A Simple Power Supply for When you Don’t Need all the Bells and Whistles.
Having a bench supply? That’s great. Moving it around? Well, it’s called a bench supply for a reason. There’s definitely a niche for a small power supply you can carry from place to place.
I’m not pushing the boat out too far here. All I want is a power supply with the absolute basics: adjustable output up to 12V, about 1A of current, built-in voltmeter, plug pack source. That’s it. It might be a bit much to slip in your pocket (plug packs are huge) but you could easily throw it in a small bag.
While the circuit isn’t much to look at, I tried out some new and unusual techniques with the case. That’s my ulterior motive here, but the end product turned out to be way more useful than expected.
Circuit Design
There’s not a whole lot to say about the circuit. You only need five or six components (excluding the connectors and plug pack) none of which are particularly interesting.
This is more or less copied directly from the LM317 datasheet with a voltmeter and protection diode tacked on. I chose R1 based on my available potentiometers; R2 makes for about 12V when R1 is 10K. Filter capacitors are pretty much arbitrary. I used what I had to hand.
For a power source, I simply bought the cheapest 15V plug pack that could source at least 1A. This one. I am not copying that incredibly long alphanumeric identifier here. You have a link if you want it.
I chose the Adafruit voltmeter again. This time it’s better matched to the application. Having 0.01V resolution would be nice, but 0.1V is more than adequate.
Circuit Construction
Since there’s so little, I could easily get away with point-to-point wiring. My problem with that is it would be exceptionally hard to service. PCB construction is out because it would be excessively expensive for such a simple one-off project.
Instead I grabbed the Chip Quik SBB1002-1. This is the smallest “breadboard” style protoboard I can find. Having learned from Simple Load, I invested on off-board plug pack connectors. I only have to fit the LM317, one fixed resistor, two capacitors, and four connectors.
It’s hard to overstate just how small this board is- it’s just a tad bigger than the TO-247 heatsink!
Soldering was a simple job, but where’s the LM317?
On the other side of the board, that’s where!
Well that’s not how you usually mount things! I realized that by putting the LM317 on the bottom of the board folded over backwards I can use the board itself as a heatsink clamp. I have to keep the area above clear, but even with a tiny board it’s not a problem.
One actual problem with this clever solution is my standoffs were too high. Some caliper measurements showed I needed a spacer about 2mm thick. The case I’m using is about 2mm thick, so I just recycled a chunk from the meter cutout. Wrapped in insulating tape, of course.
If I had to pin down the one thing that makes a successful engineer, it would be pragmatism.
After soldering up the various parts, there’s nothing much to do but plug stuff in. One connector was loose, but that was the only real problem.
Case Construction
Pre-made electronics cases can be quite expensive. This time though, I spoiled myself and decided to get a proper aluminum box. It’s remarkably ugly, has no style or character, and costs four and a half bucks. My main motivation to buy it is laziness. Making a box on this scale would be tricky with my current tools. Next time, perhaps.
The good thing about a pre-made box is you don’t have to do nearly as much measuring. I have to do a little to get the connectors in the right place. 19mm is the standard for double banana plugs. Everything else can be done with less accuracy. If it fits, it works.
Internally, I divided the box into rough quarters. Connectors at the ends, one section each for the circuit and meter. I would run into vertical clearance problems if I tried to fit the circuit and meter over each other. Mounting them on the same side would cause similar issues. This box isn’t quite as big as it looks.
Shiny dark pencil on shiny light aluminum was never going to photograph well. You can still see the four sections, which is what’s actually important.
Connectors are placed so you can hold the case in your left hand and adjust output with your right. Works okay the other way around too.
Clearly I should have flipped the position of the potentiometer and pluck-pack connector.
Eh, slightly longer wires isn’t an unmanageable solution.
Originally, I didn’t plan on adding a power switch. Practical experience quickly showed how short-sighted this was. I duly added one, but failed to position it properly:
Yes, they are touching. I ended up bending the contacts down to squeeze it in.
If it fits by twisting parts in a way the manufacturer never intended which likely negatively impacts future reliability, it works.
One more reason to double check your measurements before cutting. Oh, and the switch is a little loose because I over-filed the hole. Two more reasons to double check your measurements before cutting.
