TJIIRRS: Number 13,
Handy Techniques and Devices:

A Simple Mirror Mount,
A Simple Translation Stage,
A Simple Differential Screw for Fine Adjustments
A Simple Hand-held Spectroscope
a Link to our Vacuum System

A Simple Mirror Mount

(21-22 August, 2007, et seq.)

I have been working on various projects, some of which require mirror mounts. I never seem to have enough of these, so last night I started thinking about whether there might be a relatively easy way to build them. I have a few fancy commercial mounts for times when precision is more important, so I decided to go for a design that may not be quite as precise, but requires only parts that can be purchased at hardware stores. I later amended that to allow parts purchased at hobby shops and perhaps even bicycle shops, and I should probably note that I regard thrift stores as being included in the list of permitted sources almost automatically.

As I proceeded, I realized that I also wanted it to be possible to build this mount without any use of machine tools. I do not currently have ready access to a lathe or a mill, and it is clear that many other DIY folks are in the same boat. Although I eventually used a drillpress to drill a few holes, that is not essential; you could easily drill them by hand. (In plain point of fact, most of the holes are not essential — there are simple workarounds for them.)

This page goes through my entire building/learning process, and it is a bit long. If you don’t want to slog through the entire megillah, you can cut to the chase. On the other hand, I found the process interesting enough to leave in place for anyone who wants to read it. Besides, if you just look at the latest and spiffiest version, you won’t necessarily know how to put a steel ball on the end of an adjustment screw. Of course, if you prefer to solve that puzzle (or any of the other puzzles involved here) yourself, you are encouraged to skip ahead.

First, here’s a photo of the parts I bought when I started the project:

The total cost of what you see here was just under $17; but you also have to add the cost of glue and a few nuts and bolts that aren&squo;t shown here, so let’s call it $20. That’s unrealistically high, though, because there is more here than you need for one mount.

This mount is very easy to construct; it took us (I had help and advice from our loyal Range Safety Officer, LisaJulie Peoples) only 2 or 3 hours to assemble the prototype, even though we had to figure out what we were doing as we did it.

Please note that this first version is what I like to refer to as a “PoP machine”; that is, it is intended to provide Proof of Principle, and is not as robust as a final version. For example, this one is glued together with CA (Cyanoacrylate) glue; the real thing, further down the page, is glued with epoxy, which is much stronger and less brittle. It also doesn’t deposit white powder all over everything in its vicinity as it cures. (Ahem. Grump.)

Not Quite Step by Step:

I was too excited while building the prototype to take photos of the early steps, but I have annotated the first one to make this a bit easier to follow.

Draw or scribe lines on the back of the rider. They should be in line with the holes that define the X and Y rotation axes. (These lines are indicated in purple in the photo, immediately below.) The pivot goes where the lines cross; glue on either a steel ball (which can sit on a washer for added gluing strength and ease of positioning) or an acorn nut (which is easier to find, and will work adequately well). There is a detail photo, further down, showing the way I made the pivot on the prototype.

[I did not, btw, perform any machining on these pieces; all modifications were made by gluing things on. The one bit of machining shows up later, on the mirror holder.]

Glue nuts on the outermost holes of the rider. The adjustment screws will pass through these. (Later, we will add a secondary nut, to take out most of the backlash; for now, though, a single nut should do.) If you have trouble with the countersinking, just glue a washer on first. Then glue the nut over the hole in the middle of the washer. Remember to keep things as well centered as possible.

[NOTE: I describe a considerably more precise version of this mount later; you probably want to read through the entire description before you start acquiring parts or doing any construction.]

Position the rider on the stand, with a washer under the steel ball or acorn nut. (Your choice about exact positioning, but I will suggest that it is probably easiest and most effective to match the upper/inner edges.) Glue the washer into place, being careful not to get the acorn nut or steel ball stuck to it.

It’s a good idea to round the ends of the adjustment screws a bit, so that they move smoothly on the surface of the stand. You may also want to file or sand the surface smooth. For the proto, however, I didn’t bother. (Please note that we will get to a considerably better way to deal with this issue in a later version.)

I put smaller washers under the ends of the adjustment screws, to keep them from wandering about; but if you can arrange short narrow channels for them to “walk” along, that would be significantly less wobbly and more precise. (I actually did that in a later version of the mount, which is described further down the page.)

I glued the gusset and stand to the baseplate together; then I attached the rider to the stand.

You need at least one spring to hold the rider onto the stand, so that the adjustment screws will have something to push against. If you can attach a short strong spring in the inside corner, that will do. I didn’t have a spring of the right type, so instead I took a pair of o-rings from the assortment, and used one on each axis. They were just a bit long, so I put a twist into each one to tighten it. (Again, see the photos.)

