Building the AstroMedia Wimshurst Machine
While I waited for my kit to arrive I started gathering the tools and materials recommended on the maker’s web page. Since the site is in German, and I wasn’t sure about Google’s translation or what the US equivalent would be for some European products, I asked for help on http://www.papermodelers.com/ . I ended up buying Beacon 527, UHU all-purpose adhesive, super glue gel, and E6000. I already had epoxy, Scotch tape, and all the tools I needed. I ended up not using the E6000, and the epoxy gave way the first time I turned the crank, so I rebuilt the crank with a #6 machine screw and J-B Weld Steel-Reinforced Epoxy.
The instruction booklet was 16 pages, 8.25” x 11.625” (210 mm x 296 mm). I assume it was translated from German to English. It was a good translation, quite clear, but more pictures would have been helpful. I didn’t have to look far: the maker put step-by step photos on his website: http://michelswunderland.de/solderiron/wims.html . Very helpful. If you’re considering this kit, you can follow every step in these photos, and see exactly what construction entails.
I had planned to take photos during construction, but this is a very time-intensive build. The maker says to allow 8 to 15 hours’ construction time. If I built several in a row, I might approach those times, but this is the most I’ve ever spent on a model, so I proceeded with caution. I had two weeks’ vacation, and I worked on it several hours every day (and night, often into the next day). I finished on the tenth day at 8 pm. If I’d stopped for photos, I don’t know when I would have finished, but not during this vacation, and I don’t think I could have added much to the step-by-step photos linked above.
I don’t know what the parts count was, but it’s high. I was glad the parts were clearly labeled. The instructions are broken into chapters A through S, and each part is printed with a letter and number. The letter matches the corresponding chapter, and the numbers generally follow the order of assembly. Each step calls out every part by letter and number. Better yet, on each side of each part is a symbol followed by the letter and number of the part that gets glued to that side. I still made a few errors, occasionally gluing a part with the wrong side out, but these were minor cosmetic flaws, not functional problems. I give this numbering system credit.
The instructions recommend cutting, rather than tearing out the parts (they are die-cut, with multiple >1mm tabs on their perimeters holding them in place) and I started cutting each part carefully with a sharpened (I use a diamond hone) #11 X Acto blade. I later found that if I pressed on the edge of the part next to each tab with a toothpick, the tabs would tear neatly. The toothpick concentrated the force so that if I stopped pressing at the first sign of bending, I wouldn’t crease or curl the part. This greatly reduced cutting time, and the tabs cleaned up quickly with sandpaper or a nail file.
Almost every part is built up from multiple layers of card glued together. I knew it would be critical to spread a layer of glue that would cover the entire surface, but thin enough to avoid wrinkling. I also knew the glues I had experience with would not do. I cut a sheet of Bristol Board from my stockpile in four pieces, and glued two with UHU and two with Beacon 527, and pressed them between two boards to dry. The next day, both were well-bonded, with no hint of a wrinkle. Their respective web sites list UHU dry time 10 minutes, cure time 24 hours, and 527 dry time 1 hour, cure time 24 hours. For some parts, I was glad for Beacon’s longer working time. For others, I was glad for UHU’s quicker dry time. Thanks to rickstef on papermodelers.com for recommending Beacon 527. It works great, I can get it at my local Walmart, (I used two tubes of it on this model) and UHU costs four times as much.
As the major structural members, the base and supports require strength and rigidity. each starts with gluing two layers of card together, adding perpendicular zig-zag pieces for support, then gluing another double layer on top. Then a strip around the edge covers everything neatly to make strong and good-looking slab. It’s an impressive bit of paper engineering. I probably would have cut them from wood to save time.
I found a small kitchen spatula with a flexible 1” x 2” (2.5 cm x 5 cm) silicone blade an effective tool for spreading a thin, even layer of glue over a large surface. as soon as the glue dries, flex the blade, and the glue peels off.
The web instructions suggest the option of mounting the base on a slightly larger piece of plywood for improved stability, and adding rubber feet to the plywood to keep it from sliding. I did, and I’m happy with the result.