Mounting the display was tricky- tape won’t cut it this time. I planned to use some plastic standoffs to hold it down. But, I drilled the mounting holes a little too close together. Not wanting to modify the standoffs, I designed and 3D printed some of my own. My design ended up a little far back, but it’s perfectly usable.
I could just leave the LED display exposed. On the other hand, I have some castable resin that could create a neat little custom window. Might as well give it a go.
I overcompensated a little bit with keeping the resin contained.
Initial results were a bit rough. It took some serious work to clean it up.
A little plastic polish goes a long ways.
My camera hates bright green LEDs for some reason. You can see the difference, but it’s far more dramatic in real life. With my eyes, the unpolished display was blurry yet still readable. Post-polish it’s crystal clear.
Except for the giant gouge. You can see that regardless.
Bare aluminum is ugly as hell, but at the very least I can polish it up a bit. I’d consider anodizing it, but I don’t have the equipment set up right now. Maybe later. EDIT FROM THE FUTURE: I went on to do exactly that.
*sigh* This is as good as it’s going to get folks.
Testing
There’s not a lot to test. Just make sure nothing explodes under load and it actually fits in a pocket or the like.
Since the LM317 is already stable I don’t need to do any loop analysis. The less complex mathematics I have to do (both kinds), the happier I am.
Thermally speaking, the box has a surface area of about 68 sqcm. I calculate this to have a thermal resistance of about 6 C/W. Unfortunately I don’t have a good number for the LM317. Datasheet says 5 C/W, but the TO-220 can be anywhere from 0.3 C/W upwards.
Using the datasheet value means a combined resistance of about 11 C/W. For the 125C/W junction limit at 50C ambient, a little less than 7W of power can be continuously dissipated.
I could stop here. I did build a constant current load for a reason though. This is one of the ideal times to use it.
Unlike the bench supply, I can’t get clever with the output current/voltage controls. Simple Power adjusts the voltage, but Simple Load has to handle the current. Finally, a good reason to actually use Simple Load!
Based on these tests, I got a value of 11.5C/W. In more direct terms, that’s only about 6.5W capacity. These terms match up well with the estimated values.
Then I added a silicone thermal pad for electrical isolation. I expected this to increase thermal resistance by 1 or 2 C/W. Instead it dropped to 7C/W. This little addition almost doubled the power capacity! Goes to show what the little details can affect, and how inaccurate datasheets can be.
Finishing Up
Simple Power isn’t a particularly glamorous project. It’s very much a “my first project” sort of thing. Not exactly reaching for the stars as it were. To that end, I got to try some new ideas out on a low stakes project.
Using a pre-made case cut down on a lot of production work. I don’t really think it justifies the added cost though. You have to balance it against the cost of building one from scratch, including the time and tools. I already have (or want) the tools, time isn’t that important for a personal project, so building my own is always going to be the better option.
I overheated the power switch during soldering. It still works, it just doesn’t “click” properly. I’m planning to replace it with a lower profile switch later on.
Being tethered to a wall socket is annoying. While it would be impractical to just give Simple Power a battery, rest assured I’m keeping the concept on file. It would be significantly more complex, and I’m still working through the details.
Most small projects only need a fixed voltage- e.g. 5V for a microcontroller. As soon as I figure out how to build cheap, small cases I’ll probably build a few based on Simple Power.
Cost wise, Simple Power was more expensive than I first thought. The most expensive part is the plug pack(9.03USD), followed by the voltmeter(7.95USD) and the box(4.5USD). Percentage wise, they are about 70% of the cost. I had most of the parts to hand, but they were the cheap ones- which explains why I have them already.
Cast resin needs some special handling (and it’s far more viscous than I expect), but it’s probably the best option for small, clear windows. A full plastic sheet would work better for a larger project. You’ll want to invest in some polishing compound.
Using the circuit board as a heatsink clamp might have been a little too clever. For a proper PCB with real mounting points it would work much better. Either way it’s an unusual (though not unheard of) design trick.
Simple power was built to allow me to power stuff up away from the bench. Really though, I intended it as a low-risk test bed for some neat ideas. Honestly, I wondered if it would really be all that useful. However you might just have noticed it hanging around before…
It’s always the simple things you never really notice that make the most difference, isn’t it?
Have a question? Comment? Insight? Post below!