Making DC Power Supply from ATX PSU

Thread Starter

chedarmaster

Joined Oct 22, 2011
1
Greetings!

I'm thinking about doing a power supply for some projects to help me learn more about electronics and I have 2 options: a laptop brick or an ATX power supply.

I have read this article about converting an ATX: http://www.wikihow.com/Convert-a-Computer-ATX-Power-Supply-to-a-Lab-Power-Supply

and would like your opinion about it, please.

To me it seems that using the brick would limit me a lot, in terms of output voltages and also the available current for output (it reads: Max 60W). The ATX provides me more voltages to work with and more output power.

About this ATX project, because I have no formal training with electronics I also ask:

1. Those different -12V and +12V, -5V and +5V can really be used in combination to obtain 24V, 7V and so on, just like that? Or do I need to worry about the amperage output of each as well? Because I see written on the ATX: +5V 30A and -5V 0.3A. Does this mean I will only be able to use a combination of +5 -5 (i.e. 10V) with devices that use top 0.3A?

2. Can I use a big screw driver to discharge the 400V capacitor? Go ahead, laugh all you want, but I would like to know of a way. I have many resistors, but I heard one has to consider their level of energy dissipation (1/4W, 1/2W, etc.). How can I find one "strong" enough to discharge that capacitor?

Any usefull remarks are very wellcome.

Thank you
 

Adjuster

Joined Dec 26, 2010
2,148
DO NOT DISCHARGE A LINE RECTIFIER FED CAPACITOR WITH A SCREWDRIVER.

The capacitor should be drained anyway, but if by any chance it is still up, there will be a loud BANG, and the capacitor, the screwdriver, and possibly other things may be damaged. You may also get a shock.

A large wirewound resistor, perhaps 10kohm 10W might be safer. You could probably use a bit lower wattage as the use will be very intermittent.
 

SgtWookie

Joined Jul 17, 2007
22,230
I'm thinking about doing a power supply for some projects to help me learn more about electronics and I have 2 options: a laptop brick or an ATX power supply.
Why not both, if they're not being used for anything else? :)

I have read this article about converting an ATX: http://www.wikihow.com/Convert-a-Computer-ATX-Power-Supply-to-a-Lab-Power-Supply

and would like your opinion about it, please.
Ignore the part about "tape". Don't use "electrical tape" for insulation on anything; it'll get gummy and fall off after a couple years. Use heat shrink tubing, preferably a couple of layers. Use nylon zip-ties instead of tape to bundle wires together. You can get them in various colors, but the wires in an ATX PSU are color-coded anyway, so why worry?

You may want to leave some of the MOLEX connectors on the harness; they can be handy for testing things on the bench.

To me it seems that using the brick would limit me a lot, in terms of output voltages and also the available current for output (it reads: Max 60W). The ATX provides me more voltages to work with and more output power.
Sometimes you just need one supply, but a converted ATX supply is quite versatile.

About this ATX project, because I have no formal training with electronics I also ask:

1. Those different -12V and +12V, -5V and +5V can really be used in combination to obtain 24V, 7V and so on, just like that? Or do I need to worry about the amperage output of each as well? Because I see written on the ATX: +5V 30A and -5V 0.3A. Does this mean I will only be able to use a combination of +5 -5 (i.e. 10V) with devices that use top 0.3A?
You really should look for a supply that has a bit more current available for the negative voltages. I converted an old 250W Compaq supply I had; the -5v and -12v are rated for 800mA. Not terribly surprisingly, they use 7905 and 7912 regulators respectively for those voltages.

And if you want to use both a positive and negative voltage, then the limit will be the output with the lowest current rating.

BTW, my Compaq supply will trip during turn-on if there's a load connected between +5v and +12v, but it works fine if you connect it after it has stabilized.

