TJIIRRS: Number 19

(04 July, 2007)

There are lots of interesting designs for electrostatic generators. The most common and familiar type of “influence machine”, these days, is the Wimshurst machine, but that has counterrotating disks, and is slightly complex. This is not to say that Wimshurst machines are really all that difficult to construct (see this page by Jarrod Kinsey for an example); but I decided to go with something simpler for now. This page records my construction of a Voss machine, with photos and with my comments and occasional suggestions about the various issues I encountered.

I have also provided a link (near the bottom of the page) to a site where you can find a lot of information about many kinds of electrostatic generators.

In its simplest form, the Voss or Töpler-Holtz machine has only a single rotating disk, and when correctly designed and constructed it provides more than adequate performance. In fact, multiple Voss machines (another example) were sometimes used at the turn of the 20th century to power X-ray machines.

(Unless they have corrected it, btw, there is an error in the description of the first photo on the Kenyon page. It says that the rotating disk is larger, but in fact, and as you can see in all of the photos, the rotating disk is almost invariably the smaller of the two. I strongly suspect that this has to do with the mounting flanges of the appropriator brushes; you can see them in several of the photos. Because the front disk has to be close to the rear one, the flanges do not permit the front disk to be as wide as the rear one. I will return to this issue in a few paragraphs.)

I found all of the rotating-disk electrostatic “influence” machines rather difficult to understand, btw, until I wrote my own version of this French page about the Wimshurst machine. In the course of making my translation I found that I needed to describe the action a bit more fully than the original page did, so I added specific notes about what happens when a sector is at various positions, and I also added a section about induction (formerly called influence, which may possibly be a better term). The result is really more of an adaptation than a translation.

The act of writing that adaptation was what really permitted me to understand the way the machine works, and I note this because it is something many people have found: if you want to understand something better, try explaining it to someone else. This will also let you know, in a big hurry, just how underappreciated teachers are in the US (or wherever you are). Technical writing is also difficult and frustrating; one of the few compensations is the fact that the writer gets to understand the material better by documenting it. If the writer is good (and lucky), other folks also get to understand it.

Sidebar: Construction of the Collet

If you decide that you want to build a Voss or other static generator, there is no reason why you need to do it the same way I did; that said, however, some people may want a bit more information about the collet, because it seems to be a decent way to hold the disk in place. I made it from two small wooden wheels that I bought at a hobby shop. (These typically come in packages of 4, which is handy.) Here is one of the wheels:

The back side is flat and very plain.

I slid two of these onto the shaft with their flat sides apart for maximum stability, and used J-B Weld epoxy to glue them together. That gave me a disk of double thickness, prealigned to the shaft. Here is a cross-section:

(I did not take photos as I was doing all of this, so I was obliged to make these illustrations from the completed piece.)

Then I drew a line across one face of the disk, through the center hole (the horizontal line that separates the two pieces, though I did not saw them apart at that point), and two more lines, to indicate the positions of the screw holes, at right angles to the first one:

Then I drilled 4 holes into the wheels, so that they were in wood all the way through, and some distance away from the shaft, using the lines as guides. You can see the holes in the cross-section, above, and here:

Originally, these holes were all the correct diameter for a tapped 8-32 hole. After I drilled them I sawed through the middle of the collet, which gave me two halves, each of which had 4 holes in it. Then I drilled out 4 of the holes to the body size of the 8-32 screw, and tapped the other 4. I built this collet with two screws facing up and two facing down, but I doubt that it makes any real difference.

(10 July, 2007)

I noticed that the screws were bending when I tightened them, so I went back to the bandsaw and created flat places for the heads to rest on. Here are two views of the modified collet:

(06 July, 2007)

Here is the assembly, static and rotating, though not yet in its actual bearings. You can see that there is little or no wobble as the disk rotates. (The blurry dark-gray vertical line in these two photos is the phonograph record, which you are seeing edge-on.)

The small wooden wheel at the left is from an earlier version of the design; I decided to use wheels of the next larger size, both for better angular control and because they have center holes of the correct size to fit the shaft, and I don’t have to drill them out.

