Lab power supply

26.02.2010 20:56

I've been wanting to design and build a new 50 W lab power supply for some time now. It has turned out to be one of those projects that you think will take a month tops and then the lack of time stretches it to half a year and counting.

The minimum requirements are 0 - 25 V and 2 A with an adjustable current limit. I've considered four approaches for such a design:

  • A plain linear regulator,
  • linear regulator combined with a transformer with multiple taps,
  • linear regulator with a thyristor pre-regulator and
  • switched-mode regulator.

These are pretty much sorted by ascending complexity and efficiency.

The power requirements are just barely within the reach of the first option. However that would require a big passive heat sink (I want to keep away as far away from unreliable fans as possible). Plus building a new device that would operate around 10 - 20% efficiency most of the time doesn't really feel right. So scratch that.

I spent quite a bit of time researching the second option. In fact, I have an almost completed design for it on the drawing board right now. It uses a two-tap transformer with a relay to switch between them - transformers with more taps aren't easy to find. The regulator part is roughly based on the 0-30 V power supply from Electronics lab.

Still, I'm not really happy with it. I have doubts about the longevity of the relay and worst-case heat dissipation is still uncomfortably high.

By the way, the original Electronics lab design is pretty broken in several ways and I strongly doubt that it meets its specifications - but that is perhaps a topic for another post.

I'm not going to even consider building a switcher for this purpose. It's noisy and has worse regulation characteristics than a linear design. Not really something I would want in a lab supply. Plus finding appropriate ferrite cores for switchers is always a pain.

So, right now I'm looking into a thyristor pre-regulator. There's a pretty good application note from Linear technology that has a basic circuit. It looks solid on paper and I'm going to give it a try tomorrow to see how it behaves in practice. If it works as advertised I'm more than prepared to go back to the drawing board with this.

Posted by Tomaž | Categories: Analog


Have 30V / 10A regulator at home made with scr preregulator. It's heavy (choke, transformer) and noisy (scr s tend to produce elmg noise), but it is very relieable.

Posted by Gašper Korinšek (via Facebook)

Interesting. I was under the impression that these preregulators don't cause much more noise than an ordinary rectifier.

Posted by Tomaž (via Facebook)

Ordinary rectifiers perform "switching" at zero crossing. Phase regulation on the other hand can switch anywhere during sine wave (in worst case this is at peak).

Posted by Gašper Korinšek (via Facebook)

Well, if we're talking about RC rectifiers the diodes will mostly switch on at the peak of the sine wave, since that's when the input voltage is higher than the capacitor voltage on the other side. If the regulator is done right, it will also only switch thyristors on when the voltage on them isn't significant.

Posted by Tomaž (via Facebook)

Actually this is LC rectifier (output waveform is identical to RC). Problem is noise injected back to AC line (output of power supply itself is good).

Posted by Gašper Korinšek (via Facebook)

Look in "Everyday with Practical Electronics", April, 1994 - "Mosfet Variable Bench Power Supply"-Mark Stuart.
The scheme is designed with very good SCR pre-regulator. comments and schematic i fond at

Posted by Vesko

Vesko, thanks for the tip.

This looks like an interesting design: I wouldn't call it a SCR regulator because it looks like it uses bipolar transistors as switching elements but it works on a similar principle.

For any one else interested, this is the exact discussion thread Vesko is talking about:

My lab power supply however is almost finished at this point. Certainly too late to make such radical changes to its design.

Posted by Tomaž

Add a new comment

(No HTML tags allowed. Separate paragraphs with a blank line.)