(Republished from another web site by permission of the author - me)
A variable voltage power supply is an asset to any bench. Even if a commercial lab supply is available, another power supply is always useful to operate those things that require an odd voltage or to check performance under different voltages. When combined with the Power Monitor, this ten buck power supply will keep up with most lab-type power supplies.
The major component of a power supply is just that - the line-powered AC to low voltage DC power supply. We won't waste our time designing this part. The local Goodwill Store or Value Village should have any number of nicely designed, regulated switching power supplies. What we want is a supply that provides a couple more volts output than we need at an amp or more. The supply I found for this project is an HP printer supply rated at 18 volts, 1.1 amps. It set me back $1.99 at the local Value Village. A regulated switching power supply rather than an unregulated, transformer-based supply is desirable for this application. Sparkfun has a tutorial about power supplies that is useful.
The other major component to make this variable regulated supply is a voltage regulator. This is a case where the tried and true is still the best, the LM317. It's been around for over thirty years and is still going strong. This chip offers tight regulation over a 1.25 - 37 volt range with short circuit and overheating protection. It's made by a number of manufacturers and will cost between 42 and 74 cents at Mousers. While I have to acknowledge that Radio Shack is almost never the right answer, they even carry this (Catalog #: 276-1778), although at 3x - 5x Mouser's price ($2.29). National Semiconductor has a nice application note for this great device.
A capacitor, a fixed resistor and a pot (variable resistor) complete the parts list. Here's the schematic (and a link for the schematic & parts list).

The final item necessary to contain all this magic is an enclosure. Radio Shack (still not the right answer) has a bunch of different project boxes but they all share a certain cheap feeling producing a "home brew" result. In pursuit of a more substantial, nicer looking enclosure, I again turned to the folks at Goodwill for help. I found an computer keyboard/monitor switch in a metal enclosure. You may have one of these laying around for serial or printer connections. I had many of them until a couple years ago when I donated them to Goodwill, but I look at it this way. The $3 I paid for it was pretty cheap storage!
I removed the switch and connectors from the enclosure and drilled the front panel to accept two binding posts on 3/4" centers. A happy discovery was that rubbing alcohol took the light blue ink labeling of the original switch right off, leaving a nice looking front panel. The pot installs in the hole formerly occupied by the switch.
The rear panel of the enclosure isn't so pretty with all the holes. A happy coincidence was the the connectors for the keyboards were on 3/4" centers that fit the binding posts with just a little Dremel tool work. I also mounted a panel-mount coaxial DC power connector in one of the reaming holes. I can plug the switching power supply in directly without modifications. I didn't worry about the other holes in the back panel - they'll be out of sight when it's being used.
Construction is a piece of cake. The LM317 should be mounted to the enclosure using an insulated mounting kit. It will get warm to touch depending on the supply voltage and load, so the enclosure acts as a heat sink. The tab of the LM317 is connected to the output terminal, so it needs to be electrically isolated from the enclosure. The rest of the components can be connected using the binding posts, pot lugs and LM317 legs as "tie points."
When all the connections are made, connect the power supply (check polarity twice before making the connection!) and check the output voltage with a DVM. When the knob is fully CCW, the voltage should be 1.25 volts. With the knob fully clockwise, the voltage will be a volt or so less than your power supply output. Setting the knob between those extremes should produce any desired voltage which won't vary much with load.
I put a sheet of paper on the front panel under the knob. It had centerlines marked for the pot shaft and a circle slightly bigger than the knob. I used a voltmeter to determine the knob position at each voltage increment and marked it on the paper.
I carefully removed the paper and taped it to my computer screen where I had drawn out the panel and a number of lines starting from the center point of the pot at various angles. I adjusted the angle of each line in the drafting software to align with the marks on the paper taped to the screen. The lines in the drafting program could be seen through the paper taped to the screen. Note that the size on the screen wasn't exactly close to 1:1 - all that mattered was the angles, and they stay the same regardless of scaling.
When all the lines were in the correct positions, I printed the page on MacTac, cut the hole for the pot (the label doesn't go all the way to the binding posts, so I didn't have to cut any holes for them), and carefully applied the label to the enclosure being sure to keep the centerlines lined up with the pot.
I was actually amazed at the accuracy of this method. In the photo, the voltmeter reads 11.57 volts, but the slot in the knob is slightly counter(anti-)clockwise from the line for 12 volts. When it's centered on any of the marks, the voltage is accurate to better than 5%.