Here are some views of the completed frame, before I added the mirror holder:


Here are two details, showing the washers I am using to keep the adjustment screws from wandering, and the pivot point:


A steel ball, if you can find one of a reasonable size, is a better pivot than an acorn nut. I found steel balls at my favorite local hardware store (an Ace Hardware called Zimmerman’s), but I built the prototype with an acorn nut because those are much more commonly available. We will get to steel balls when we do the improved versions of the mount, further down the page.

Here is a detail showing the second nut on one of the adjustment screws. I tightened this nut only until it became very slightly difficult to rotate the screw; then I glued it into place. This essentially eliminates backlash and wobble, and makes the mount considerably more precise. It also reduces the travel range slightly, but if I can find a longer screw at some point I can substitute it for the one that is currently in place.

[Note, added on 15 September, 2007: if you use sufficiently strong springs, and if the round ends of the adjustment screws are held in place, as they are in the third version of the mount, you probably don’t need anti-backlash nuts. They are, however, easy enough to make, and they do stabilize things, so I would suggest using them anyway.]

I made the mirror cell from large washers and o-rings. You can see most of the details from the photos. The one bit of machining I did was to drill holes in the mirror cell pieces for the bolts that hold it together. I used a drillpress for this, but you could do it with an ordinary drill if your hand is reasonably steady. (If you want to get fancy, you could drill both ends with the tap-drill size, then redrill one end with the body drill, and tap the other end. I just glued nuts onto one end.)

Here is what the mount looks like with the mirror holder in place:


The two washers that are visible in the photo on the left are basically gussets; the joints between the mirror holder and the frame consist of tiny dots of CA, and just aren’t strong enough, so I have reinforced them. (I would probably reinforce them even if they were epoxy, but you should probably take a look at the later versions of the mount for a slightly different and somewhat easier approach to this entire issue.)

Because I wanted to have a broad surface for the washers, I installed the mirror cell with the mirror hanging out the front. This is perhaps not the best orientation; left to its own devices, the mount falls on its face. While it is true that you will ordinarily be holding it firmly in place on an optical table of some sort, it is still a good idea to have it be stable on its own; you can use a wider baseplate, or use narrow gussets and orient the cell with the mirror behind the rider, rather than in front of it. (In fact, I reworked the prototype this way; see photos, below.)

This mirror holder is intended for 15mm optics. I also built one for 12.5mm or ½" optics, but did not put it on:

I am thinking about ways to make it easy to swap cells or to build a cell that holds multiple sizes of mirror, and if I get somewhere with either of those I will add a description.

Here are two photos of the mount on the bench, with a mirror installed in it. As you can see, I have broadened the baseplate so the mount doesn’t tip over.


After I did that I found some very nice mending plates at Zimmerman’s, and I may just swap out the base. Here are the mending plates, the single-edge blades I hope to use as replacements for the washers that the adjustment screws ride in (I was lucky enough to find very small ones, but if you can’t find little ones like these, you could probably use injector razor blades), and steel balls in several sizes:

Version 2:
Let’s Try For a Bit More Precision

Having noticed that the ends of the adjustment screws were wobbling around in the washers that I used in the prototype, I decided that I wanted to have them ride in short channels, which would constrain them along the axes. (This is common in commercial mounts.) In order for this to work, the screws must terminate in smooth rounded ends. Rather than try to round them off, which is difficult even with access to machine tools, I decided to attach small steel balls to them. That’s not entirely trivial, so I thought for a while about how to accomplish it.

My first decent idea was to take a drill that is just big enough to make a body hole for the bolt, drill into something firm, drop the ball (which, obviously, can’t be any larger than the drill size) down the hole so that it lodges in the conical end shape left by the drill (which automatically centers it), put a dab of glue on the end of the bolt, and push the bolt in until it touches the ball. When the glue has set, you can remove the bolt.

This method requires a hole that is just barely big enough. The M3 bolts I acquired (initially intending to use them in the translation stage that I describe below), are about 2.91 mm in diameter; The closest drills I had on hand were ~2.75 and ~3.13. (I measured them with cheap electronic calipers, so don’t trust the last digit.) I thought about that for a while, and decided that if I used a slightly small drill, and drilled the hole in fairly soft material, I could probably just screw the bolt into the hole until it reached the ball. This should provide better centering than a hole that is wide enough for the bolt to wiggle inside it.

The first material I tried was end-grain balsa wood. That’s probably a little bit too soft, so I went for end-grain basswood on my second attempt. I am now beginning to think that it is probably necessary to clean up the end of the bolt a bit before attempting to glue the ball to it, and I am not sure that even basswood is hard enough.

(27 August, 2007)

Yesterday I drilled into a harder piece of wood, and found that it seemed to work slightly better. I also measured a 4-40 screw; it was about 2.80 mm diameter, so the drill that measured 2.75 mm is probably a nice fit for it. Today, however, I went to Zimmerman’s and discovered that a #32 drill is just the right size for an M3/0.50 bolt: my calipers give it as 2.93 mm, and if I drill into pine (I used a scrap that had been cut from a 2x4) the screws just slip into the holes with a little effort. This ends up working extremely well, though occasionally one of the balls ends up being somewhat off-center, so you need to check them before you use them.

