Yet Another ATX Lab Bench Power Supply Conversion

This project builds upon the ideas of a previous instructables project: http://www.instructables.com/ex/i/D5FC00DAB9B110289B50001143E7E506/?ALLSTEPS
The big difference is that I decided I didn't want to destroy my ATX power supply in the conversion. Another advantage is that you can just plug in a new power supply if the old one goes caput.

Tools:
Drill and bits
Screwdriver ( for taking project box apart)
Soldering iron
Desoldering iron (optional)
Label maker (optional)

Materials
Banana jacks
Project box
A lighted 12V switch
3/4" strain relief
ATX connector (purchased or scavanged)
Miscellaneous heat shrink tubing sizes
Some wire 14-20 Ga (colors matching standard ATX if possible)

Step 1: Obtain an ATX Motherboard Connector

For my power supply I desoldered an ATX motherboard connector from a fried motherboard. An easier, less time consuming, but more expensive solution would be to buy an ATX extension cable or an ATX 20 to 24 pin adapter for about $10.

Desoldering components can be tricky, so take your time. The easiest way I have found is to first flood all the pads with as much solder as they'll hold. This allows me to heat the entire pad and pin evenly with the desoldering iron. I use a simple Radio shack model with a squeeze blub.

Comments

[Skarz88]

So how does 5v and 3.3 give 1.7? I thought it would give 8.3? Same as how does -5 and +5 give 10? I'm a bit confused lol but it would be good if I could get other voltages by using different combos.

[benjamenjohnson]

Voltage is analogous to pressure difference. Image you have two sealed adjoining rooms with a connecting window: one at 1 atm (14.7psi) and another at 2 atm. If you'd open the window between the rooms the wind (or current) would be driven by the pressure difference between the rooms or 1 atm.

We usually think about measuring voltage with respect to ground (0V), but it doesn't have to be so. Voltage is the difference in potential between two points. It is only in the special case of using ground where you can just look at one of the points and call that the voltage. So to find the voltage you take the difference of the two points whether it be 12V - 0V = 12V or 7V- 5V = 2V or 5V - (-5V) = 10V.

[benjamenjohnson]

This only works with DC voltages. AC voltages are much more complicated because you need to take phase into account -- assuming the AC voltages are even the same frequency -- and whether you are measuring RMS or peak.

[Skarz88]

How did ya get the -3.3v? So goin by what volts are available, just mix and match according to what ya need? Like +5 and +5 for 10v?

[benjamenjohnson]

I'm not sure where you see -3.3V (Maybe where I say 5V - 3.3V = 1.7V). I only have +3.3V. I just pulled all the voltages right from the power supply connector.

Also, I'm sure you just mistyped it, but you'd need -5V and +5V to get 10V

[omnibot]

Nice! Could also be a handy tool for testing PSU's.

[looking4ideas]

OHH i just thought of this: Why not make a small cut on ur PC case and use the PSU While u still can use the com?

[davidprosser]

Depending on what you use externally and what your computer is running, you might not have enough power to do both :P

[looking4ideas]

use one of those 2kW PSUs

you power it on and all the lights will go dim in your house =)

[looking4ideas]

A 2 in 1 deal

[looking4ideas]

Why dont u have a 7v? its easy to do.

[benjamenjohnson]

But I do have 7V. I also have 1.7V and 24V. You can access all of these voltages with paired banana plugs too. You just have to be careful that the circuit you're connecting isn't tied to ground.

[Crazy_Bee]

Hi, is it safe to use 24 V with this project? Would the regulator take a beating? If it is safe id be interested to know how to do it. Do i just connect -12 and +12 together? Thx for the great tutorial !

[looking4ideas]

Ok i was just giving an idea =)

[benjamenjohnson]

I have to admit that my first thought was to say "Hey dummy, take another look and do some math!'" But it was only a few years ago that I had heard of the 12V -5V = 7V trick, so that seemed a little hypocritical.