As you can see, this setup appears to work decently well. I trimmed out the wobble, and took the opportunity to sand the edge of the disk with the handy Dremel tool, so that it doesn’t move in and out quite so much when the disk rotates. (The hole in the middle of the disk is fairly well centered on the hole in the middle of the pressure plate, but at least originally it was not as nicely centered on the disk itself.)

The “front” of the disk, the side the sectors will be on, is the one that faces to the right in these photos. The inductors will be on glass plates to the left of center, which is to say behind the disk. My next step is to mount the shaft in its bearings on stands that hold it up from the base. Then I need to mount the glass plates on the base, trim them so that they are parallel to the disk, and position the inductors correctly on them.

Here is an extremely tentative partial layout, with the shaft simply inserted in holes in the vertical posts; the bearings are not yet mounted.

We are viewing this from the other side: the front of the machine, with sectors and combs and brushes, is to your left; and the rear, where the drive pulley will eventually be, is to the right. The disk (not present in this photo) goes just to the left of center, and the glass plates with the inductors on them (also absent) are just about at the middle. The first drive motor I intend to try is the small fan motor in the foreground.

It is slightly annoying that the longer end of the shaft sticks out the front, but there is nothing preventing me from sawing off the excess, and I will probably do so later on.

One reason why this is so tentative is the fact that I do not yet know how large some of the parts will be, and that makes it difficult for me to decide where to put the parts I already have. At some point, and I will reach that point fairly soon, I’m going to grit my teeth and start positioning things. I will probably wait until I have the glass plates and their mounting frame pieces cut, though, because those are crucial.

(09 July, 2007)

I have made the pieces of the frame that will hold the inductor plates, and I have marked the base to show where the plates and their frame and the posts for the shaft will go:

As you can see, the posts for the shaft are warped. Fortunately, they are fairly symmetric, and I am largely ignoring the curvature.

I have just constructed a sleeve to use as a front bearing for the shaft. It would have been nice to buy a thick-walled tube, but I didn’t find one; I did luck out, though: I found a tube that the shaft just barely fit inside, a tube the first tube just barely fit inside, and a third tube that the second one just barely fit inside. I have epoxied sections of the three tubes together to make a tube with a reasonably thick and strong wall. I have a ball bearing that I am thinking about using as the rear bearing, but it is very small, and I’m not sure whether it will handle the forces involved. I may just cut the sleeve into two pieces; it is certainly long enough to make more than one bearing out of.

(11 July, 2007)

Here is a rather tentative preassembly, with the shaft present (but not held in position, and possibly with the disk facing the wrong way):

Having gotten this far, I can tell you that it is extremely important to allow yourself some “wiggle room” in all 3 axes. If you fail to do this, you will almost certainly find that you can’t get the shaft to rotate easily, and you can’t line up the disk with the plates that the inductors are on. As things currently stand I have not allowed myself quite enough wiggle room, and I am going to have to increase the amount.

(12 July, 2007)

I have attached the two glass plates to the frame that holds them, but not to the base: I want to be able to adjust the tilt if necessary. Here is an overview from the front, and two details of the back, with everything more or less in place:

The washer assembly that is visible at the left edge of the middle photo is my tentative plan for setting the distance between the disk and the glass plates. When the machine is fully assembled the metal washer will be held in place either by a small collet or perhaps, for simplicity, by two cable ties wrapped around the shaft. (Two instead of one because the heads are wider than the tongues. It would probably be best, actually, to use three.)

It appears that the tilt issue is not going to be much of a problem; I expect to deal with it by putting a thin shim under the front shaft post. My main concern now is the fact that if I position the disk close to the glass, the phenolic pressure plate scrapes. There are three obvious possible solutions:

1. Keep the disk further from the glass. (Annoying, and may interfere with startup and operation. On the other hand, clearly the easiest option.)
   
   
2. Grind the edge of the phenolic. (Dicey, because the phonograph record is so fragile.)
   
   
3. Grind the glass down a little. (Dicey, because glass is also fragile, but actually probably the best option, and the one I decided to pursue.)

(The results are visible in the photos where I show the inductors, below.)

