PSU for stdents as excercise.

[firewalker]

In the vocational training center I teach, the syllabus predicts that the students must build a linear PSU. The guide line is a simple LM317 (actually two of them in parallel) one. No CC mode.

I want to discard the guideline. I was thinking something like the one bellow. It combines transistor, opamps and digital design. I can combine several lessons.

Any comments?



Alexander.

[Ian.M]

That's an extremely over-optimistic design for tech student construction unless your center cherry-picks the top 10% of students + assesses them with an aptitude test before enrollment.  I've sat through such a class (I needed other modules and it was a prerequisite I couldn't test out of) and had a lively (and somewhat acrimonious) debate with the college principal (ex Navy commander) about the critical flaws in the simple regulated PSU design we were supposed to build.  It had a metal lever toggle switch as a mains switch but was in a plastic enclosure.  It also couldn't deliver anywhere near the rated voltage/current because the transformer was underspecced and there wasn't enough headroom for regulation.  I proposed grounding the switch body, increasing the reservoir cap (and I did a graphical analysis to approximate the required value) and derating the PSU.  My objections were apparently not well received and most of the class were displaying extremely glazed expressions or were otherwise distracted.  However the next session, it was "Thank you Ian" from the principal and there was a pile of supplementary sheets for the lab handout detailing the chainges and two parts bins with the bigger caps and solder-tag washers to fit the switch sitting next to them on the instructors bench.

If my class had been given your schematic to build, I estimate you'd have had about two of us succeed, which was about 10% of the class.  That would have been me and the only other guy with hobby experience building circuits.   Maybe something LM723 based could be a happy compromise?

[firewalker]

They will have the PCB made by me. The only "hard" part may be the MAX4080 (although my pcb accepts both smd and a max4080 on dip-8 adapter).

I want the to build something physical with opamps with a feedback loop. And when they are going to learn about micros, use an Arduino (or something) with a display to control it.

Maybe I could make it as an extra option for the students that needs something more.

Any flaws on the design?

Alexander.

[Gyro]

When I read your original post, I interpreted "build" as design and construct a PSU. I'm just thinking that if you present them with a more complicated schematic and a Pre-designed PCB, there's only the 'construct' bit left. Does that still meet the requirements of the syllabus?

[firewalker]

Yes it does. But I don;t want to present it at the beginning as ready to solder. Nor the design, nor the pcb. I want to design it with them. To lead them to the design, exposing the problems of simpler designs, and utilizing basic opamp actions to overcome those problems.

* The LM317 psu is available as a kit they can buy...

Alexander.

[chris_leyson]

Just had a quick look at the schematic, nice design classic op-amp stuff with a pair of current steering diodes to switch between CV and CC modes. Control loop looks OK can't see it generating enough zeros to go unstable. 

[firewalker]

The design is heavily based on a psu from microsyl.com.

Alexander.

[chris_leyson]

I guess as a teaching aid you could deliberately add another capacitor somewhere to generate an open loop zero... wow now you've got overshoot when you close the loop. But seriously, you could put one or two more op-amps in there and a few more caps to make it unstable and in that respect it might be a better teaching aid. But maybe not, lazy or stupid design engineers wouldn't bother to learn or understand control loop theory but they might get a job somewhere "cos they an engineer init". Good teaching pays off in the long run Alexander 

[firewalker]

I could add a comparator for better CC detection.

Alexander.

[Kleinstein]

For a design made for learning I would reduce it to two output transistors - the main principle is learned with two as well. The idea to build a supply from ground up might be OK, but than the design should be good. At least do quite some tests in simulation and than build one or two, before starting the course. The design is not as simple as it looks at first glance.

The control loop does not look that good. The voltage control should be stable, but with the current values it could be rather slow. With extreme capacitive loads it may show excessive ringing.

The current control does not look good at all - might very well start to oscillate. This is the more difficult part with the emitter follower output stage anyway - to make this a little easier slightly larger emitter resistors could help.

Another point is, that the cross over from CC to CV mode might be really slow - also a general problem with this type of design and more so with a slow current control setting.

The max4080 is rather slow and thus would require a slow responding current limit. So you might need a faster solution there. With the extra RC filter like in the plan it is likely unstable in most cases.

With a 30 V nominal input you could run in to limits of the supply for the OPs - to be save I would stay a little lower (e.g. 24 V supply and thus 20 V max).