Here’s a photo of an M3 with a ball on it, sitting on the piece of scrap that I drilled the holes in:

I then went back to Zimmerman’s and found special wire-gauge drills. The #34 drill turns out to be just about right for a #4-40 screw, and I had no trouble putting small steel balls on the ends of the adjusting screws for the mount. (More about this, however, later.)

As long as we’re going for precision, I removed the acorn nut from the stand and put a washer in its place. This time, we will use a steel ball. I also scribed lines on the stand, to indicate positions for the V-channels that the little balls on the ends of the adjustment screws will ride in, and also for the washer that the pivot ball rides in. Here is a photo of another L, showing this:

To make the V-channels, I started with one of the tiny razorblades I show above. I snapped it in half, because the travel distance involved is so tiny that I don’t need anything like the full length. (Safety note! Use Vise-Grips® or other locking pliers, wear safety glasses or goggles, and snap away from yourself! Also, be careful of your fingers around sharp edges.) The blades are quite brittle, so it should be relatively easy to snap them. If you are having trouble, re-examine your setup.

The blades need to be tilted to create channels, so I glued a 1/16" ball under each of the pieces, and used CA to glue them down to the stand. (Remember to wipe the oil off them before you try any gluing. Again, be careful.) Here they are, before I glued them down:

As you can see from the next photo, there wasn’t enough room for a steel ball under the opposite side of the channel, so instead I used short pieces of wire-wrap post, which I happened to have lying around. This makes the channels asymmetric, but I doubt that it will be an issue.

I found that I had to snap the back edges off the blades, because otherwise they were unbalanced and fell off when I tried to position them. In fact, gluing the bits for this was very delicate work, and I had to reposition things once or twice.

As long as I had things apart, I flipped the traveller over so that the mirror cell would be over the [new] base, and the mount would not try to tip over. I also substituted M3/0.50 screws, which have finer thread pitch, for the #4-40s I was using. The 3/32" balls work very well with M3 screws, but if you go with #4-40 screws you will also want to go with 1/16" steel balls, which will fit through #4-40 nuts. That, or be sure to put the nuts on before you do your gluing. Ahem. (Yes, my mileage varied. The 4-40 socket-head capscrews look very nice with the 3/32" balls on them, but when I tried to put them back onto the mount, they, uhhh... hmmm. Yes. Well. M3 it is, then. We will, however, return to this issue later, and make good use of it.)

The positioning of the pivot appears to be even more tweaky and crucial than I had thought, and I’m not sure I have it right even now. It turns out that the channels fix the position of the ball, and if the washers for the pivot are not precisely where they need to be, one of the adjustment screws (or both) won’t sit in the center of the channel, or will walk out of the channel as you adjust the mount. This also happens if the channels are not aligned well, which is basically saying the same thing. I may have to figure out a more advanced way to build this version. (Makes me want a rather precise 3-axis positioner, but that’s a large project in and of itself.)

I am lightening the mirror cell, which is just too heavy for this somewhat more delicate design. I have ditched both of the larger/thicker washers, and I’ve been looking around for other ways to hold the o-rings in position. One thing that suggests itself is part of a faucet aerator, which you can see in the photos below. Another is the neoprene washer from a garden hose. Both are very slightly large, but I think they’ll do.

Here are two views of the nearly-completed mount. (The back half of the mirror cell is not attached.)


It occurs to me (time for a small epiphany, I guess) that the only way to prevent at least one of the adjustment screws from walking out of its track as the other screw is moved through a wide range is to put the pivot point (which in this design is at the center of the pivot ball) in the same plane as the V-channels. (You can accomplish this by drilling a hole, just almost large enough for the ball to go through, in the stand. Glue the ball in place. Let it ride in a washer that, if necessary, is held a short distance away from the rider on a stalk.)

In the meanwhile this mount is quite usable over a range of perhaps a degree or three in both axes, and I think it’s a very decent start. On the other hand, it’s annoyingly tweaky to build because everything has to be lined up really carefully and accurately, so...

Version 3:
It’s Time to Rethink (Again)

(29 August, 2007)

Version 2 of the mount is certainly viable in a general sense, but I want something simpler that works at least as well. It occurred to me that instead of trying to get the ball-ends of the screws to run in channels, it might be possible to maintain the position of the ball end, and take up the slack elsewhere.

