecycle was getting rid of a partially gutted desktop computer. I picked it up, found that it had a 400W supply, and the label showed that it had a -5V and a-12V. Neat! So, I now had everything I needed to build a brand new ATX power supply unit with two variable supply of approx. 1-10V and 10-50V at high current and fixed out puts of 3.3V, 5V, -5V, 12V and -12V at 10's of amps for the + voltages.

The second picture shows this new power supply resting on top of the old legacy ATX conversion.

I removed the ATX board with the attached fan and 115-220V switch plus the bunch of wires and quite a bit fo dust from the ATX box. I then collected all the other parts and figured out a circuit diagram. I then played with the overall box design and packaging till I was happy with the layout.

This would be a bigger box as the ATX circuit box had quite a big heatsink plus I would need space for the two other power modules and the volt-amp meters. I decided on a six inch cube as most of my other projects share the 6 inch dimension.

I then measured the main parts and drew these into the 6 inch dimensions. The final product turned out close to my original concept though the 6 inch cube turned out to be a tight fit

I directly drew the parts that would fit on the front and rear panels on 6 inch squares that I had cut out from white Pergo laminate flooring panels. I cut out the holes on my drill press and used a dremel for the square cutouts.

I used an old black can of paint to paint all the panels but somehow the paint ended up as a bubbly mess so I removed it with old rags.

As the powersupply would need adequate cooling I overdid the ventilation with holes for air circulation on the top, bottom and rear panel. I covered these holes by gluing a mesh used for windows and doors inside the panels.

I attached the front and back panels to the bottom panel by polyurethane construction adhesive which is strong enough not to need screws. The surfaces of the panels that were glued were roughened to remove the laminate coatings.

I attached the front and back panel components and then started cutting scraps of wire to connect the parts as shown in the circuit diagram previously. Won't go into too much details but the main steps were

1. Attached the fan into position to keep it out of the way (the fan was directly wired to the ATX board)
2. Then connected the live terminal from the AC IEC input socket to the fuse and then from the other end of the fuse to the power switch.
3. The power switch wire was then soldered to the ATX Live in wire.
4. I used a switch with an in built neon lamp (and resistor) so the third terminal of the switch that is connected internally to the neon lamp was soldered to neutral.
5. The ground wire from the IEC terminal went directly to the ground point on the ATX board and to the green binding post on the front panel.
6. The neutral wire from the IEC socket was connected to the ATX Neutral in wire and a small wire was tapped off the neutral wire to the neon indicator in the power switch.
7. The plugs at the end of the DC wires from the ATX board were cut off and the wires were bundled together by color.
8. The 3-way binding posts were selected to match the wire colors so hopefully the wiring would be a bit easier.
9. The orange wires were braided together, ends stripped and soldered to a ring terminal.
10. The red wires were also bundled together and soldered to a separate ring terminal.
11. One red and black wire pair was kept separate to power the volt-amp meters.
12. One black wire was saved for one end of a power resistor in the output. In the previous PSU conversion, the power supply would shut down if there was no load across it therefore the resistor in the output.
13. Another two black wires would connect to the DC on switch and to the DC on LED.
14. I started attaching the single wires for the -12V (blue) and -5V (white) to their respective binding posts.
15. Then attached the orange 3.3V bundle to the orange (yellow right now but ...) binding post.
16. The red 5V bundle to the red binding post.
17. The DC-on switch was connected to the green and black wires from the ATX board.
18. The +ve terminal of the DC-on LED was connected through a 330 ohm resistor to the gray wire from the ATX board and the other end of the end was connected to a black wire.
19. The remaining black wires were bundled together and soldered to a ring terminal and then attached to the black binding post.
20. I kept a separate orange wire and a red wire free as was not sure which wire I should be attaching to the power resistor. I found out I did not need a power resistor across the output so removed the resistor.
21. The volt-amp meters were slotted into the front panel and the plug in the back with thin red and black wires were connected to a single red and black wire from the ATX board.
22. I connected the supply to the AC power to see of everything was working. Yes, it was.

Unfortunately I did not take adequate pictures for these steps.

For the low voltage module ($4)

1. I first tested the module by connecting the input to 12V from my old PSU and then measured the output as I changed the variable resistors on the board. One var. resistor was for current adjust, other for voltage adjust. Seemed to work ok.
2. I carefully desoldered the two variable resistors with a hot air iron. The resistance on these was 10k ohms so soldered pins into the original holes on the boards and then soldered wires to the pins that were connected with standard 10k potentiometers. Tested these again to make sure that the module worked.
3. I will later connect the module to 12V and 0V wires from the ATX board and place the potentiometers into the front panel.

For the high voltage module ($12) I ran into problems

1. The module worked ok when connected to 0 & 12V with an output of 11-60V when the variable resistor was turned from zero to maximum resistance.
2. I desoldered the two variable resistors, measured these at 100k so replaced these with external 100k potentiometers. When I tested these again, the voltage went up to 90+ volts! As the capacitors at the output were labeled as 100V I was too scared to let these guys stay at 90V. Tried different pots, same results, tried an extra boost module, same results. After eliminating soldering errors, etc., finally figured out that using a 50k potentiometer gave me a maximum voltage of 50V - quite a bit safer. For the current adjust, I left the 100k pot but current cannot be adjusted down to zero - will have to check later why not.
3. I will connect these to the 12V and 0V wires from the ATX board and connect the 100k and 50k pots to the front panel.

Tried different locations to place the two power modules in the 6 inch cube till finally found two suitable locations. One on the top panel for the low voltage module; and right next to the fan on the rear panel for the high voltage module. The DC in and the DC out wires were carefully threaded through the empty spaces to the power modules and then attached to the corresponding terminals. The shielded audio cable that I had used for the pots was then passed under the bottom of the ATX board and the pots were placed into their corresponding holes in the front panel and nuts were used to attach them to the front panel.

The volt-amp meter sense wires were attached as below (see the wiring diagram)

Thin yellow wire from the meter for the low voltage variable supply went to the low voltage red binding post. Another wire from this red binding post went to the + output of the low voltage supply. The negative output of the low voltage supply was soldered to the thick red wire of the volt-amp meter. The thick black wire from the volt-amp-meter was connected the black binding post of the low voltage supply.

The high-voltage module was wired similarly to its set of binding posts and volt-amp meter. Red binding post connected to yellow wire from volt-amp meter and to + output screw terminal of the high voltage module. Black binding post connected to the thick black wire from the volt-amp meter. Negative output from the high voltage module connected to the thick red wire of the volt-amp meter,

The modules were tested again.

The low voltage module was held in place with a screw. The high voltage panel was held in place with double sided tapes and hot glue. To be safe, a wooden support block was also added to hold the high voltage module in place.

The ATX board was then attached to the front panel with screws into wooden standoffs. A folded sheet of plastic was used between the board and the front panel to prevent any electrical mis-connections.

One more functional check.

The final set of images show how tightly packed all the components are. Electrical tape definitely required to keep the wires under control.

Voltages for all the binding posts were checked and were within tolerance. The low voltage variable supply behaved very well with good adjustment of voltage from approximately 1.2V to 10.6V that was constant even when the load was changed. Current adjustment also worked well.

The high voltage power supply showed about a 1V increase when there was no load across the output. And current adjustment could not go below 0.4A. The range of voltage was about 10.8 to 51 volts.

It is pretty impressive that you can light a regular 110V bulb from an ATX supply.

It finally took its place on my lab bench just above the old but still very useful ATX supply.