Created
01/15/08

I used individual polycrystalline silicon solar cells purchased from Electronic Goldmine. Each cell is rated .5V, 1.0A. Of course, that's somewhat optimistic...

I planned out my array ahead of time on notebook paper. I chose a series string of ten cells, for a nominal 5.0V open circuit, 1.0A short circuit.

The panel frame was made from cheap wood moulding. Woodworking is not my specialty. I screwed up the measurements and had to re-do some of the mitre joints. I ran out of material so I was forced to fill in a gap at one of the joints with a tiny little piece that, perhaps, no one will notice.

It turns out that glass is expensive when you need it cut to specification. Two panes cost me $20. I dropped and shattered one, of course. Now you see why I bought two of them! The glass is mounted to the frame with Lexel caulking goo. It dries clear and reasonably waterproof.

Array construction was the real hard part. Silicon cells are a real bitch to solder. The back side is slightly porous and sputter-coated with some metal - aluminum, perhaps? - this side, if sufficiently heated, sucks up solder with no difficulty. Sufficiently heating it is tricky. This requires a long dwell time with your soldering iron, and you run the risk of cracking the cell with every solder joint.

Topside connections are way worse. These are made to thin deposited metal traces which are coated with an oxide. You have to scrape away the oxide with a knife, without accidentally scraping through the metal. Then you have to solder it pretty quick, or it re-oxidizes.

After some trial and error, the method I had the best luck with was to melt a hefty glob of solder onto a trace and then remelt it just long enough to drop the pre-tinned end of a short length of wire onto it. Because the wire is pretty thin and I'm pushing an amp of current (optimistically!) I decided to attach two wires to each cell to make two parallel power busses and share the load between them. I harvested the wire from some scrap CAT5 networking cable.

Once I had attached topside wires to every cell it was time to connect cells together. For this operation I set the cells side by side and upside down with the topside wires from each cell hanging over the backside contact of the adjacent cell, adjusted the spacing until I was satisfied that the entire array would properly fit within the space dictated by the frame dimensions, and soldered up the backside connections to make two series chains of five cells each.

These cells use up a whole lot of silicon, so they make them as thin as they can in order to save on materials. They are terribly fragile. I was afraid that they might break if they were in contact with both a rigid front panel and back panel of wildly differing flexibility and thermal expansion coefficients, and I didn't want unexpected jars or shocks on the back panel to snap apart my solder joints. A sheet of shipping foam from a recent Internet book order was harvested to yield a couple of layers of shock-absorber. I tacked down the edges with Lexel, and then applied several strips of Lexel over the foam and put down my back panel. The panel is a piece of styrene sheet originally meant to serve as optical diffuser for overhead flourescent lighting. It looks and feels nice, but it turned out to be ridiculously brittle and has a habit of shattering when cut. After I managed, on my third attempt, to cut out a square piece of the appropriate dimensions without it breaking into little shards, I sealed it down with a continuous bead of Lexel around the edges and placed a few microwave oven transformers on it for weight. As it happens, I have a lot of them laying around my basement - they make excellent doorstops!

Here is the final result. It looks alright, I think, and moreso given that it involves wood and I can never make anything out of wood look right.

A couple pages from my notebook showing values measured on a somewhat sunny day with different load resistances. The highest power I observed was about 2.9 watts. This was definitely more work than should be necessary to build a fragile 2.9W power source! Still, it was fun.