2. Can I use a big screw driver to discharge the 400V capacitor? Go ahead, laugh all you want, but I would like to know of a way. I have many resistors, but I heard one has to consider their level of energy dissipation (1/4W, 1/2W, etc.). How can I find one "strong" enough to discharge that capacitor?
You could use a 120v night light bulb. Light bulbs are really good for discharging caps, as they give a visual indication as well as discharging the caps pretty quickly; as their resistance is not fixed; when the filament is cold, it's just a few Ohms, but when at operating temperature, it may be hundreds of Ohms.
 
Last edited:

K7GUH

Joined Jan 28, 2011
190
Converting an AT supply is not complicated. There are several sources of accurate info on the web. Read two or three of them, pick the one you most readily understand. Don't worry about current limitations until you need a voltage which is not provided in the specs. The incandescent bulb used to discharge a capacitor is one of the safest methods available. Although I have used a Christmas tree light, I prefer a 25 watt (or greater) ordinary incandescent bulb. If you can locate or scrounge a fixture to mount it in, that's also a good idea. Use only one hand to connect wires, keep the other in your back pocket.
 
Greetings,

I am currently trying to convert a Shark Tech. 1200W ATX PSU and have come across yellow/green wires (Protective Ground?) and have no idea how to treat them.

Any help would be greatly appreciated.
 

Evil Lurker

Joined Aug 25, 2011
116
I have done both, and IMO laptop bricks hooked to either a linear or switching dc regulator are far superior as the are smaller and will allow for higher voltage applications wheras any decent ATX psu will have too many protections that will need to be removed before you can change its output voltage by more than a couple volts tops.

Furthermore, with an ATX PSU there should be nothing to change at all on the inside. Get yourself a dead or obsolete motherboard from a computer repair shop, yank off any electrolytic capacitors (save the ones that are not leaking or bulged if you can) since some of the smaller ones can blow up, then take a heat gun to the solder side of the power input header (Harbor Freight $10 on sale) and start heating the board up while giving it the occasional whack on something. If you do it right it should pop right out after the solder has liquified when you whack the board. Don't even bother with using braid or a pump... motherboards are multilayered PCB's using high temperature lead free RoHS solder along with tight fitting through holes and the ground plane on the inside will suck the heat away from the joint long before you will ever get them off with anything but a beast of a soldering iron.

Once you get the header in hand you can buy a perfboard PCB, do a bit of drilling with a dremel rotary tool to get it to fit, run some wire traces to proper terminal block headers or whatever, and slap on a power switch that connects the green wire to ground. Also you should take note that if your just piddling with 5v applications using less than one amp you can run them off the standby rail without actually turning the entire PSU on. Two other things... most ATX PSU's like to have a minimum load on them so on your breakout board consider slapping some sand bar resistors on it... this will also help increase your output voltage and that if your PSU has a 5 or 12V sensing wire on the motherboard connector you will have to attach a load to that or else the PSU won't come on at all. One last thing... unless your actually going to use 1200W worth of power the PWM controller will have to cut the duty cycle so low it might not even run stable if its too low.

With a laptop brick all you have to do is slap a LM350 or the like three terminal regulator on it an you can have whatever output voltage you want, or, if that isn't efficient enough (linear regulators put out a crapload of heat) you can go with a switcher like the LM2575 or some other higher rated amperage in its family. If you really want to get a wide range of output voltage you can go with a HP power brick which are normally rated at 32V and set your regulator up accordingly.

Hope this helps.
 

Rubnalq

Joined May 11, 2012
4
Hi! sorry for resurrecting this thread, however I'm sure this belongs to this thread:

I'm currently thinking of converting one semi-old ATX power supply to use voltages between 0 and 12 volts for starting in electronics! the thing is I don't have too much knowledge yet so here it goes:

1. The PSU is a 500W 2009 one, from one old computer I repaired that didn't want to start with this power supply, so it may be damaged. My plan is not to drill or open the power supply yet, just short the green wire to a black one and use one of the colored cables (for example the 3.3 or the 5 volt ones). Is this too risky considering the old pc didn't start?.

