Modifying Computer Power Supplies

Computer power supplies are inherently fast switching hence the smaller mains transformer – not the linear type which is inefficient but serves as a great heater for winter months. Can these power supplies be used elsewhere – for instance to drive a car stereo system?  This is a very good question.
 To answer this one needs to look at the loading on the power supply and and what it was designed for. A computer power supply will almost definietly run a car stereo but the disadvantages are over current tripping. A car stereo will not draw a constant current because the amplifier requires changes in current to drive the loudspeaker and unfortunately these currents can vary from a few amperes to anything up to 50 to 100 Amperes.  The power supply will stall under heavy load conditions – it’s designed to do this. Unfortunately car stereos are often driven to it’s maximum potential and when hitting transients the power supply may not be able to cope.  A 12V rail feeding an amplifier which has been bridged (max 20W into 4 Ohm load) means that the user requires about 2A per amplifier * 4 or about 10 Amperes. So the minimum power dissipation one would be looking at is about 120W to run a standard, as advertised 25W * 4 car system and one isn’t even looking at the efficiency of this system.  But yes, in my experience the 20A 12V rail will supply the power required without tripping.  (not the same power supply feeding your computer though – this is inviting problems).   

What about modifying the power supply to run your home stereo system which requires higher voltages. Now this is something that only the experienced electronics buffs should try and should have a great deal of experience in the design and construction of switched mode power supplies. There are circuits shown on the web, one of which I link to but please be cautious. If you have had no experience with switched mode power supplies I would not advise you to attempt any project where you are looking at changing the output voltages.
First of all, the power supply is going to need some changes to the filtering, which means high voltage electrolytics – these, because of their capacity of possibly 10 000uFds and up have a very low reactance when looking at it from the output of the power supply side (you would be using lower capacity than 10 000 uFD BTW because of the SMPS switching frequency) – and in fact is pretty close to a short – you will need to charge these capacitors through a resistance network first and then bypass the network otherwise the power supply will trip. I mention this because often the experimenter overlooks this issue. Remember that you will be powering an amplifier so this should form part of the slow switch on circuit. ATX power supplies do need some sort of load initially to start up – I do this via the slow switch-on circuit as well. Load 500mA until the rails are stable and then switchover to the power amplifier. 

Now what about that voltage you are trying to get?

There are two very important aspects of your design you need to look at a) is this split rail or single rail you require b) what is the voltage you require and c) your SMPSU has an over voltage protection circuit – is this on the 5V rail or 12V rail, in most cases it is on both. You will need a schematic – modern power supplies are cramped and can take you ages to trace. 

Regarding (a), most power amplifiers are directly coupled and need a split rail supply so you will need to start looking at rebuilding the secondary windings. See below.  (b) Regarding your voltages, remember that the amplifier you are delivering power into most probably did NOT have a regulated power supply which means under heavy load conditions the voltage rails drop reducing power dissipation. Regulated supplies usually have excellent results but are also brilliant at destroying output devices if they cannot handle the voltages AND current across them. (Bipolar transistors are a case in point). Consult the specification sheet and check the load line graphs otherwise you may end up with an expensive mess. Regarding the final point (c), your power supply needs a reference voltage – it will hunt for that voltage and lock on. In most power supplies this reference is from both the 5V and 12V rails so you will need to trace and remove these feedback lines. Get the pdf data sheet on the pulse with modulator from the manufacturer. Texas Instruments and International Rectifier readily submit their sheets on the web, it’s easy to trace back  from the control circuit case you don’t have the schematic.  Go to this web site for some details – thanks guys.  The 200W power supply is a great learning tool, uses the ever popular TL494 (spec sheet readily available as you can see). I’m not going to bore you with the details but experimenting with these supplies you must remember that…

The output rails are low voltage. You will need to change the output rectifiers to fast switching devices with the proper PIV rating.
Your filter capacitors need to be changed as well.
You should modify the over current circuitry to suit your needs.
Possibly use the +5V standby circuit to power up a 110/220V fan.
Use the soft start circuit to power up the main switching circuit (the 200W power supply gives a very good indication of what one would normally find) and not hard wire the main switcher because in most cases this circuit also forms part of the over voltage circuit. i.e. look at the base of Q1.
The secondary transformers in PCs are normally centre tapped where the centre tap is grounded (and earthed). You will need to isolate this part and full wave rectify the output (remember that these diodes must be fast switching).  I mentioned split rail power supplies earlier on – this is where you will need to start looking at rewinding the secondary part of the main switcher. I believe the more professional experimenters normally opt for two power supplies in series – obvioulsy you will then need to start looking at the polarities of the output. Myself, although I no longer have the project running any more – I rewound the secondary winding and with a little bit of asistance from an engineer whom specialises in UPS design we managed to get a running +/-40V 8A  from an older 400W power supply. There was a lot more to this though – I had issues with the main switchers blowing because of a saturation issue. I’ll try to find the plans, should have it still. I will still be redoing this project within the next month or two and give full plans. Power supplies are so much cheaper these days.  BTW – do add a choke / pi filter to the output.

I came across this very interesting forum which essentially covers what has been said here but please remember that you are working with mains voltages and very importantly remember to filter out noise.  You may not worry about it but your neighbours will.  To the budding experimenters out there always use fuses that are not going to explode and leave glass fragments all over the place, normally your arms and face. I have repaired hundreds of PC power supplies over the years and more often than not taking short cuts have their complications. You can end up losing an eye if you are not careful. Do use HRC or high rupture fuses – they are there also for your own safety. They are usually ceramic or sand filled.

Many of the circuits shown on the internet are designed by passionate people, often engineers and very experienced hobbyists – they expect you to respect electricity.