Converting a PC power supply (ATX) to a Variable Power supply.

[raynerd]

I was in need of a 5v power supply for my microcontroller circuits as the supply I have was not working with them for some reason and I was not happy keep using batteries. There are many write-ups online so if your interested, do a google and don`t trust this. This is just a log of what I am doing and have done so far. This is a work in progress although I do now have it working as a +5v, -5v, +12v, -12v supply. I hope to include a 3.3v supply and also a variable pot to have a true range of supplies.

This is the inside of the PSU and looking at other examples, mine seems jam packed with components. It made it hard to decide where to mount my switch and sockets. Not ideal but it is going to have to be on the side of the unit but I prefered this to other examples where people had been in similar situations and taken the plugs to an additional piece of angle iron bolted to the top of the supply. At least this it will all be contained.



The main jist, is that when you take a few hours   to untangle the mass of wires, there is:

red: +5v
white: -5v
yellow +12v
blue: -12v
black: gnd
gray: DC-OK  (power on)
green: PS-on (turn-DC on)
Purple - VSB+5v (standby)

Here is the 10w, 10ohm resistor connected across the 5v supply to ensure there is always a load so the supply does not shut down with nothing connected.



And once I`d added a few posts to the case along with the Power On LED:



And here it is providing a voltage as close as dandy to 5v:



Still quite a bit to do but that is where I left it this evening. Clearly the top is not down fully but I want to finish it before I start to use it. I need to differentiate between the various voltages, also I am ultimately going to have +12, -12, +5, -5, +3.3 volts, and also ground. That totals 6 binding posts but I think because of the large current this can provide, I guess you could run multiple sources therefore is it worth having a couple of grounds?

Then I need to think about this variable voltage pot. Again lots of stuff on the net regarding that but it would then make it very useful! Once finished it should give me direct: 24v (+12, -12), 17v (+5, -12), 12v (+12, 0), 10v (+5, -5), 7v (+12, +5), 5v (+5, 0) and then fully variable 0-11v.

Chris


 

[websterz]

Been running my x2 powerfeed on a modded ATX for 6 or 7 months now...works a treat. 

[Bernd]

Chris,

The gray,green and purple. What voltage is supplied to those three wires to make the power supply work without being plugged into the mother board?

Bernd

[michvhf]

You need to put a load on the 5V line (Chris is using the 10 ohm 10 watt resistor for this), then you should only need to ground the green wire.

[Bernd]

You need to put a load on the 5V line (Chris is using the 10 ohm 10 watt resistor for this), then you should only need to ground the green wire.

Thanks. I figured it was something like that.

Bernd

[madjackghengis]

Hi Bernd and interested parties, I've built two different Vellman electronic kits which use solid state chips to make variable speed "chopper circuits" specifically for driving motors with full torque and variable speed.  One is specifically designed for twelve volts, and is limited to 1 amp of load current, the other is useful from five to thirty five volts, determined by the power supply used.  Both are easy to build and quite reliable and responsive.  They were built to use to drive a surplus wiper motor for a power feed on the Y axis of my mill, and one to drive a drill-driver motor whose batteries gave up the ghost, to drive the top slide of my lathe.  The wiper motor works fine, but I have not yet connected the other, as I was looking for a power supply for other than twelve volts, and now have an answer.  Both kits are very easy to assemble, and are well proven out circuits that are reliable, and give repeatable performance.  I will be trying out the second one on power out of an old computer power supply, and wish I'd thought of that first.  Vellman also makes an electronic ignition kit for converting points to electronics, which allows one to use micro switches for "points" without having the ignition current burning out the micro switches.  It also works on differing voltages, although I have not tried it to find how low one can go and still have reliable sparks.  I have a couple Jap motorcycle coils which are about 0.9 ohms, and will try them on five volts through the vellman kit and report back as soon as I have five volts to try it with.  Mad Jack

[raynerd]

Bernd, sorry for the delay, I read your question this morning but I wanted to check something on the power supply before I answered it and I`ve only just got into the shop now.

Green - PWR_ON is an active high and must be grounded for the psu to turn on DC. For testing I just connected to gnd and it boots the supply right away however on the conversion it is the green that the switch is connecting to - grounding it to turn the supply ON.

Purple is +5v Standby - this line is ALWAYS 5v whilest you have a AC power supply to the AXT unit. It can be ignored or connected to an led as I have to show power is being fed to the PSU.