But the more I thought about it the more I realized that your comment was a good opportunity to add information that might not be immediately obvious to everybody .

So:
12V - (-12V) = 24V
5V - 3.3V =1.7V
you can't use paired plugs but
12V - 3.3V = 8.7V
etc...

Just be careful using pulling any of these non-rated voltages out of the supply, it wasn't designed from it and there's no guarantee that the voltage will remain stable if you pull any sort of current doing so. I would recommend piggybacking a voltmeter to watch the voltage if you're going to use this trick.

[davidprosser]

can anybody tell me if it's alright to not have minus terminals, but take the lowest pin (-12v) to be ground, and work the rest up from there. Then I don't have to calculate all the other voltages. Anybody get me?

[DualPhase]

Clean solders dude. nice job

[Myself]

The equidistant-from-Ground layout for paired banana plugs is genius! Add a second ground point up near the 'O' of ON, and you could run independent paired cables to +5 and +12 loads. If I built one I'd probably have multiple sets of +5 points. Add a USB jack or two, with just the power and ground pins connected, so you can use those little flex lights on your bench. I find 'em really handy for fine soldering work when the weather's too warm to get out the big incandescent. They also make fine night lights, and you might bring out the +5 standby line for an always-on USB socket for just that reason. Toss in a few cigarette lighter sockets, and your handy-dandy 12-volt source just got even handier. (I'm not sure what would make it dandier. Designer shoes?) Although I guess almost anything you might charge from a car is also available with a USB charger now, eh? But half my 12-volt gear has Anderson Powerpole connectors on it now anyway, so I'd have to include a few of those... (While writing the rest of this comment, I realized I'd strayed far into overkill territory. So I came back and added this warning: WHAT FOLLOWS IS ABSURD.) For fan testing, I'd want a variable voltage output, so I can ramp up from nothing and see the lowest voltage at which the fan will start, or back off once it's started, and see the lowest voltage at which it will remain spinning. Also a milliammeter for load figuring, and a pulse counter on the third pin to test the internal tach sensor. Hook up a car's Mass Air Flow sensor and it'd become an extremely versatile fan evaluation rig. Gather and plot all that data, and ... (Cheat: Buffering the tach line and feeding it into a soundcard, then doing frequency counting in software, might be a cheap way to figure RPM.)

[rocketman221]

I try to avoid using those cigarette lighter sockets they are just a pain to work with and the plugs never make good contact. binding posts are the best because you can connect a bare wire to them or use bannana plugs.

[_soapy_]

You could simply count the blades, then use a microphone to listen to the fan, determine the frequency, and then divide by the blade count. This would work using your regular microphone, rather than messing about.

[Yerboogieman]

i've seen those switches, do they really light up?

[benjamenjohnson]

Yes, as long as you buy one that has the right voltage bulb for the application. In this case I used one with a 12V bulb. The one I used isn't super bright, just bright enough to see that the power supply is on.

[cvxdes]

It should say how many amps you can draw from it on the bottom, or somewhere. Mine was 1.5 amps, it's an old 60 watt from like 1999.
Anyways, I teamed mine up with the Third Hand++, to power the fan and LEDs that I have lighting it. Here's some pics of the one I made:

[drcrash]

I second the question "How many amps can you draw from it." I would like to heat up a shaped piece of stainless steel wire (thick enough to hold its shape, plus a bit) by running lots of amps through it. (This is a hot wire cutter for cutting plastic, but not the usual kind of thin-taut-wire thing.) I built a supply from a 350-watt ATX, rated at 30 amps @ 3.3 V. Can I actually use nearly that many amps? I could get by with half that, maybe even a third, but 20+ amps would be much better.

