Yet another power supply thread

[Frenchie]

Hi all, first up, I'm a Chemical & Process Engineer who tinkers with Electronics sporadically as a hobby so please forgive any ignorance on my part.

I've somewhat outgrown my existing bench supply, (1.5-15V @ 800mA) and am looking to build a new one to supplement it. I know that bench supplies are fairly inexpensive to buy off the shelf, but where is the fun in that?.

My requirements are a variable supply with a maximum voltage output of about 15-30V @ 2-3A with current limiting. Having the output voltage go all the way down to 0V would be nice, but not essential (and I know a few ways of doing that), a minimum output of 1.5-3V will be fine. The voltage side of things is fairly easy to do and I'm comfortable designing a linear regulator to do it. Adding in the current limiter is the bit I'm struggling to find a good solution for.

The simplest solution I've looked at is using a transistor as an emitter follower and force a voltage drop (either with diodes or an LED) over a sense resistor. At least from simulations that seems to be the cheap and easiest way to do it but is a fairly rough solution to the problem. I've seen a few variations on this theme that are better designs (usually using multiple signal transistors in addition to a Power transistor) and I would probably use one of those if I were to go that route

I've also considered using an LM317 as a constant current source followed by another to regulate the output voltage which seems to come up quite often as a common design. Alternatively, using a regulator with a current sense function (such as the L200) seems to be another way to tackle the problem from this angle.

Lastly, the final option I've come across is to get away from the constraints of dedicated regulators entirely and do the same thing with Op Amps. I was thinking of taking something like http://www.electronics-lab.com/projects/power/001/index.html or http://electronic-circuit.net/bench-power-supply/ and modifying it to better suit my needs. This does seem to be a bit over the top for a hobbyist supply but hey, it'll be a fun build.

Basically, I'm not sure which way to jump between those options. I'm sure people here have a much better idea of the pros/cons of each one than I do. Is there anything else I should be considering which would give me better results or am I going about this problem entirely the wrong way? Thanks!

[cboy2us]

It's been my experience that building a moderately large bench power supply, it is hard to foresee your future demand.  I would recommend an op amp based design with separate sensing inputs so that you can accommodate voltage drop in your power bus.

If you are seriously planning on something that will have a long useful life, you might consider at least a switching input stage followed by linear regulation so your linear isn't dissipating a lot of wasted energy.

[cyberfish]

I'm thinking about building one, too.

Found a power op amp that has easily adjustable current limit.

http://focus.ti.com/docs/prod/folders/print/opa548.html

With a microcontroller, cheap DAC (to set both input voltage and current limit), and huge heatsink, that should be all you need (and maybe a fixed switching regulator before it if your source voltage is way too high). I'm planning to use an old laptop adapter that can do 18V/4A.

[Balaur]

The OPA548 seems to be a very nice, integrated solution. Where's the fun in that? 

Linear voltage-regulated supplies are quite easy to build. It's the current regulation that could be a bit more difficult to implement.

If you want to go with a discrete solution, then one of the usual implementations looks like that:
http://www.mobileword.ru/index.php/download/doc_download/25----hy1803d.html

In addition, you can also use the LM723 chip quite nicely:
http://www.ve2ums.ca/chasse/Serge/Atelier/Projets/Membres/VE2EMM/alimentation_ang.htm

At work, I had also designed some power supplies based on TI POLs: big current in a very small packages, nice efficiency. My implementation featured: output voltage control using digital potentiometers; load-side voltage sensing; current sensing through Hall sensors and current regulation through a (sluggish but OK) microcontroller loop if needed. By using a comparator, it was much easier (and instantaneous) to power off the output when the current was bigger than a pre-set limit.

In any case, you must know that I chickened out when thinking about building a PSU for my home lab, and chosen to buy a commercial one instead.

[david77]

I'd also go the discrete route.
When I was starting to look into building a new PSU for myself some time ago I found that the usual
linear regulators had too many limitations that I didn't want in my main bench supply.
The common ones like LM317 or L200 take too little amps, have a too low max. input voltage, don't go
down to 0V, etc. To overcome all those problems would take nearly the same effort as building something
from scratch.
I found a very nice design in a 30 years old ELEKTOR magazine, it's a 723 as reference voltage source, two
741's - one for voltage, one for current limiting - and a couple of 2N3055's in the power stage. All basic
parts most hobbyist have lying around anyway.
It is very customizable, power and voltage are only limited by the output devices.

The only downside I can see is the enormous heat the power trannies have to sink at high load. At times I
could keep a mug of tea warm on the thing. That's because it nearly impossible to find a transformer
with more than one tap. But with enormous heatsinks that's not a real problem, I think.

I'd definitely go for more power than you may think you need now. You never know what you'll get into
in the future. My last PSU was a 30V/2A affair, I thought that was plenty. But it turned out there were
occasions when more power would've been welcome . And, of course one supply is never enough. For
versatilities sake you need at least two. So this time around I went for dual 30V/5A. (That mainly had something to do with the transformer I got for free, though.)
Oh, and sense inputs are not a bad thing.

Over all the home built PSU will never be cheaper than buying one off the shelf, unless you have the
parts or can get them cheap/second hand/surplus.

EDIT: Btw. the beefy PSU has come in handy one cold winter morning when my car battery was dead.
20min. at 12A pushed enough juice in it to cranck the engine .

[Frenchie]

Thanks for the input so far. I was already leaning towards building it from discrete parts and the responses seem to think that's a good way to go. I'm going to go spend some more time looking at coming from that direction. Always happy to hear further discussion though .

[saturation]

If your purpose is to have fun and learn, by all means.

But if you need a PSU, as you know its cheaper to buy a Mastech or similar clone then upgrade it as you see fit.

If you itemize the parts count, you'll spend 2-3x more dollars and still not get it looking as professional.

Enjoy!

[quantumfall]

I'm going to build this one. http://www.magenta2000.co.uk/acatalog/Legacy_Parts.html  There is a PDF of the original magazine build project at the bottom of the page.

Its linear with 0-25V  at 2.5 Amp and current limiting, uses standard parts and they have the PCB for sale. I did build this before and it worked well.  At low voltage and high current mean it will need a good heatsink is the only drawback.  But its relatively simple to understand and work with.

[tinsmith]

If you want to go discrete, this may help.

http://tangentsoft.net/elec/opamp-linreg.html

Also, this (may need modification).

http://tangentsoft.net/elec/tread/

One thing - you should probably fish up a linear AC/DC converter rather than playing with raw AC.

[FreeThinker]

Just my 2cents but in all honesty building your own psu is no-no. It will mostly cost you more than a decent unit in parts cost and will most likely not perform as well. My route would be to buy a non worker from ebay.If you fix it you learn from it and get a working unit .if you don't you can use it for parts to make a good one (same result) either way you win, just get it at the right price .

[House91320]

Use a liner regulator and a junk computer power supply.