Grey - PWR_OK - this goes to +5v when there is a supply to the DC power lines.  I`m going to add a second LED here as this is a confirmation that the power lines have a feed.

To test the PSU, I simply grounded green and purple and attached a load across the +5v. So that you don`t have to have anything connected on the actual thing when it is up and running, I`m using the 10w, 10ohm resistor.

Hope that is of some use.

Chris

[Bernd]

Thanks Chris. That's exactly what I needed. Thanks also goes out to the others that answered. Much appreciated.

Bernd

[DMIOM]

.......
Green - PWR_ON is an active high and must be grounded for the psu to turn on DC. For testing I just connected to gnd and it boots the supply right away however on the conversion it is the green that the switch is connecting to - grounding it to turn the supply ON.

Purple is +5v Standby - this line is ALWAYS 5v whilest you have a AC power supply to the AXT unit. It can be ignored or connected to an led as I have to show power is being fed to the PSU.

Grey - PWR_OK - this goes to +5v when there is a supply to the DC power lines.  I`m going to add a second LED here as this is a confirmation that the power lines have a feed.

To test the PSU, I simply grounded green and purple and attached a load across the +5v. So that you don`t have to have anything connected on the actual thing when it is up and running, I`m using the 10w, 10ohm resistor. .......

Chris - a couple of points :

1/ If the Purple is the +5v standby (which is what is often used to power the Ethernet port and was used for the serial port UART, so you can have "wake on LAN" and used to have "wake on Ring") then that +5v is, as you say, live all the time - so grounding the Purple is in fact putting a dead short across one of the supplies!

2/ You mention the Green is grounded to turn the PSU on, but describe it as Active High? I suspect it has a pull-up so it floats high, but is actually Active Low? (i.e. take it low / to ground) to operate.

3/ Have seen in photos you're fitting terminals etc. in through sides of the case - I'm sure you're careful but just in case anyone else doesn't know, switched mode power supplies can have lethal hundreds of volts DC in them .... 

Dave

[raynerd]

Dave, 1. Yes, in retrospect, think I should have connected across LED even when testing. 2. Green is called Active High on a few sites but I see what your saying.

OK, carrying on today, I realised that in actual fact, once I`ve made this I don`t want to keep pulling it apart and adding to it as was my initial thoughts. So, if I`m going to try and build a circuit that gives me a variable supply as well, I need to do it now so that I can correctly position the rest of the components and front binding posts and have this finished. I wanted to add a fully variable output and followed these notes to build a suitable circuit on breadboard for testing purposes:

http://www.wikihow.com/Add-Variable-Voltage-to-Your-ATX-Based-Bench-Power-Supply

The main component is the LM317T voltage regulator. Here is the trial circuit:





The pot is quite sensitive and I need to sort this out as it is difficult to control the voltage but it is stable once you get it there. This is the lowest voltage possible 1.26v


At 10.00V

At maximum - 19.81V. Really this should be 24v so I think I need to change the load resistor.

Again, this is just a log of what I am doing…google and Do Your Own Research first!

[andyf]

Chris, it might be an idea to build the LM317 circuit in a separate disconnectable box, so that if you ever decide to use the ATX-based PSU to power a motor at variable speed, you could easilly substitute another box containing one of the PWM controllers that Mad Jack suggests.

I understand that LM317Ts handle 1.5A for sure, 2.2A average and over 3A if you are lucky enough to get a good 'un. They can be paralleled for more oomph as shown on page 15 here: http://www.datasheetcatalog.org/datasheet2/1/03cgthpfat4t4ly5kfp5lpwladfy.pdf
 
Andy

[raynerd]

That is actually a really super idea - it`ll just plug straight into the -12 +12v binding posts! Hey, it would even make a nice PIC project if I added a LCD to display the voltage! Cracking idea and I`m also running low on space in there.

[75Plus]

Chris,

I posted this variable power supply a few weeks ago. It uses 2 LM317T regulators, one controls the voltage and the other controls current. The circuit is all hard wired and no circuit board is required. You can omit the transformer and feed it from the ATX supply. Using +12v and -12v as your input you will have a greater range of voltages available. Actually from 1 to 21 or 22 volts. Have a look.

http://madmodder.net/index.php?topic=2551.0

Joe

Andy and I were typing at the same time thinking the same way.

[DMIOM]

Chris, it might be an idea to build the LM317 circuit in a separate disconnectable box,....

Another benefit of doing that would be that if you used, say, a diecast box you could use it to heatsink the regulators

Dave