I am designing a power supply for a computer, the 12V rail needs to be capable of supplying 12v@40A continuous power. I will be using an iron core transformer to step down the current, and a full wave rectifier consisting of 4 diodes and a capacitor to get 12vDC. I can't count on the input for the power supply to be completely consistent, so I will need a voltage regulator on the 12v power rail, or an AC voltage regulator on the mains. The regulator should be capable of compensating for small voltage drops as well if possible.

If you have any suggestions on how to build a regulator to those specifications, I will greatly appreciate any help.

PS. change the title to: "12v 40A voltage regulated power supply", I got a black screen every time I used a long title, so I shortened it.

 

I just can't persuade myself to design from scratch, something you can buy retail for $30.

http://www.newegg.com/Product/Product.aspx?Item=N82E16817822013

 

mmmm let me think how about an Atx off the shelf.

http://www.ebay.co.uk/itm/110V-220V-AC-to-DC-Converter-power-supply-12V-40A-/390078610583

 

I know I can BUY a PSU, the problem is the quality control and the amount of power the store-bought power supplies can deliver is lacking. I already killed one 500W power supply by drawing no more than 160W, and faulty PSUs have cost me a few motherboards. I would rather design and build a power supply then spend hundreds of dollars in parts later.

 

sadly you will spend hundreds of dollars designing one..

 

sadly you will spend hundreds of dollars designing one..

The design has yet to cost more than 30$, unless building a voltage regulator has a crazy high cost to build, then the design will not be expensive to build. Voltage regulation is the final part of the design. It is not difficult to get/build a transformer, it is not expensive to get the diodes, low voltage capacitors are easy to get, and can be fabricated if necessary. The voltage regulator design will be used for other projects. I just need some help locating, or designing a voltage regulator that will be capable of supplying enough power. I have no experience with voltage regulators, so I need help with that part of the design.

Would 4 12v 10A voltage regulators in parallel do the trick?, if so what brand of regulators would be acceptable?

 

You want a linear regulator. Even a low-droput one will need arpoximately 1.5-2V higher input voltage than output to maintain regulation. 2V*40A is an appreciable amount of power to be dissipated, but it likely will be more beacuse you don´t want to stay on the verge of dropping out of regulation.
My guess is at least $50 for the transformer.

Another thing is what happens after you build it, and the regulator becomes unstable when you finally connect the pc, are you prepared to loose the motherboard plus some other components again? Maybe just buying some high quality psu may be the better option.

No, you cant simply parallel regulator together.

 

You want a linear regulator. Even a low-droput one will need arpoximately 1.5-2V higher input voltage than output to maintain regulation. 2V*40A is an appreciable amount of power to be dissipated, but it likely will be more beacuse you don´t want to stay on the verge of dropping out of regulation.
My guess is at least $50 for the transformer.

Another thing is what happens after you build it, and the regulator becomes unstable when you finally connect the pc, are you prepared to loose the motherboard plus some other components again? Maybe just buying some high quality psu may be the better option.

No, you cant simply parallel regulator together.

I see..., would a line conditioner across the mains work better?
Also, could a mag amp be used to regulate the power supply without the resistive losses that appear to be tied to silicon regulators? If so how.

I am planning to wind the transformer, or use one that came out of an uninterpretable power supply that has a messed up battery. either way it won't be too expensive.

 

sadly you will spend hundreds of dollars designing one..

Actually, it would be more like $50k. And you will need to hire an experienced PS designer as a contractor, somebody to design the PCBs, somebody to design the mechanical assemblies. Actually, $50k probably won't cover it.

 

The design has yet to cost more than 30$, unless building a voltage regulator has a crazy high cost to build, then the design will not be expensive to build.

This is what happens when digital people stray into the mine field of power electronics.

Take some very knowledgable advice: BUY ONE. You can not possibly build one for anywhere near what they cost to buy, and you can get quality units. Just because the made in China junk is not reliable, don't assume all power supplies will fail.

 

If you mean a production ready, then yes that would be a reasonable price, but for a home made one-off the costs will be lower. But still I don´t think that they will be lower than buying some higher-end ATX supply off the shelf, off course with reasonably 1.5-2x higher wattage than what the PC actually needs.

Mind you guys that OP wants a linear supply, not a switch mode.

 

A 600VA transformer, 50A diodes for the bridge and a very sizeable smoothing capacitor (let's say for the sake of argument 2000uF/A giving 80,000uF but will probably need more) and regulators for the +12V, +5.0V and +3.3V rails that will have significant amounts of heat to shift are why linear power supplies are never used for computer power supplies. Factor in the necessary safety circuits like over voltage, current and temperature protection and your design will end up costing a significant amount of money and be the size of shoe box if not more.

 

Mind you guys that OP wants a linear supply, not a switch mode.

