LM338 Power Supply

After many years of employing this ugly and clumsy bench power supply, I decided it was time to build something better, smaller, and well - prettier. This is the result. (I love the DVM).  It began as a variable power supply based on an LM338 5A voltage regulator with input from the shack’s 13.8 volt 20A power supply.

Click on any of the pictures to the right for full size images. More photos are here., and please scroll down for the whole story.

Several years ago Phil Salas, AD5X, published an article describing a circuit which added variable output to a fixed output power supply. The article is here on his web site.  Recognizing the value of the concept, I mentally filed it away for future use. This became the basic idea for my meager attempt at a small footprint variable power supply for the work bench.

The circuit for mine incorporates the LM338 adjustable regulator. The Data Sheet has several very helpful application notes and circuits. I chose one that illustrated variable output and included protection diodes. Here’s the resulting schematic of my version.

A few words about it are in order.
First notice the two diodes, they’re included to protect the regulator from damage in case the input is accidentally shorted to ground. This is a distinct possibility if using jumpers to attach it to the shack's power supply. Also the output of station supply may be shorted if some other device fails. Without the diodes, if this happens, the capacitors will dump their charge back through the regulator. Since the current spike may be many amps, the regulator may fail. The diodes steer the current around the regulator and into ground, thereby protecting it from damage.

A major consideration is the “tuning rate” and linearity of the output voltage. Much time was spent in choosing the adjustment pot. This was not a trivial task. With a linear pot (implied by the data sheet I thought) the output voltage changed dramatically in the first 30° of rotation, and almost not at all in the last half of rotation. So I conjectured that a log taper pot of some kind was needed. It had to change the voltage less during the first 30° and more during the last 180° This implied an audio taper or similar, although that wasn’t immediately apparent to me. I tried linear tapers, audio tapers, played with custom tapers, and wasn't satisfied with any of them. The audio taper was the best of the bunch though. I would also add that a larger diameter pot "tunes" smoother than a smaller diameter one.

Speaking of custom tapers, the Secret Life of Pots web page is a treasure trove of information on potentiometers. The author states that there is no way to create an audio taper variable resistor using a linear pot and a parallel resistor. Guess what, he's right! Don’t ask how much time I spent trying to confirm/refute that statement. A reverse log taper is easily accomplished, if you ever need one, but not an audio taper. Many thanks to R.G. Keen for creating this page way back in 1999.

After much experimentation, I settled on an audio taper pot. It doesn't perfectly linearize the output, but it’s close enough, and is easy to set the desired voltage. It’s a far cry from the touchy, poor action of a linear pot. An audio taper is an absolute necessity with this regulator, IMHO.

Unfotunately while testing with various loads, the LM338 with the station ps didn't live up to expectations. Even with a small load of about 1 amp, the voltage sagged and wouldn't reach 12 volts. It obviously didnt have enough headroom to regulate well. I had posted the project on the Four State QRP Group's reflector and several members offered their experiences. Tommy Henderson, WD5AGO, has extensive experience with the 338 and suggested feeding it much higher voltage to ensure plenty of headroom. He's an instructor at Tulsa Community College and has his students build these with plenty of input.

I wanted several amps capability at 12 volts so I scrounged around to find one with higher DC voltage output. John Lonigro, AAØVE, had suggested that a laptop power supply might be adequate, and I found good one for a Toshiba. It's a switcher of course, but seems to be RF quiet, and it's rated at 19 volts at 4.7 amps(!) Using that rather than the station ps as the input, the 388 provides 4.4A at 17.6 volts - success! The regulator is dissipating 5.7W at that current level, ((19-17.6)x4.4)=5.7, so heat sinking becomes important. I mounted mine on the rear panel of the aluminum enclosure using a TO-220 mounting kit from Diz at kitsandparts.com and plenty of heat sink compound, and it's adequate. Did you ever notice how heak sink compund goes everywhere? I even found some on the steering wheel of my car.

Bottom Line: With adequate input the LM388 makes a FB variable power supply. This small supply is user friendly and fits nicely on my cluttered bench. For using the station 13.8V power supply as the input, try the same regulator that Phil Salas references in his article. While the LD1085V and the LM338 have nearly identical input/output voltages (1.2 and 1.5), the LM388 doesnt live up to that at moderate current requirements with my 13.5 volt Station power supply driving it.

If you build this circuit, please let me know how it performs for you.
Return to Accessories Page  Return to HOME page
Updated 30 November 2012 ... WAØITP

Click for larger image. JPEG Not Uploaded Yet JPEG Not Uploaded Yet JPEG Not Uploaded Yet