The issue here is that if I position the ball end of the screw firmly, the angle that the screw makes with the rider must change as the angle of the rider changes. If the pivot point is not coplanar with the ball ends of the screws, the angle changes in both axes. I originally thought about bearings and/or slides, but the aforementioned Range Safety Officer suggested neoprene, in the form of o-rings, grommets, or faucet washers, for the compliant couplings. At first the neoprene idea seemed like it might be too loose, but after some thought I decided to give it a try. I found, somewhat to my surprise, that faucet washers are probably too stiff, and I have switched to o-rings. O-rings appear to be just about right for this, and the revised design is noticeably simpler than the previous version. (There will, however, be further changes, so please keep reading.)

I am working up a diagram to illustrate the angle issue, but it isn’t finished yet. When it is, I will post it here. In the meanwhile, here are some photos of the mount as it goes together. First, a detail of the pivot point and a detail of one of the adjustment screws.


As you can see, I have drilled a hole in the stand. I could have put the holders for the adjustment screw ends on blocks as an alternative, but this was easier. (The holders, btw, are just #4-40 nuts. Inasmuch as it was clear that the little steel balls wouldn’t go through them, they seemed like the obvious thing to use here, and they work quite well.) If you build one of these, it’s your choice. I will note, btw, that the column you see on top of the pivot ball in these photos is a little bit too long for the adjustment screws I’m using now; I will be removing the nut that is currently acting as a spacer. (In the photo on the left, and in the inverted photo below, also on the left, this nut appears at the base of the column.)

In the photo on the right you can see the o-ring, just below and to the left of the anti-backlash nut assembly. (There is a washer between it and the nut; I used CA to attach the o-ring to the rider and the washer, and then epoxy to attach the first nut to the washer and the second nut to the first one.) Here are two overviews of the mechanism:


Here is one of the screws just about fully extended, and just about fully retracted:


The dark liquid on the screw in the photo on the left is a wonderful lubricant called Dri-Slide®, which is made by Lilly Industries. It contains molybdenum disulfide, and will even work in vacuum after the carrier evaporates, which takes a while.

Here is the basic mechanism on the base; I have removed the hex nut (and the washer) from the column extending “up” (in this photo, to the right) from the pivot point to the rider:

I would not have bothered to remove those extensions if my M3 adjustment screws had been 25 mm long; but the longest I could find at the store was 20 mm.

It’s not quite over yet, though — I still have to decide how I will make the mirror cell. I may bite the bullet and design a cell that can be swapped out, so I can change mirror sizes easily.

(30 August, 2007)

After considerable fussing in the plumbing department at the hardware store, and then searching through my odds and ends here, I have come up with what I think is a viable mirror cell, at least for 15 mm mirrors. (It may be possible to adapt it to 12.5 mm mirrors, depending on the availability of appropriate washers and o-rings.) Photos:


This is the body of a sink aerator, with a washer dropped into the bottom as a stop; on top of the washer are two o-rings (obviously, if I just put the smaller one in by itself, it would not remain centered). The mirror is centered by a neoprene washer, and is held in by an adapter onto which I epoxied another washer (again, as a stop), and into which I dropped another o-ring. (Adapter, washer, and o-ring are on the left in the first picture; mirror in the middle; aerator body, washer, o-ring, and neoprene washer on the right.)

(Note, added on 31 August, 2007)

I went back to Zimmerman’s and found that they sell aerators with the adapters already in them, which makes this a whole lot simpler. Here are two types:

These do require a bit of care when you go to attach them to the mount, as they are tapered, but it isn’t particularly difficult.

(end of note)

The mount is now complete; let’s put the mirror in, and see how well it works...

...The first thing I noticed, after I got it onto the bench, was that the adjustment range is huge; it seems to be about +/- 5° in each axis, though I haven’t actually measured it yet. (I put a HeNe with a white card on the front a little less than a foot and a half away from the mount, so I could see the return, and at the end of the adjustment range the return was off on the wall someplace, a surprising angular distance from the laser.)

Coarse adjustment was easy, but when it came to fine tweaking I found myself wishing I had socket-head cap screws. Such is life: I used what I could find. (You can always go to MSC Industrial Supply if you decide you want stuff that is fancier than whatever your local hardware store has in stock.)

Here are two photos of the front of the laser, with the return beam from the mirror going straight back into it:


The laser did a lot of mode-hopping at first, even though the mirror is quite blue and does not reflect much in the red part of the spectrum. You can see some of the appearance of mode-hopping in the photos above, though of course it is a dynamic process, and I should probably take a video of it. What really caught me by surprise was something I was not able to capture directly with the camera: the laser also changed color for short periods (!).


Sidebar: A Brief Note about HeNe Lasers

If you were not already aware of this, HeNe can put out several visible lines, including green (543.5nm), yellow (594nm), and orange (612nm), in addition to the usual red at 633 nm, a slightly longer red, and some lines in the IR, of which I think 1523 nm is the most common. In fact, the first gas laser was a HeNe, and it ran on one of the strong IR lines, not the red we are now so familiar with. (That came along a month or two later.)