The OP never stated he wanted a linear, one of the respondents jumped to that conclusion. A 500W linear is highly impractical and not any more reliable than a WELL BUILT switcher which would be about 1/20 the size and weight of a linear.

If this is for development work, just buy a good 1kW lab supply that will come with built in protection and is virtually indestructable. Or, keep trying to kluge something cheap together and spend a lifetime fighting with it.

 

In his first post OP says he wants iron core transformer with full bridge rectifier and output cap. You don´t use iron core transformers for switchmode supplies.

 

In his first post OP says he wants iron core transformer with full bridge rectifier and output cap. You don´t use iron core transformers for switchmode supplies.

Yes, you do if you use a switcher to buck down the unregulated DC supply coming from the FWB/cap. Not all switchers are offline, but 500W power supplies certainly should be offline switchers unless enough money in the budget is allowed for a wheelbarrow to haul this boat anchor around.

A 500W linear is so impractical in this case it is virtually impossible, the power dissipation would be ridiculous..

 

I do not plan to use a printed circuit board unless I think I might be able to sell the PSUs in high volume, and have the cash to back it up. That isn't going to happen in the foreseeable future. I agree the cheap crap made in china isn't going to hold up, but it doesn't change the fact that most PSU suppliers appear to have a quality control problem, and also like to lie about the maximum continuous power their products are able to supply. I agree it might be cheaper to buy one, but if I can figure out how to prevent the output voltage from fluctuating, then I can build a cheap, reliable power supply. that I know how to fix. I don't care how the voltage is stabilized, I don't claim to know the best method to stabilize the output of a power supply. I can see that a regular voltage regulator will dissipate the extra power as heat, and thus are not a good solution. Pulse width modulation, and mag amps appear to be better solutions, but they require control circuitry.

Is there a device/circuit that senses the difference between a reference voltage and the voltage of some output, and increases the resistance across a third set of legs if the output voltage is greater than the reference and decreases it's resistance when the reference voltage is greater? The device/circuit would need to be able to handle about 100mA across the third set of legs max.


Also, anything I said in my previous posts is subject to change if it is too impractical to implement, a 500W iron-core transformer isn't THAT big, the rectifier can be replaced with something better, or moved further down the line if need be.

 

I know I can BUY a PSU, the problem is the quality control and the amount of power the store-bought power supplies can deliver is lacking. I already killed one 500W power supply by drawing no more than 160W, and faulty PSUs have cost me a few motherboards. I would rather design and build a power supply then spend hundreds of dollars in parts later.

Buying the cheapest power supply is a sure way to end in failure.
There are plenty of high quality computer supplies that will have 12V rails that meet your current demands.
I don't even look at power supplies that run in the 500W range that cost less than $150.00.
Garbage in > garbage out, get something good.
http://www.newegg.com/Product/Product.aspx?Item=N82E16817371052

 

Also, a switch mode power supply would also work, i hadn't considered that option before. Those appear to require fairly complex control circuitry, and have some notable issues.

 

I suggest that you take Bountyhunter and Sue_AF6LJ's advice, and go with a well-designed high-end COTS (commercial, off-the-shelf) power supply. If you have only recently learned of SMPSs (switch mode power supplies), then you are quite a ways from actually building a working supply of ANY power out, much less to meet the specifications required of an ATXplus12 supply.

Since a computer will usually spend most of the time being switched on, you really need it to be as energy-efficient as possible. This excludes linear power supplies, because they will waste a great deal of power in the pass transistors as heat.

Along with a well-designed power supply that exceeds your power requirements by at least 20%, you should also use an uninterruptable power supply up front.

If you'd like to learn some basics about SMPS's, I suggest that you start with Ronald Dekkers' fine "Flyback Converters for Dummies" page:
http://www.dos4ever.com/flyback/flyback.html

If you decide to try either the boost or flyback circuits, I suggest that you replace R4 with a 47k resistor, and omit R7 and the neon lamp. The R4 change limits the output voltage to ~50v max, which is much safer; and a neon lamp usually needs 55v or more to light.

Note that flyback SMPS's are only good for up to around 150W outputs; there are a great many more topologies that exist. This is just for a relatively inexpensive introduction to SMPS's; there is a lot to learn about them.

 

If I cannot find an acceptable DIY solution, I will take Sue_AF6LJ's solution. Even if I do I still would like to know how to construct/acquire a magnetic amplifier control module (for voltage regulation, other types of control modules appear to be much easier to construct)

This is basically what the control module should be capable of:

A device/circuit that senses the difference between a reference voltage and the voltage of some output, and increases the resistance across a third set of legs if the output voltage is greater than the reference and decreases it's resistance when the reference voltage is greater. The device/circuit would need to be able to handle about 100mA across the third set of legs max.

The final product will be tested before I stick it into my machine, even if it doesn't work out, I will have learned something useful.

here is a link to the mag amp Wikipedia page: http://en.wikipedia.org/wiki/Magnetic_amplifier