Power supply designing

[Erik]

Hello all,

A couple of years ago i got a big transformer from a friend (2x18V@30A) and wanted to design a power supply for it but i failed because i didn't/don't have the knowledge for it. Recently i picked up electronics
again and started from scratch learning theory. I did some self study to get atleast some basic skills but because i'm too eager to design a power supply myself i started looking at some schematics figuring out how they work and applying theory to that instead of reading 1000 pages of theory.

The main problem i'm facing is the diversity of designs available on the net.
I also looked at some forums about designing a power supply but that didn't helped me out much
because everyone has another opinion about what's good and what's not. The first one says "use an LM317" the other one says "no!, use a LM723" The next one says "no! use a L200" and the last one says "Strap some op-amps together". Perhaps you can understand that this doesn't help out a newbie like me to determine whats the best base to start with.

So my question is if someone has a good base design to start off with for a (lab) power supply which can be regulated from +/- 0V upwards and 0A upwards. I'm not asking for a complete design but just a simple but good base circuit for learning purposes which can be expanded eventually.

Greets,

Erik

[septer012]

The reason everyone has different opinions on the regulators is that they all waste different amounts of power to heat.  This is a trade off with current output and performance of these linear type regulators.  The LM317 for instance requires something like Vin=3v+Vout to operate.  In a very simple model lets say you input 9V as Vin to the regulator and you have it set up to output 3V.  Whatever circuit you are running with the power supply is consuming .200 Amps of current.  Power Wasted due to heat is: (Vin-Vout)Current = (9-3)*0.200 = 1.2 Watts.  This is wasted power.  Wasted power is converted almost entirely to heat energy and dumped through the Vout pin/heatsink on the package.  Not too worrysome if you have a good heatsink in place, but it can be far worse and you can even desolder solder joints. The regulators datasheet should have a thermal joint plot which shows you watts vs temperature.

An LM317 should probably not be used in a bench supply as it was really designed to be setup/designed per application and not such that its a variable supply with many varied conditions if you know what I am saying.

I suggest you read this: http://www.dimensionengineering.com/switchingregulators.htm

And maybe they explain things a little better than me.  They are promoting switching regulators over linear regulators.  There are two things that you have to worry about if switching to one of these type of regulators.  1) Noise in to your circuit on switching frequency 2) cost, they are very expensive compared to their linear counterparts 35 cents verse 15 dollars.

[jimmc]

Just to add to what septer012 said, if you want to make full use of your transformer using a linear regulator then, under worst case conditions (max current, 0v out), you will have to dissipate about 1000w in your power supply (2 x 18V x 30A). Not impossible but not trivial either.

Try looking in some of the amateur radio books, such as the ARRL handbook, hefty low voltage power supplies are popular for powering their portable gear.

Jim

[Erik]

Gonna take a look at that site for sure septer012!

In the mean time i couldn't wait untill someone replied so i've built build this power supply http://www.electronics-lab.com/projects/power/003/index.html  for practice. I did some quick testing and at the loos of it it works like a charm. Although there are some things i don't like for example they didn't use a bridge rectifier and didn't incorporate a fuse within, another thing i don't like is the transistor used to drive the 2n3055(2n2219) even though i'm still in the learning process it looks to me there should be a better alternative. It does have a nice led that lights up when your in current limit "mode" and seems pretty stable too.

What do you think about the schematic? Like how it's designed and what could use improvements. Well, in the mean time i'll do some calculations to figure out how it's exactly working.


Edit @ jimmc

That would be the worst case indeed and would need some very good cooling. But in my case i think it would be better to make a dual output psu to limit the dissipation a bit but it sure is gonna need a big fat heat sink. 

[desolatordan]

Gonna take a look at that site for sure septer012!

In the mean time i couldn't wait untill someone replied so i've built build this power supply http://www.electronics-lab.com/projects/power/003/index.html  for practice. I did some quick testing and at the loos of it it works like a charm. Although there are some things i don't like for example they didn't use a bridge rectifier and didn't incorporate a fuse within, another thing i don't like is the transistor used to drive the 2n3055(2n2219) even though i'm still in the learning process it looks to me there should be a better alternative. It does have a nice led that lights up when your in current limit "mode" and seems pretty stable too.