In connection with this, btw, I seem to recall that 6 or 7 of the lines share a common upper state. Because the cross-section for stimulated emission decreases as the wavelength gets shorter, it is easiest to get lasing on the longest line of the set, which happens to be an IR line. In fact, if you have a very high-power HeNe, with a long tube, you have to take steps to get much power in the visible, as the IR line tends to operate even with red mirrors, and it steals power from the red line.

The green line, being the shortest, is the most difficult to lase, and 543.5 nm “GreNe” [that may be a trademark] lasers generally don’t put out much power — I think half a milliwatt is fairly common. They also tend to be rather expensive.


After lots of fruitless attempts to show the color change by pointing the camera at the laser, I put the spectroscope (bottom of this page) behind the mirror, and managed to get some evidence.

Here are 5 photos. The first shows just the regular HeNe red at ~633 nm; if you look at the second carefully, you may be able to make out just the ghost of a second line, a little to the left of the main one. In the third image there is a line to the right of the main one (possibly 640.2 nm), and at least two ghosts to the left. The fourth image clearly shows the orange line to the left, and the line to the right; in the fifth, the line to the right of 633 has gone away, but there is still some orange coming out of the laser. (The camera does not capture the colors well; the line that is well separated from the main line on the left is distinctly orange or yellow-orange to the eye, and the main beam, which is overexposed here and looks bright yellow or almost white, is the usual HeNe red color.)


(These are full-size crops from the original images, so unfortunately I don’t have any that are larger.)

The laser did less and less mode-hopping as it warmed up and stabilized, and there were fewer and fewer moments when it changed color, though I found that I could sometimes get a certain amount of color change for a brief period by tweaking the mirror a little.

These photos are not really what this is about; but this is the first time I’ve been entirely sure that I was seeing a plain-vanilla red HeNe putting out other colors, and I thought that was pretty interesting.

In any case, it is abundantly clear that although the mount is a bit soft, in the sense that it bends a little if you push with the screwdriver when you are adjusting it, it is a pretty decent mount. It is reasonably easy to adjust and set, and it holds a setting fairly well.

If you don’t need the very wide adjustment range that this specific version provides, you could easily just widen the holes in the L bracket a bit, epoxy thin washers over them, and put the adjustment screws through those. Remember to retract the screws most of the way when you attach the receivers that the balls ride in, to minimize the maximum angular offset. That should give you a stiffer mount. In fact, see below — I decided to try doing this.

Revisionist Tendencies

Here is the beginning of what I am currently calling Rev 3A. It will have less range, but should be more precise, as I am using larger brackets. It should also be slightly stiffer; instead of o-rings, I am using the very slight springiness of brass (bronze?) washers. Here are some of the pieces:

The first step is to attach the large washers to the traveller; then the small washers go on top of those. In the photo on the left, I have some machine screws as positioning aids, and the epoxy is on the large washers, waiting for the small ones to be placed. Once all of that has set, the anti-backlash nuts go on the small washers (which, btw, are #6), and the anti-backlash nuts go on the small washers. In the second photo, I have the first half of each anti-backlash nut in place.


Notice the fact that I have (carefully!) sawed off the lower edges of the washers, so that the bottom of the horizontal-axis one won’t hit the base. I haven’t decided which one will end up being horizontal, so I sawed them both.

(03 September, 2007)

The above version of the mount, when I tested it, worked fairly well; but I did notice that it was still a bit more rubbery than I would have liked. When I tried to adjust it, the pressure of the screwdriver made a noticeable difference in the setting, which made it hard to get precise control. I did succeed, but it was by guessing where the aiming point would be when I let go.

That just isn’t good enough to satisfy, so I thought it out, and concluded that at this scale even steel is fairly soft if it’s thin enough. For decent stiffness you want either thicker sections, box beams, or even Buckminster-Fuller-Style Octet Trusses, if you have the materials and the patience it would take to build them. I decided on box beams, for simplicity and ease of construction. (That involved the hobby shop in this project — only a few hardware stores carry the rectangular-cross-section brass tubing that I used.)

It also seemed like a good idea to have the screws go through the stator and to put the pivot point on the rider. I am calling this improvement Rev 3B rather than V4, as I could have tweaked any of the versions this way, and also because I built it by reworking an existing mount instead of starting afresh.

Swapping the two pieces was easy. I did drill out the holes in the backup plate for the stator, because they were just a bit too small for the heads of the screws to fit through, but this hardly counts, because you could do it with a file in a few minutes. Here are the constructed boxes:


Here is the mount on a base, before I built and installed the mirror cell:


The mirror cell for this mount consists of 3 main pieces, as that was about the easiest thing I could put together from what I had on hand. 2 of the pieces are a faucet aerator and adapter, as pictured above; the third is another adapter. I also used a metal washer or two, some o-rings, and a neoprene washer. These are to hold the mirror in place and prevent it from contacting metal, which might chip it.

