M42club.com - Home of the BMW E30/E36 318i/iS
FAQ / REFERENCE => How-To's => Topic started by: DesktopDave on October 31, 2012, 02:44:44 PM
-
The weather is pretty dreary here now, so I figured I'd spend some quality time converting an old, useless PC power supply into a useful bench-top power supply. I've been redoing the shop and a bench P/S is a great diagnostic tool to have. Often with electrical parts, eliminating the unknowns can solve the problem. With a good, known, clean 12VDC you can eliminate all sorts of problems right off the bat.
And you can do it in the comfort of your own man cave. ;)
It's easy to do, but now for the warnings. First, the capacitors inside any modern PC P/S can store enough power to kill you. Please keep this in mind, and be safe. Safety first! Leave the power supply unplugged for a day or two before taking it apart; only have one hand inside the box at any given time, wear gloves, etc. Second, while I've tried my best to make sure this is typed up correctly, a typo or two could really be a big problem. I assume no responsibility for what you do with this information...I'm just putting it out there to share since I'm a (fairly) nice guy - I love to help out - but any mistakes/misunderstandings/errors/guesses/estimations, etc. are entirely your responsibility to double-check. :mad:
Having said that, I'm pretty confident it's all correct. :cool:
Tools I needed included a decent soldering iron (>20watt does a great job), electrical pliers, cutting dykes and a cordless drill with bits.
Supplies I used included 60/40 solder, shrink tube, small crimp-on loop terminals, "liquid" electrical tape & BMW-style fabric 'friction' tape.
Parts I needed include a working desktop PC ATX power supply (especially one with a dedicated power switch - free salvage :D), a set of 4-6 multicolored 4mm banana clip posts ($5 on Ebay (http://www.ebay.com/itm/400196339809?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1497.l2649) :rolleyes:), a big 10 ohm resistor rated 10watts or better ($9 on Ebay (http://www.ebay.com/itm/320757994835?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1497.l2649)). I plan on adding a few LEDs and matching resistors just for kicks later on. So what did I get for a couple hours of research, a bit of work and $14? Well, one more useful tool. And who ever has enough of those? :D
-
Next, I'll get into the theory of the whole deal. It's pretty simple. All PC P/S output a variety of DC voltages to power the accessories in any PC. Those voltages include +12, +5, +3.3, -5 and even -12. Combining these voltages can effectively test a lot of battery-driven items, ranging from 3volt flashlights all the way up to small 24 volt loads. Just be sure not to exceed the output amperage you have available. It'll be posted on the side of the supply:
(http://i729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/th_2012-10-28182837.jpg) (http://s729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/?action=view¤t=2012-10-28182837.jpg)
On this supply, I can test 12v loads up to 120 watts (12vdc*10amps=120watts). I would test headlights or HIDs with this, but not a starter motor. However, testing 24v loads (connecting +12v to the load's positive lead and connecting -12v to the load's negative lead) maxes out at only six watts. I won't be testing any golf cart motors with this...
Furthermore, any decent quality P/S will have resettable solid-state power protection (polyfuses), so a short won't kill the P/S. It should just turn itself off. Keep in mind that the Chinese kids that originally assembled these units aren't really interested in your safety. So safely clear the short, wait for the polyfuse to cool off, cautiously power the P/S back up & test it for proper output. Then you should be ready to go.
The big downside is that these P/S don't usually have any type of current limiting. That's one of the many reasons why a decent bench-top supply costs so much. What does this mean to you? Well, a short can put an entire rail's rated wattage into your newly-created circuit. So mixing this P/S with parts of your body, puddles of water or the great outdoors is a no-no. If you complete a circuit with your own body...well...hopefully you'll live & learn not to do that again.
The idea is basically to cut all of the wires off the outside of the P/S and rewire it in a safer way. There are a few considerations to make sure the P/S will function. Most of my ideas came from this article (http://www.wikihow.com/Convert-a-Computer-ATX-Power-Supply-to-a-Lab-Power-Supply); this one is even sweeter (http://hackaday.com/2010/12/09/atx-psu-turned-into-an-adjustable-voltage-bench-supply/)...here's an even better solution, and only $37 shipped! (http://jordandsp.com/ATX-bench-top-power-supply-adapter.php) Anyhow, there are about ten thousand other tutorials out there. Read several of them to be sure you have an idea of what you need to do.
1. Make sure you have a generic ATX power supply. Total wattage is not that important...IMHO anything over about 150 watts will work just fine. Although OEM parts like Dell and Apple can be made to work, they occasionally do not follow international standards regarding wire colors and patterns. I'd avoid them.
2. All P/S will have a sticker on the exterior rating the output of each voltage. These are also called 'rails,' be sure you note which voltage carries the highest rating. Typically it'll be the 3.3 or 5 volt rail. Note whichever one it is - the P/S usually needs a small load on that rail to properly regulate the other rails.
3. All ATX P/S will have a dedicated wire that tells the P/S when to leave standby and go to full power. This is almost always a green wire. When grounded, this signal will power the unit up.
4. Most all P/S will also have a brown or orange 3.3v 'sense' wire. This must be connected to the normal 3.3v bundle, or else the P/S will power back down.
-
OK - directions out of the way - onto the pictures!