Sidebar: Why We Love Japanese Saws

In the course of assembling the frame, I found that the crosspiece at the top was a bit too long. I went back to the miter box and sawed the end off. It was still too long, but only by a very small amount, so I sawed the end off again. Here are two views of the resulting chip; notice how nice and smooth the cut surface is:

The chip is thin enough that it is quite translucent:

You can see where I have carved away a small amount of glass, in the middle, to make a larger space for the pressure plate on the back of the rotating disk.

(16 July, 2007)

I need to be able to make contact with the foil, but that side of the paper is stuck to the glass. Carbon-based ink (Sumi ink or India ink, for example) is known to be conductive enough to serve as the sector material, but would it soak through the paper enough to let me contact the foil? I took a small circle of the same kind of origami paper the inductors are made from, and painted Sumi ink on the back. After it dried I scraped a bit of the anodized surface off the foil side so I could touch actual metal with the probe of my meter, and was very happy to discover that the answer is a clear “Yes.”

Then I thought about the fact that the disk may end up rotating either way, depending on whether I’m turning it by hand or with the motor, and that means I have to be able to swap the positions of the appropriators and neutralizers. Usually, each appropriator structure is built at the correct end of the inductor it’s in front of, and makes contact there; but (particularly in the case of metal-foil inductors) there is no earthly reason why it has to be that way, so I made my Sumi-ink dots at the outer edge of each inductor, halfway up. We’ll see how well that works when I get to the point of testing the machine.

Usually, only acupuncturists can get acupuncture needles. I managed to get my hands on one for long enough to take this photo, but I will be using sewing needles for the combs on my machine unless I luck out at a garage sale or something. (I suppose I could sharpen the sewing needles; but I am going to need about 4 dozen, and I’m not willing to spend the time it would take unless I am very sure that the results will be worth it.) It is definitely possible, btw, to sharpen one of these things. Here is the same sewing needle (upper) after I chucked it into the Dremel and rubbed it on one of my favorite ultrafine abrasives, compared to the same acupuncture needle (lower) —

(Sorry about the focus.)

Here’s the abrasive, a brown ceramic integrated circuit package (left). The purple (right) and white ones and most gray ones are too coarse for this, though I do use them for sharpening calligraphic pen nibs, which I then polish with the brown ones...

To whatever extent I can optimize charge collection it seems pointless to lose it to corona, and I plan on surrounding the needles (except, of course, the region directly in front of them) with a structure intended to minimize losses. Wood turns out to be a vaguely semiconducting material, and is very good for equalizing field contours and thus performing this function. I expect to build the brushes and combs of this machine from it, and I will include photographs as I construct those parts. In the meanwhile, you can find out more about this use of wood by reading the description of Winter’s machine at the beginning of Homemade Lightning, by R. A. Ford.

A typical Voss machine (as on the Kenyon page I link to above) has 6 sectors, but I am initially using only 4 on this device, for ease of construction. The sectors are present partly to get the machine to self-start, and partly to charge the inductors; the surface of the disk (at least, the part that is in front of the inductors) is the primary charge-carrying region. In this regard, the Voss machine is somewhat like a hybrid between the Wimshurst (which uses its sectors as charge-carriers) and the Bonetti (which uses an area of the disk’s surface as its main charge-carrier, and lacks sectors).

(16 July, 2007)

I don’t really need anything fancy for a first test, so last night I put a sewing pin under the microscope, next to the appliqué sharp needle. To my surprise, I found that the points were quite similar in size, though the barrel of the pin is thicker. Clearly, pins will do for a first test, so I took my Dremel and a #67 drillbit, and made 4 combs. Here are two of them:

The pins are 1/4" apart, and there is 1/2" of clear space at each end. There are 11 pins in each comb.

I made these by pushing the pins too far through the wooden bar and then pressing the bar down onto two pieces of glass about 3" apart on the bench, with a washer in the middle of the space between them. (You will notice that the 5 points in the middle do not stick out as far as the 3 on each end. This is so they won’t collide with the sectors if I position the combs very close to the disk.) Once everything was positioned nicely, I put a dot of cyanoacrylate glue around the place where each pin went into the wood. I will probably just use thin wire to connect to these; there isn’t really any reason to get fancy about it at this stage. The “real thing” will be entirely encased in wood in any case (except for the points of the needles), and thin wire will work just fine. The current involved here is measured in microamperes, after all.