Here are photos of the cell being glued into place, with both adapters on & the aerator body removed (for balance), and of the completed mount:


I have tested this mount, and as far as I’m concerned, it is a decent and usable design. As with the earlier versions, it can be built entirely from parts and materials that are available at hardware stores and hobby shops, though some of you may have to go to a bicycle store or some other odd place for small steel balls; and it can be built with nothing more than a jeweler’s saw, a file or two, a hobby knife, and some sandpaper, though I will admit that it is rather more convenient if you can drill a few holes fairly accurately. (You do need to drill at least one hole in order to glue the balls on the ends of the adjustment screws, but it should be possible to build a jig to do that from a few pieces of paper, which would eliminate all need for drilling. I may write that method up and post it here at some point, if I have time.)

Here, in case anyone wants it, is an “X” (cross-eye) mode pseudostereo (that is, I took the two photos separately) macrophoto:

...And now, on to the next item in this set.

A Simple Translation Stage

(23 August, 2007)

In connection with one of my projects, I find that I need some translation stages. (I need to position a pinhole for a spatial filter, and I need to position one or two iris diaphragms.) In order to fill these needs I have designed a translation stage that can be built for one or more axes, uses only materials that are available at hobby shops and hardware stores, and does not require any machine tools beyond a Dremel® or other rotary tool. You will, however, need a jeweler’s saw or something similar; a very small bandsaw might do.

Here is the basic structure I am making use of; I bought these at a local hobby shop:

In each case, what we have is a tube that fits very nicely over either a rod or another tube. With a square cross-section, rotation around the axis is not much of a problem: the fit is fairly decent. Still, if you need precision, the square configuration probably has more wiggle-room that you are going to be comfortable with.

In order to use the round cross-section you must resort to low cunning, and use them in pairs. The easiest way is to glue the two sleeves together; they have to be very nicely parallel, but that should be fairly easy to arrange. Another advantage of the circular cross-section is that the rods, especially in pairs, are considerably stiffer than hollow tubes. A third advantage of doing things in pairs is that you can push (or pull) the rods apart a little at their ends, which should eliminate most of the runout/wobble.


Sidebar: A Slightly Simpler Alternative

You do not have to use the concentric structures I’ve shown above; it is perfectly reasonable to use pieces of angle iron (or of aluminum angle extrusion) that ride on the rods. (I would, though, use stainless steel rods rather than any type of thin-wall tubing for this version.) You will, of course, have to make sure the riders are held onto the rods on any axes where gravity won’t hold them on for you. It is also probably a good idea to be sure that rods and riders are all as straight as possible.

Here is a rude ASCII-pic cross-section. The /\ structures with the little ^ “hats” are the angle pieces; the Os are the rods or tubes.

     ^  ^

(The hats should actually be integral with the /\ structures, but of course ASCII is not conducive to the accurate representation of detail.)

You could put a “floor” between the two riders to make it easy to mount things on them, just as you can with the concentric structure:

     ^     ^

Back to the main flow, and to actual construction...


My initial needs are not stringent, and for a first prototype I am going to make a 2-axis stage using the uppermost pair of square tubes from the photo, above. (The outer sleeve is 1/4" square and the inner is 7/32", which should give me reasonable strength and stiffness.) I would like to have perhaps half an inch of travel or a little less, so I am making the sleeves 2" long and the shafts 2.5" long.

It is rather important to ensure that the second axis is orthogonal to the first one. Fortunately, I have a small machinist’s square here. If you don’t have a square you can acquire one, or find something that has a reasonably accurate right angle. For most purposes we are not talking about microradians.

Here is a photo of the setup, with the second-axis shaft in position to be glued onto the first-axis sleeve:

Once I have the two axes connected, I need to be able to adjust their positions. Zimmerman’s (mentioned above, in the section about the mirror mount) carries M3-0.50 screws and nuts, which suit this application; 0.50 mm is just under 0.020 inch, and that is probably adequate precision for positioning an iris diaphragm. When I build the 3-axis stage for the pinhole of the spatial filter, however, I will need either micrometer heads, the differential screw arrangement that is further down this page, or both. (I think I have only two small mike heads, which leaves me using a differential screw for the third axis.)

To return to the prototype: rather than letting the first-axis screw push on the base of the second-axis shaft, which would reduce the usable travel of the second axis, I am thinking about gluing a small stop into place on the first-axis sleeve.

(Evening of 24 August, 2007)

I have checked to see where I need the iris on the laser, and I have epoxied the shaft of axis 2 onto the rider of axis 1, using two small washers as gussets. Here are two slightly blurry (sorry) photos of the gluing setup, showing the square I’m using as a reference


(The little spring clamp is holding the washers in place.)