Step one - locate a decent P/S, test to see if the PC works, then leave it unplugged for a few days. Pull the P/S, cut the warranty labels (what, no warranty on a fifteen-year-old P/S?), unscrew the top cover. You'll see something like this:
(http://i729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/th_2012-10-28122825.jpg) (http://s729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/?action=view¤t=2012-10-28122825.jpg)
Next, I decided that I was going to pull most of those wires out. This was an old unit so it still has decently sized 16-gauge wires. Since 16-gauge can safely carry 14 amps @ 12 volts for short lengths (< 3 feet) (http://www.offroaders.com/tech/12-volt-wire-gauge-amps.htm), I figured two would provide a decent margin of safety but still be reasonably easy to fit into the crimp-on loop terminals I had. I unscrewed the main circuit board screws and lifted the board out a bit. I know it's overkill, but then I used a large screwdriver to short both primary capacitor's leads under the board with one arm behind my back. You can never be too sure about stuff like this - those big caps pack a hell of a punch!
The wires we're interested in are the yellow (12v), red (5v) orange (3.3v) and black (ground ,0v). I needed one extra red wire (load resistor) and two extra black wires (one for the load resistor and one for the power-on signal). Here's what it looks like now:
(http://i729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/th_2012-10-28131758.jpg) (http://s729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/?action=view¤t=2012-10-28131758.jpg)
-
Next, we have to complete the internal mods so that the external power switch will turn on the unit.
First, the green wire has to be spliced with a black ground wire. I used a lineman's splice, then soldered it, finally covered it with some nice heatshrink tube (any black wire will do, BTW):
(http://i729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/th_2012-10-28132624.jpg) (http://s729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/?action=view¤t=2012-10-28132624.jpg)
Then, the load resistor has to be spliced into the primary rail. On mine it happened to be the 5v rail, so I ran a red wire to one terminal. The other terminal gets a black ground wire. I looped the wires through the terminals, soldered it firmly, and used a bit of heat-shrink tube to insulate it nicely. Finally, I screwed it onto the rear inner wall of the case. It needs to be screwed down as it will be dissipating a bit of heat. IMHO don't screw it into the heat sinks...those sinks are not necessarily grounded, and in cheap P/S are barely sufficient to cool the existing transistors. Overheating those will kill the P/S and waste all your work. Here's a shot of the installed resistor:
(http://i729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/th_2012-10-28153711.jpg) (http://s729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/?action=view¤t=2012-10-28153711.jpg)
-
I'm planning on changing a few things once I make sure this all works, so I left quite a few wires for later uses. I sealed the ends with liquid tape, then taped them up to prevent any shorts.
In case you're wondering, the wires are as follows:
Grey is the "power good" signal - I'll be using that to light up a green LED. I didn't have the right 150ohm resistor on hand, so I left that for later.
Purple is the "standby" signal line. Likewise, I'll use that to light up a yellow LED.
Blue is the -12v line.
White is the -5v line.
(http://i729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/th_2012-10-28153722.jpg) (http://s729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/?action=view¤t=2012-10-28153722.jpg)
You can also see that I've stripped & twisted the other pairs before I crimp & solder loop (eyelet) terminals onto them.
-
I went the extra mile with these loop terminals, since they're not really designed to carry larger loads. First, I twisted two wires together for each circuit. Then, I soldered these splices together. Next, I crimped a loop terminal onto each one. Finally, I filled the terminals with solder.
(http://i729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/th_2012-10-28154335.jpg) (http://s729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/?action=view¤t=2012-10-28154335.jpg) (http://i729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/th_2012-10-28154357.jpg) (http://s729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/?action=view¤t=2012-10-28154357.jpg) (http://i729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/th_2012-10-28154425.jpg) (http://s729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/?action=view¤t=2012-10-28154425.jpg)
Finally, I wrapped each circuit in fabric tape:
(http://i729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/th_2012-10-28154808.jpg) (http://s729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/?action=view¤t=2012-10-28154808.jpg)
My next task was to prepare the case for the terminal lugs. Clearance was a problem - first off, I didn't want the lugs too close to each other. That creates the potential for shorting stuff out. Also, between the heat sinks, ventilation slots and internal components, I was left with about a 2" x 2" space. I drilled a few suitable holes & threaded the lug posts nuts on. Finally, I attached the circuits to the posts, and tightened everything up. Since I'm a really cheap bastid, I had to be especially careful with those posts. I managed to strip a pinch nut on one and had to leave it out. No -12v for me! If you consider the set I bought, I'd suggest getting two right off the bat.
(http://i729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/th_2012-10-28160250.jpg) (http://s729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/?action=view¤t=2012-10-28160250.jpg)
-
Finally...time to button it all up & test it out! I shot some video, 'cause I was feeling special. Here's a direct link (http://s729.photobucket.com/albums/ww299/desktopdave/Projects/Benchtop%20PS/) to the whole album.
The results? 3.37v...5.14v...11.96v Even better, the case reports 0v. :cool: No shorts from the posts, plugs or leads either. I'll report back on performance after I fire something up with it.
Close enough for gov't work. We have a winner! :D I tested it by firing up a Lexus HID projector I had laying around. Works just fine and I didn't get shocked.
Feel free to post some replies if you have questions, suggestions, or if you're an overachiever that's built a better mousetrap.
:D :cool:
-
Nice work, Dave. Nothing beats having a power source with a couple different voltages handy.
Could this be used as the power supply for an electrolysis setup?
-
I'd prefer to use a dumb battery charger, as they usually have built-in current limitation. Given the relatively weak electrolyte, I don't see why it wouldn't work unless you were doing large surface area items like engine blocks or whole body panels.
The only safety concern I'd have is a hard short. I'd definitely wire up a shunt ammeter/DMM in series to keep tabs on how much current the tank uses, or wire it off a protected GFCI circuit.
-
I'm using a battery charger my dad bought back in the early seventies. It has an ammeter, but that's a good suggestion for the GFCI circuit.
-
I should go over and pick up my Dad's old charger. It's much better than my cheap "Schumacher" 30 amp buzz box. It manages to be both noisy and inaccurate.