Two of the combs need to have brushes added to them so they can serve as neutralizers. I am thinking about using some very thin metallized Mylar™ for this, as I happen to have a scrap of it here.

If I can figure out how to position these things, it won’t be long before I make a preliminary test.

(19 July, 2007)

Here are views of some of the bits, as constructed and in place. The machine is lying on its back, which accounts for the screwy angle. The handle is exactly what it looks like (a window crank), but I have adapted it by gluing a brass tube into it with J-B Weld epoxy.

(Those last two were taken while the epoxy was still wet.)

Here are two of the back side, showing the way I am connecting the inductors to the appropriator brushes:

Note that these are all temporary, for testing. The wooden bars are stuck to the microscope slides with RTV, and can easily be removed.

One important thing to remember: this machine is set up for clockwise rotation of the disk. As the sectors begin to pass in front of the inductors they pass under the appropriator brushes, and then as they begin to move away from the inductors they pass under the neutralizer brush/combs.

(evening of 19 July, 2007)

I tried running the machine, to no avail. I tried starting it with a piece of PVC pipe that I had rubbed with paper, but that didn’t work; I even tried connecting the appropriators (and thus the inductors) across the output of the little Wimshurst machine that I’ve been running my TEA nitrogen laser with. Still nothing. It behaves as if somewhere, somehow, charge is leaking fairly rapidly and preventing the machine from doing what it should. I have already checked a fragment of another similar disk with the Wimshurst, and it did not leak, so I doubt that the disk is the issue, but I will be checking it with the Wimshurst anyway, to be sure. (I did. It wasn’t.)

(Early morning, 21 July, 2007)

I smell ozone.

Two things (at least) are going on here. First: in the process of checking the appropriators and inductors, I discovered that the ink spot on the back of one of the inductors was not really making decent contact with the metal foil on the front. I took an X-Acto® knife and carved a tiny bit of paper away from the foil of the inductor at the outer edge, near the ink spot. Then I smeared silver conductive adhesive on the scrape and over to the ink spot. End of that problem. For obvious reasons, I then repeated this with the other inductor.

Second: it turns out that metallized Mylar probably isn’t the absolute best material to make the brushes out of. I discovered this by verifying correct conductivity everywhere else, and then finally putting one of the meter probes on one of the sectors, which was in contact with a brush.

Nada. Open circuit.

I ended up putting silver conductive paint on all four brushes, several times, after which I finally began to see some charging. I can smell ozone, and if I hold my finger near the disk while I rotate it, I get tiny shocks. It’s a Voss machine! Now I have to give it some output combs and see whether I can get reasonable performance out of it. (The RTV is setting, and I should be able to do more testing during the day.)

(some hours later)

Here is a photo of the machine with the collector combs in place:

(Sorry that’s a bit fuzzy — it was a longish hand-held exposure.)

With the combs in place, I was (just barely) able to use the machine to operate a TEA nitrogen laser, which was the original design intention. I suspect that things will be easier with “final” brushes and combs instead of the temporary versions I’m currently using, and with a motor drive instead of hand cranking. Things will also be easier when I find a better material to make the brushes out of. I will probably be testing some possibilities as my next step, unless I just decide to build a new machine. (I have a glass disk that is 4 mm thick and 50 cm across...)

(04 August, 2007)

Well, I had a disk like that. Turns out that it was was tempered, so when the glass shop tried to drill a hole in it, it turned into many tiny pieces. They are going to make me another one, 20" across, 3/16" thick, and with a 3/4" hole in the middle.

In the meanwhile, I have acquired some acupuncture needles. These are 0.12mm diameter (as thin as I’ve seen so far), and impressively sharp. Here is my specially sharpened “appliqué sharp” sewing needle, next to one of them:

I am constructing new combs with these needles, to see whether I can improve the operation of the machine before I build a larger one. Here are some early stages:

I am planning on covering the tops of the cutoff needles with another piece of wood, and I hope to connect to this array with wire that is insulated for 40 kV.

Here are the motor I expect to use, and its pulley:

The motor comes from a small ventilation fan or humidifier, acquired at a local thrift store. The pulley is a small one, removed from a little sheave that I got at the hardware store. I drilled it out so it would fit onto the motor shaft, and cleaned it up slightly so that it is fairly well balanced. It is held onto the shaft with J-B Weld epoxy.