Notice that the square is generating 2 axes of reference. The shaft is in contact with it all the way up, which is one axis; in addition, I centered the square on the bottom shaft as well as I could by eye, and then I centered the top of the vertical shaft against the square as well as I could. (Again, my needs here are not terribly stringent; if they were, I would have acquired at least one more square, and possibly two.)

I have also purchased some small springs, to use as tensioners; still thinking about how to attach those, how to mount the adjustment screws, and how to mount the stage on its base. I will have to put the base on a pedestal for this laser, but that’s a separate issue.

I need angle brackets to hold various pieces of this stage. Here is the glued assembly for the two axes (with riders on the shafts), as well as some possible angle brackets and sources:

As you can see, old computer parts turn out to be valuable in new ways. Here is the first foot being glued onto the stage:

Here is the control mechanism for the Y axis:

(I epoxied a plate with some holes in it on top of the shaft, and attached the relevant items. I put an L on the traveller, with a bar behind it for the springs to attach to. I also added a stop nut on the adjustment screw, because if I try to push the axis past its travel limit I will detach the plate from the end of the shaft, which is a Bozo No-No and is severely deprecated.)

Here is the completed stage. As you can see, I did the control for the X axis slightly differently; but the principle is the same. (I will probably add a stop nut on this axis to match the one on the Y axis, though because of the construction it isn’t really necessary.)

This device is now ready to go into the laser. If and when I get any results worth reporting, I will add a link to the relevant page.


Two further notions:

1) You can always acquire a single axis by disassembling a dead CD-ROM or CD/DVD player. The laser head rides on rails, at least in the players that I find inside computers. Here is a photo of an assembly that I took out of an ancient Mitsumi ROM reader; it still has the disk rotation motor on it, but that was easy to remove. The laser and other irrelevant parts of the head should also be easy to get rid of, if you don’t just decide to ignore them.

This has the advantage of giving you an actuator, though you may choose not to use it. You could fairly easily build a multi-axis stage from these if you don’t need it to be particularly compact. Alternatively, you could use one of them as the lowest level of a 2- or 3-axis stage.

2) A chap named Brian Manning has made his own version of the ruling engine that appeared in the Amateur Scientist column of Scientific American in April, 1975. His page about it is a gem.

A Simple Differential Screw

(25 May, 2007)

It turns out that there is a nifty way to make a differential screw — you butt two screws together (with a compliant coupling, unless you are abundantly certain that they are rather precisely collinear and well aligned with whatever you are using them to move), and use a fixed nut on one & a travelling nut on the other. You can think of the screws as being there primarily to connect the nuts together; the fixed nut is the reference, and the other is the driver. In practice, it’s not quite as simple as it sounds; but it is nonetheless a remarkably easy way to permit relatively fine adjustments. (We are not talking single wavelengths of light here. That takes slightly more advanced techniques, or really careful construction, and good temperature control.)

I desperately wish I’d made this up, because it is so ripping brilliant, but I didn’t; someone told me about it, perhaps a decade ago. Somebody almost certainly has (or had — probably long expired) a patent on it.

[I’m going to assume, for the purposes of this explanation, that the coarse screw rides in the fixed nut, so that the final motion is “forward” when you rotate the coarse screw clockwise. There is, however, absolutely no reason why you would have to do it that way if you didn’t want to.]

When you rotate the coarse screw, it moves in the reference nut. That pushes (or pulls) and rotates the fine screw, which thus pulls (or pushes) the travelling nut almost back to where it was before you rotated the screws. How much of a change in position you get depends on the ratio between the thread pitches. If you have, for example, M5-1.0 and 10-24, you’re looking at about 1.4 mm for 24 turns, which is quite decent. If you can make everything out of the same material, and if you can control the temperature in the room precisely, you may want to try M5-0.8 (fine pitch) and 10-32, which gives you 0.2 mm travel for 32 turns of the 10-32 screw. (It’s 0.8 mm/turn vs 0.79375 mm/turn.)

I don’t need anything like that kind of precision for my current application (positioning the ends of the electrodes of a TEA nitrogen laser), so I went with a much bigger difference. My coarser screw is 10-24 NC and the finer one is 10-32 NF. The 10-24 moves forward 1" if I rotate it through 24 turns, and the 10-32 pulls the travelling nut backward 24/32", which = 3/4", so the pushrod I’m using as an actuator ends up going forward by 1/4".

That is, 1 turn gives me a little over 10 mils, and one “hour” (if you think of the thing in terms of a clock) gets me just under 1 mil.

Here are two preliminary photos, with the electrode not yet attached to the pushrod:


You’ll notice that the fixed and travelling nuts are composed of two hexnuts each. This is a crude anti-backlash mechanism. To do it right, you actually have to make an extra doubled nut & run it up and down the screw a few times with some extremely fine abrasive, so it takes out all of the irregularities in the threads. Then you set that nut aside, remove ALL of the abrasive from the screw, and put on the real anti-backlash nut. I’m not going to bother, and I’m not going to tighten up the anti-backlash very much — I just want to take out a little of the slop.