2. The etiquette in the PSU says for example that the output of the 3.3V rail (orange) is 30A. If I get a diode that works with around 3 volts and connect it to the orange and a black (ground) what will happen? 30 amps sound like a very big current, will the diode die on the spot using maximum current possible? or will it only use the current that it needs? what if connect a small circuit to those 3.3v, if the circuit uses for example a couple 1/2 W resistors... the circuit will die on the spot too? or just use the needed mAmp to work?
 

MaxHeadRoom

Joined Jul 18, 2013
28,617
The common or black is usually at ground potential already in the supply.
Also the 500w pertains to the 5v supply, the 12v supplies are usually much less wattage/current rating.
Max.
 

paulktreg

Joined Jun 2, 2008
833
Max

I'll have to correct you there. The 500W figure is for all rails combined.

Modern quality ATX power supplies, this may include this 2009 model, are designed very differently these days and the full wattage rating is often available via the +12V.
 

wayneh

Joined Sep 9, 2010
17,496
If I get a diode that works with around 3 volts and connect it to the orange and a black (ground) what will happen? 30 amps sound like a very big current, will the diode die on the spot using maximum current possible? or will it only use the current that it needs? what if connect a small circuit to those 3.3v, if the circuit uses for example a couple 1/2 W resistors... the circuit will die on the spot too? or just use the needed mAmp to work?
By "diode", do you mean LED? Do NOT connect an LED (or diode) directly across the power supply without something to limit (control) the current flow. Otherwise, it may explode in your face and/or release the magic smoke.
 

Rubnalq

Joined May 11, 2012
4
By "diode", do you mean LED? Do NOT connect an LED (or diode) directly across the power supply without something to limit (control) the current flow. Otherwise, it may explode in your face and/or release the magic smoke.
I guessed this, however, if my question is this: if its a led and a 5k resistor, what can I expect in terms of amps? (again 3.3 rail and ground rail).

Will the resistor die too? when will the resistor not die?
 

AnalogKid

Joined Aug 1, 2013
10,986
I've never been a big fan of this approach. The price is right, but getting that price means a design that is so highly optimized for a single task that it can be cranky when pressed into some other kind of service.

For example, most atx supplies have a minimum load requirement. This makes the magnetics (and hence the price) cheaper. Usually the +5V output or the +5 and +3.3 combined must be loaded to 10% of their rated current before the magnetics stabilize enough for the other outputs to come into regulation.

And to be clear about one of the earlier questions - all of the outputs are common-grounded. That means that you can combine any positive output with any negative output to get other effective output voltages. but you can not combine outputs of the same polarity. +12 and -5 will get you 17 volts; +12 and +5 will get you smoke.

ak
 

Rubnalq

Joined May 11, 2012
4
I've never been a big fan of this approach. The price is right, but getting that price means a design that is so highly optimized for a single task that it can be cranky when pressed into some other kind of service.

For example, most atx supplies have a minimum load requirement. This makes the magnetics (and hence the price) cheaper. Usually the +5V output or the +5 and +3.3 combined must be loaded to 10% of their rated current before the magnetics stabilize enough for the other outputs to come into regulation.

And to be clear about one of the earlier questions - all of the outputs are common-grounded. That means that you can combine any positive output with any negative output to get other effective output voltages. but you can not combine outputs of the same polarity. +12 and -5 will get you 17 volts; +12 and +5 will get you smoke.

ak
Okay... so +12 and +5 will not work... but what about the amps? if I have +3.3V with 30A output, this means I need at least a 9.9 watt ressistor right? (10% of 99 Watts?).

If I put in a circuit that is very simple afterwards it will use just the needed amps or the full 30amps will run throught and burn everything?
 

AnalogKid

Joined Aug 1, 2013
10,986
All of the outputs function as constant-voltage sources. They make their output voltages no matter what the current draw is, as long as it's not too little (minimum load) and not to much (overload).

A 30 amp output means that that much current is available if the load draws it, not that that current is forced through everything. If the load is 1 ohm, or a bunch of electronics that is equivalent, then the current from a 5V output will be 5 amps, and the current from a 3.3V output will be 3.3 A. Read up on Ohm's Law.

If you're sure your electronics will draw the minimum load current, no other external resistor is required.

ak
 
Top