What do you think about the schematic? Like how it's designed and what could use improvements. Well, in the mean time i'll do some calculations to figure out how it's exactly working.


Edit @ jimmc

That would be the worst case indeed and would need some very good cooling. But in my case i think it would be better to make a dual output psu to limit the dissipation a bit but it sure is gonna need a big fat heat sink. 

Here's a topic about how bad the design for that linear power supply is, and the necessary changes needed to make it safe and usable.

http://www.electronics-lab.com/forum/index.php?topic=19066.0

[Zero999]

I wish the webmaster would fix or remove all the broken projects on that site. It gets annoying having to tell eveyone who's made a bad project and makes a thread about it, how to fix it. It's even worse when the project is dangerous, i.e. the PCB exposure unit which uses dangerous carcinogenic UVC tubes which are the wrong wavelength for exposing PCBs rather than safe UVA tubes which work perfectly.

[Erik]

Hmm, i wasn't aware this was such a "bad" design as some claim on that forum, just came from my work so i'm gonna take a look at what kind of improvements they made to it. I got interested in it because it was such a simple design with at the looks of it good specs but now i'm starting to doubt.

[Simon]

what are your current supply requirements ? work that out and make something to satisfy it then move on. your looking at a very powerful unit which you will probably not need for some time, just make a circuit capable of doing what you currently need. there are plaent of 1-5 amp SMPS chips that are not too costly.

[rossmoffett]

When all else fails, buy a book, you're in luck, only 300 pages!  

http://www.amazon.com/gp/product/0750674458/ref=oss_product

If you read the first 150 pages I think you'd be doing fine, the rest is the finer points of component selection.  For 18V @ 30A you're going to need some very beefy heat sinks or a switched-mode power supply, the most complicated type.  I'm sure there are lots of stock ICs for this, but not necessarily for an adjustable regulated supply.  If not that, you could build a simple beefy rectifier with a capacitive filter and throttle it with a variac.

Google switched-mode power supply for a good kick.  Here's a great URL:
http://www.smps.us/smpsdesign.html

You might consider rectifying, then charging a beefy capacitor bank, pulse-width modulating the output from that into another beefy capacitor bank, and then running that output through an inductor for good measure.  At the end you'll want a simple panel meter (cheap <$10 multimeters work great for a .1V accuracy - or a digital panel meter) to give you a rough idea of what you've got.  This kind of designs voltage output would be dependent on the load, however.. it would be pretty simple.

[Zero999]

Just to add to what septer012 said, if you want to make full use of your transformer using a linear regulator then, under worst case conditions (max current, 0v out), you will have to dissipate about 1000w in your power supply (2 x 18V x 30A). Not impossible but not trivial either.

One solution is to use a tap changer. Connect the two secondariness in series and when you set the output to a lower voltage you draw power from the centre tap rather than the higher voltage rail but then you've still got to dissipate over 500W which is still a problem.

Then there are all sorts of other tricks such as connecting resistors in series with the pass transistor so power can be dissipated using cheap resistors rather than semiconductors. Another possible solution is to connect large choke in series with the AC side which can bypassed when at higher voltage settings: this has the advantage of using very little power. You might also be able to do a similar technique using capacitors but ripple current is an issue.

Then there's the switching regulator, the output is noisy so might it might not be a good idea for a lab power supply but it can be used in conjunction with a linear regulator to form what is known as a hybrid power supply. The idea is to connect a linear regulator to the output of the SMPS and set output voltage of the linear regulator to a volt or so lower than the SMPS. This combines the advantages of the linear and SMPS: you get a power supply with a reasonable efficiency with a clean and smooth output. A tap changer would still be a good idea for an SMPS because they're more efficient and easier to build when Vin is near Vout.

[qno]

The problem with your question is the controls from 0 to max voltage and 0 to max current.

Look for the HP power supply schematics and the HP power supply reference book. this will get you started.
They use a separate power supply to power the control circuit to control the main output.

If you want to supply 10 or 20 amps at lower voltages (0 to 12) volts you need some sort of preregulation to
prevent the 500 or 1000 Watt made by the transformer to be converted to heat.

The easiest and sometimes cheapest way is perhaps to buy a used one on e-pay or your local HAM swap meeting.