Speaking of slop, this version is probably not really accurate for anything as tiny as 1 mil, because there is only RTV [in this case, aquarium caulk] between the nuts I’ve glued to the ends of the screws to provide sufficient area for good coupling; in order to get enough stiffness for real accuracy, I would have to make sure they were always in metal-to-metal contact. (Care to think up a nice simple way to achieve that and still have a compliant coupling that allows for [or prevents] all of the possible misalignments? Definitely a fun project.) OTOH, once the RTV stiffens up fully, it should certainly be good enough for folk music.

It is not trivial to figure out the details on one of these, and I have probably made some mistakes here, as this is really the first time I’ve tried to build one; but I’m sure that a bit of hands-on experience will get me where I want to go. I don’t really need even as much as 1/4" of travel for the laser I’m working on right now, which helps.

A Simple Hand Spectroscope

(24 August, 2007)

I noticed that The Surplus Shed has some inexpensive diffraction gratings, so I bought one and used it to build a little hand-held spectroscope out of Schedule 40 PVC plumbing pipe and fittings. Here are two photos, showing the way I constructed the ends:


I have angled the slit, so that one end is a bit wider than the other. I hope this will let me see some of the Fraunhofer absorption lines in the spectrum of the sun. Note, btw, the fact that the slit end consists of two threaded fittings. The one that the slit is mounted on screws onto the one that is glued to the main tube. Doing it this way lets me line up the slit with the grating easily. (It also lets me replace the slit with an adjustable one, should I ever decide to do so.) The two pieces of flat plastic under the razorblades are there for two reasons: first, to exclude extraneous light that would otherwise go around the back edges of the blades. Second, the blades are not actually wide enough to reach the edge of the fitting, so I had to support them. You can also see a small piece of wood at the edge, where there is a space between the two flat pieces. Again, it is there to exclude extraneous light.

The other end may be slightly unusual; the hand spectroscope I had when I was a kid was simpler — it had both ends cut off square, so its grating was parallel to its slit, and there was nothing past the grating at the eye end. I set the grating of this one at the angle that gave me the brightest spectra, and I added the extra tube to prevent other lights from interfering with my view of the grating. (Since this picture was taken I have changed the angle at the end slightly, so that if I hold the filter on the front of the camera up against the tube, the lens points more or less directly at the spectrum I’m trying to photograph.)

Here are a few spectra. First, a halogen bulb produces the expectable continuum, which may let you get a sense of the bands that the camera’s sensor is picking up. Then the discharge inside a HeNe laser. Third, a compact fluorescent bulb; and finally a spectrum of one of the HID headlights of my car:


(The middle two are somewhat underexposed; the left edge of the HID headlight one is grossly overexposed.)

The grating is definitely not the greatest I’ve ever seen, but it didn’t cost very much, and it clearly works. (Note, added on 15 September, 2007: I have purchased what I hope is a better grating on eBay, and if it works out well I will post a revision here.)

The new grating does, indeed, work pretty well, but it will require two photographs to cover the visible spectrum until and unless I shorten the sighting tube. This is because the grating, a rectangle that is about twice as long as it is wide, is ruled with the lines going the wrong way. At some point I may acquire a better one, but this one serves reasonably well for just looking at the spectra of reasonably bright sources.

Improving a Small Vacuum System

(09 January, 2011, ff)

I have been building some gas lasers, or trying to, and as part of that effort I have had some difficulty achieving a sufficiently good vacuum. I am fairly sure that the roughing pump I’m using is good enough, but I found that I couldn’t even get the laser head down to 1 Torr.

Part of this appears to be a materials issue: silicone rubber caulk appears to continue to outgas for quite some time after it is apparently fully cured, or perhaps it allows some air to diffuse through; either way, I think that it is not particularly suitable for use in vacuum systems, even crude ones.

There is a special epoxy called Torr-Seal, which is good down to about 1.0X10-9 Torr; it is rather expensive, but it does work. If you look on the Web, you will find claims that Loctite’s Hysol 1C (which is far less expensive) is very similar, but I don’t see any actual performance test results. My current conjecture is that just about any properly-mixed epoxy should be good down to better than 1 micron (10-3 Torr), and I am guessing that J-B Weld, which contains fillers of some sort, may be slightly better than most.

Then there is the question of teflon pipe tape. How good is it? It certainly appears to be adequate in the low-milliTorr range, which is good enough for me, at least for now. I am, though, curious about whether it can be used at lower pressures.

In any case, here is a page on my rebuild of the manifold for this setup.

Back to the Index

To the Joss Research Institute Website

To my [updated] mirror

To my current research homepage


This work was supported by
the Joss Research Institute

Contact Information:

My email address is, 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: Wed May 10 15:03:28 EDT 2017