I am still thinking about the shaft pulley on the machine, and how to attach it. (More about this later, as it begins to take shape.)

(13 August, 2007)

I have built a new set of brushes and combs for this machine, using 0.012mm acupuncture needles for the combs. Brushes and combs are encased in wood. (I will take photos as time permits.) I tried this setup with a second phonograph record, and was not satisfied, so I bought a piece 0.093" Lexan™ (polycarbonate); marked a 12" circle on it by tracing around a phono record; cut out the circle with a scroll saw and filed the edge a bit to smooth it; attached 6 washers for sectors; and installed it. A Neoprene™ washer (1/4" hole, 2" diameter) replaces the hard white foam and cork composition circle that I was using between the original disks and the collet. The pressure plate for this disk is the one that I made for the second phono record; it is cut from a plastic cap that originally graced a spice jar. (Again, photos later.)

There is no guarantee that this will work well, but I should note that when I removed the plastic protective sheet from the front of the disk it took on a nice static charge, which certainly bodes well. I have not bothered to remove the sheet from the back of the disk; I doubt that it makes any real difference. (If I change my mind later, I will probably be able to remove it without destroying the setup.) I should note that this new disk is not flat, and also that it is considerably more flexible than the phonograph records were. I don’t think wobble will be a huge issue, but the center of the disk bulges out (toward the front of the machine). This may worsen as the disk becomes charged and is attracted to the inductors... We Shall See.

As I write this, on the evening of August 13th, the RTV (aquarium caulk, in this case) is still curing. I hope to be able to test the machine later in the evening, but it may have to wait until morning. I will report the results if they are worth reporting.

(14 August, 2007)

While I am having a bit of trouble with the brushes and the bearings, this unmistakeably works. I ran a small TEA nitrogen laser with it, as a test. It is slower than the commercial Wimshurst machine that I normally use for that purpose, perhaps partly because it has only one active disk. There may also be issues with conductivity inside the neutralizers. Either way, it is obvious that polycarbonate is viable as a disk material, though it is not as stiff as I’d like.

(08 August, 2007)

In a paper that is available on the Web, Noël Félici offers some interesting insights on charge transfer. Our machines operate in what he calls “Low Mobility” mode. Under somewhat different circumstances, it becomes possible to transfer considerably more charge. Félici was granted several patents for his work; his machines produced as much as ~2 W/cm², with maximum power outputs ranging as high as kilowatts (!).

I would like to investigate the low end of high mobility (as it were), and I hope to rebuild this machine in a way that will permit me to do so.

First, however, I am going to replace the original disk with another, because something is wrong now and the machine isn’t working. As long as I’m doing that I will also be replacing the brushes and combs with new ones of a different design, to see whether I get good performance.

Once I’m reasonably satisfied (or the machine fails again and I conclude that phonograph records of the type I’m using are poor candidates), I will begin to think about a glass disk and more stringent conditions. I will continue to post as things proceed.

(16 August, 2007)

I bought a sheet of Lexan™ (polycarbonate), 12"x24", 0.093" thick, and made a disk from it. This clearly works, but I am still having some trouble with brushes and combs. I conclude that Litz wire is suboptimal for brushes, as it tends to be hard to strip (tiny wires with plastic coating, unless you can get ancient stuff with varnish/lacquer instead) and not springy enough.

Even so, I have been able to power my little TEA nitrogen laser with this disk, and I am once again thinking about motorizing this machine.

The best source of information about the Voss machine (and other electrostatic machines) on the Web that I am currently aware of is a set of pages by Professor Antonio Carlos M. de Queiroz. There are occasionally issues with the server, so if you cannot see these pages the first time you try, wait a while and try again. They are well worth it: there is a huge amount of information on them and in the links that Professor de Queiroz provides, and he has built a wide variety of extremely good electrostatic machines himself.

Back to the Index

Home

Contact Information:

My email address is a@b.com, where a is my first name (just jon, only 3 letters, no “h”), and b is joss.

My phone number is +1 240 604 4495.

Last modified: Thu Jun 23 16:04:29 CDT 2016