We’ve been hit with two big storms here in Boulder, Colorado, so I’ve had a lot of time indoors to work on the project.

The first step was to verify that my 12-volt power supply design will work. Normally you’d think this would be pretty simple, but there are all sorts of little problems that can creep in. To make this design small, I didn’t overbuild it too much -- that could cause problems.

Most power supplies are rated in terms of DC voltage and current. This is the amount of power it can provide to a device that is constantly on. Conversely, displays are rated at the amount of power they consume if they were on continuously.

The reality is that neither of these devices operate at a constant rate. The power supply puts out power in discrete bundles -- up to 1.6 million “bundles” a second. A capacitor on the output is supposed to provide energy in-between these bundles, but it’s not perfect. The result is that the output is 12 volts with a little bit of a noise that looks like a sawtooth.

The display doesn’t light all LEDs at the same time. In reality, it lights up only one row at a time. Like a TV, it scans the entire display, and your persistence of vision forms a non-flickering image.

A problem can occur if there is an interaction between the sawtooth noise on the power supply and the LED’s refresh rate. Under the right circumstances, the low part of the power supply waveform could be synchronized to the display in such a way that causes horrible flicker.  This is a little like when a computer monitor image is recorded on a camcorder.

I also tried inverting the picture so that most of the display is lit up. This requires much more current than normally used (so much that the battery would last only 30 minutes instead of 25 hours), but it also worked. The power supply made a lot of noise (which is typical of switching power supplies), but the output was a clean and stable at 12 volts.

- - - -

So, now that my power supply is verified, time to build a PC board!

In high school, I made a lot of PC boards in the lab. I used all sorts of nasty chemicals, spent lots of time, and in the end I had amateur quality boards. At the time, professional quality boards cost a lot of money because board making was very labor-intensive.  The boards were hand-drawn with rulers, black tape, and an exact-o knife, and keeping the original mylar sheets clean was a challenge.

Times have changed. Most design is done with CAD programs and the designs are submitted to PC Board houses electronically.  With CAD programs comes standardization, and with that have come lower prices.  Quality has improved, and one day turnarounds are no longer exorbitantly expensive. It is now often cheaper to send out a small board than to make it your self -- especially if if is two or more layers and requires drilling.

This leads me to the snow storm and the winter vacation: I wanted to work on this board to keep from going insane, and even with one-day-turns, there was going to be no way I could get a board back in time to play with. Luckily, I did a good job on the layout so that I only needed a single-sided PC board.  So, on the day that things cleared up, I bought some PC board making supplies at the store.

All PC boards start off as an even sheet of copper bonded to a fiberglass core. This is coated with a special light-sensitive film (called “resist”) that is impervious to acid. The resist is exposed with the desired pattern, and photographically developed so that portions of it wash away. Then, the whole board is placed in an acid bath where all the unprotected copper is dissolved away.

Boards start off on clear plastic (left). I used an ordinary laser printer and special film.  Then, resist is placed on the copper (middle). This is a photographic process that’s pretty simple. Finally, the excess copper is etched away with acid (right). Note that the green resist material is still on the board.

The final two steps are removing the resist film (I used “goof off” to dissolve it), and then plating the board in tin.  Without the plating, the bare copper will quickly anodize (a.k.a. rust) and could fail.

The original clear film I used had a few problems. The toner didn’t adhere perfectly, so some of the lines had holes in them. I could have touched this up with a dark pen, but I was afraid I’d flake off more toner particles. That, and these lines are really thin so I was afraid of “coloring outside the lines”

As you can see, the holes in the original made holes in the final product:

This is a little sloppy, but good enough!  There were no major gaps, and I used solder to reinforce the wires that were partially compromised.