Power Supply project

[cksa]

So I managed to get hold of a transformer for free. http://img153.imageshack.us/img153/5761/mg9048.jpg

It's has two secondary coils. 23.7V each and 2.43A

I want to build a lab power supply out of it as a project. Minimum goals I would set are:
*dual rails output, one negative one positive. up to +- 15V, so I can do basic opamp work. I guess I would want a minimum of one AMP. Adjustable through potentiometers. Also adjustable current limit.

Ultimate goals:
*Have it controlled by a microcontroller - how can I do this? =\
*Have a lot more current capacity.

Is it an idiotic idea to whack a switching buck regulator after I've rectified the voltage from the secondary's instead of a linear regulator?, so I can hopefully get more amps at low voltages?

EDIT: As an EE student, i'm not really cared about the practicalities of making one myself. I really want to make this a learning experience. I will probably design up a PCB for this at the end to get more experience at laying out components.

[alm]

It's has two secondary coils. 23.7V each and 2.43A

I want to build a lab power supply out of it as a project. Minimum goals I would set are:
*dual rails output, one negative one positive. up to +- 15V, so I can do basic opamp work. I guess I would want a minimum of one AMP. Adjustable through potentiometers. Also adjustable current limit.

Something like 2x25V 1.4A would match the transformer, at least for linear regulation. One useful option for op-amps is tracking, so you can adjust both supplies symmetrically (or with a fixed ratio).

*Have it controlled by a microcontroller - how can I do this? =\

Replace any pots you see by DAC's. Use rotary encoders or buttons to control the micro. You could also connect pots to an ADC, but then the controls won't reflect any changes you make some other way (assuming you want serial/USB/LAN control, which seems easy if you control everything from a micro). You can use ADC inputs and an LCD display instead of expensive, heavy panel meters. Don't be tempted to involve the micro into the control loop, see the other topic about a power supply kit in this forum .

*Have a lot more current capacity.

For linear supplies, get a larger transformer. Adding more pass transistors, more cooling and heavier resistors is easy (though not free). Another option would be a parallel button, which parallels both supplies (doubles the current), if you don't need high current and symmetrical supply at the same time.

Is it an idiotic idea to whack a switching buck regulator after I've rectified the voltage from the secondary's instead of a linear regulator?, so I can hopefully get more amps at low voltages?

Nothing wrong with that idea. One issue with switching regulators is that it produces a fair amount of broadband noise, which exceeds the 1mVrms or so expected from lab supplies. A tactic that is sometimes used is a switching pre-regulator which produces a voltage just high enough for the switching regulator. That way the linear regulator doesn't have to dissipate 90% of the energy at low voltage levels, but it does take care of some of the noise. Elektor and ELV both published designs for adjustable supplies with buck regulators (without linear regulation). The ELV one is available for free (in German, but you can look at the schematic), the Elektor one costs a few bucks. Not suggesting that you have to clone these designs, but they may give some ideas.

EDIT: As an EE student, i'm not really cared about the practicalities of making one myself. I really want to make this a learning experience. I will probably design up a PCB for this at the end to get more experience at laying out components.

Nothing wrong with building for the sake of learning and building, that's 90% of building electronics yourself anyway, no way you can compete with that $5 gadget from China.

[david77]

I say go for it and build your own PSU!
I've built quite a few over the years and everytime you learn something new. It's not just electronics, there's so much more involved in building your own stuff. All the mechanical work to make a nice case, fitting the boards inside, choosing knobs and buttons and what have you is also good experience.

You could go the simple (and frankly boring) way and just use some LM317T for regulation. That would give you a 1,2-22V / 1,5A dual power supply. It would have reasonably good performance at lowest cost and lowest part count. You could add a little circuitry to make it go down to 0V, wich is a nice touch  .

I'm using a circuit wich was published in Elektor in 1982 (!), it's rather discrete with LM741's, LM723 and a bunch of 2N3055. My unit gives 2x 0-35V/5A. The only drawback is that the secondaries are not switched to reduce power dissipation, so I can use it as a heater, too. But other than that it works brilliantly and without flaw.

At the moment I'm about to replace the crappy analogue meters with new and shiny LED Displays wich hopefully will make it even more usefull.

Personally I wouldn't want a SMPS as my main bench supply. Two reasons really: Higher noise floor due to switching frequency and maybe I'm a bit oldfashioned about it .
The SMPS solution is sureley not bad, it's certainly much more efficient than linear. It always depends on what you want to do with it, I suppose.

David

[cksa]

hmm. i found the other power supply threads on this forum and had a read.

I'll probably stick with linear then....going to attempt to read up on how you would interface a DAC from a micro....into adjusting voltage and current on a regulator..

[alm]

There are also more stability issues with switchers, some commercial switching lab supplies get screwed up when RF is conducted into the supply. It's harder to confuse a linear supply.

One issue with the LM723 is that its constant current mode is not great, it's more like a current limit than a real constant current mode. But it's possible that the Elektor design worked around this.

[david77]

Yup, they did. The 723 is only used to stabilise the reference voltage. The voltage and current regulation is done by two 741's.