The brass inserts in the terminal strips that hold the brushes appear to have been replaced with steel. I couldn’t get the inserts out of the plastic strip, but I cut the strip into six pieces, and it works.
The “Adjustment” or “Calibration” bench is a clever way of squaring the disks and hubs to the axle, so they run true, an important and sometimes tricky aspect of any Wimshurst design.
The disks are 210 mm (8.27”) in diameter. The disk segments are precut in adhesive-backed aluminum foil. They are cut with four straight sides. This will work, but the sharp corners will be prone to corona discharge, reducing the output voltage. I used Dr. Antônio Carlos M. de Queiroz’ WMD program http://www.coe.ufrj.br/%7Eacmq/programs/wmd.zip to design the sectors. Using the numbers from WMD, I drew a template with a ruler and compass, traced it on the precut sectors with an Ultrafine Sharpie, and cut them out. Because the sectors were precut, I could only make them rounded by removing material. Dissatisfied with that result, I took my template and marked out new sectors on aluminum foil tape, and cut them by hand. The sector pattern in the book was still helpful for positioning the sectors, and I was happier with the rounded sectors I made. Not perfect, but no sharp corners. I found a soft pencil eraser an ideal tool for smoothing the sectors flat against the dish. It left some smudges, some I removed with a damp cloth, and the rest with a Mr. Clean MagicEraser. Then I wiped the disks with a dry towel, and dried them with a hair dryer.
One problem I had when I tried to build a Wimshurst from scratch was getting the belts to run on the pulleys. They would slip and not turn, or the belts would slide off entirely. This kit builds up the pulleys from card. They have a deep groove, and the tension is just right; they do not slip at all.
The Leyden Jars are rolled from a sheet of transparent PVC, with adhesive foil inside and out. The ends are plugged with card. The PVC is temporarily kept from unrolling with cardboard rings, then the seam is taped, and the layers glued with Super Glue (cyanoacrylate). The Super Glue fogged the PVC, so mine aren’t a s pretty as the ones on the web page, but they work.
Of all the parts in this machine, the crank took the longest to build. The handle has a wooden axle, on which the handle (rolled from paper) rotates freely. A cap on the end of the axle keeps the handle from sliding off. The body of the crank is built up from many layers of card, the wooden axle is glued into a hole at one end, and the aluminum drive shaft is epoxied into the hole at the other end. I allowed the epoxy to cure, but it broke loose at the first turn of the crank. I found I could fit a 6-32 machine screw into the hollow drive shaft, so I tightened the crank on the screw between two nuts, coated the entire screw with J-B Weld Steel-Reinforced Epoxy, and slid it into the aluminum drive shaft. I let it cure for a full 24 hours before moving it, and then tried the crank. It held.
I noticed later that, unlike most Wimshurst machines, this one only collects charges from one of the disks. There is a pair of neutralizers for each disk, but only one pair of collectors. It should be possible to increase the current output by adding a pair of collectors to the other side. Probably not twice as much, but I’m not sure how to calculate the amount.
The large and small spherical electrodes are made from wooden balls wrapped in adhesive aluminum foil. I was skeptical. I know how hard it is to wrap a flat sheet of anything around a compound curve, how easy it is to wrinkle foil, and how hard it is to get it smooth again. I thought about simply painting the spheres with some conductive nickel paint I had, but when I tried it on a scrap of wood, it was ugly. The instructions say that because the foil is only 0.1 mm thick, wrinkles can be smoothed flat by rolling the spheres on a hard surface with your hand. I figured if it didn’t work, I could replace the spheres with ball bearings, so I tried it. It worked remarkably well. Not as smooth as a polished ball bearing, but there weren’t any sharp edges to speak of.
At this point, all that was left was final assembly. Everything fit together neatly. It took a bit of stretching to get the rubber band belts in position, and I worried that they might be too short but, once in position, the tension was just right.
I turned the crank, and it worked on the first try. I adjusted the neutralizers and the electrodes, and found with a one inch gap, I got one spark for every turn of the crank.