[Mr. Smart Kid]

what the 5v line i use that for my heater thiggeyss

[benjamenjohnson]

The highest amp load I've pulled from this gizmo was when I used it to try to shock some NimH AAA cells with the 12V taps, ala another instructable. I didn't measure the current but the supply overloaded and shut down. I was lucky it reset itself after I unplugged it for about a minute. By the way shocking the AAA cell worked for about 5 minutes. It measures .5V before and 1.4V after shocking it, but after about 5 minuted use the voltage dropped back below 1V.

[Derin]

i shocked lead acid wit the charger no luck

[Mr. Smart Kid]

NOT NIMI!!!!!!!!!!!!!!!!!!!!1

[wa3rey]

>By the way shocking the AAA cell worked for about 5 minutes. Did you charge the battery after shocking it? Shocking the battery blows out the short circuits in the battery but then you must charge it using a conventional charger.

[Mr. Smart Kid]

btw

[benjamenjohnson]

I failed to mention that I tried charging the cells in my archos with the Rockbox firmware installed after I shocked them. Rockbox showed the voltage of the 4 AAA cells when charging. Normally this voltage is about 5.7-5.8 volts throughout the charging cycle. After about 5 minuted the voltage started showing 4.8V. I removed the batteries and tested the 2 AAA's I shocked, they were reading less than 1V. I concluded that they weren't going to take a charge. I'm sure it works for some cells, but it didn't work for mine. No harm trying.

[drcrash]

I would think that it you want high amps, you shouldn't cut most of the same-colored wires and only use one or two. I'd guess that you need several conductors to handle the amperage of the ground, +3.3, and +5 volt connections. I did use a minimum load resistor. Rather than using one 10 ohm resistor, I used two 5-ohm resistors in series. That way, the same amount of heat gets distributed over twice as much resistor, and it doesn't heat up as much, so I don't have to worry about heat-sinking them. They get warm to the touch, but not hot.

[n0ukf]

No mere guess there, you do need more (or thicker) wires to handle more current. If you test your PSU with various load resistors, you may find you can use a higher resistance and get by with using a lower wattage resistor (this leaves more current headroom for the project you're powering).

[Avi_Y]

A neat job. How much current can you draw from it? Forgive my ignorance, and I do know these suppliers are supposed to work with computers, but would it fit for using other sensitive electronics? I would like to build such a supplier as a power source for a computerized DC charger, and was wondering if it would work. Thanks, Avi.

[Mr. Smart Kid]

yes you can, amps deapend on the power supply

[drcrash]

The really easy way to do this is to start with an ATX cable of some sort, with all the right-colored wires in place, rather than an ATX connector and various colored wire. Just cut the cable, and you have all the wires. (I got an ATX extension cable for $2 from a bin at the Goodwill Computer Store.) Then it's mostly a matter of soldering the wires together by color, and connecting them to the binding posts. I would not recommend cutting most of the wires of each color, if you're planning to draw serious amps. I'd solder together all the wires of each color. (In a tree pattern, rather than trying to solder together a whole bunch of wire ends in a lump.)

[gabolema]

Hi! What would be the difference (besides the current) between that and a variable transformer (the one we with our electric appliances (1.5 to 12V) . Couldn't the transformer be used as well? (they aren't that expensive)... Thanks!

[benjamenjohnson]

Sure you can use one of those, but here are some reasons you would want to use this over a variable transformer from say Radio Shack. 1) Current. 2) Those variable transformers might not be regulated or at least not regulated very well -- they read 5V when you measure them, but they drop in voltage as soon as they are connected to a load. 3) Negative voltages 4) Multiple voltages at one time. 5) Banana plugs are standard test leads. 6) It's just convenient sometimes to be able to plus into 5V without fiddling with a tiny switch. 7) This isn't very expensive if you have an old power supply lying around, in fact making a cheap alternative to a $100 dollar or more bench power supply is the point of this instructable.

[radiorental]

very cool, will definitely do this and add in 'myself' suggestion on cig & usb sockets - thanks

[godsdog]

Hey, this is great. I think I'm gonna drill me up one. Thanks. I am thinking about adding a 4 pin molex for connecting IDE devices.