Mirror Drums | televisionmachine

My Experience With Mirror Drums

A version of this article also appears in the NBTV newsletter Vol. 43 No. 4

Over the last few years I have attempted to make several mirror drums. All were intended to go into working machines but, so far, none have made it to completion. This page documents the design progression. They all follow the general process outlined by Dave Gentle in his article describing his homemade mirror drum in the NBTVA Newsletter, Vol. 34 No. 3.

Attempt 1: This is the most simplistic of my attempts, done with tin snips and an old paper cutter. It's more or less an exact copy of Dave's technique. The center support disc (the aluminum circle that the mirror clips attach to) is made from a single layer of .020" aluminum sheet, so needless to say it's more than a little wobbly. To make the support disc, a paper guide pattern marking the location of both the mounting and alignment holes (the outer and inner rings, respectively) was generated on a computer, printed, stuck to the disc with adhesive spray, then drilled with an ordinary drill bit. Pre-cut 1" x 1/2" back-silvered glass craft mirrors were purchased from eBay. They were secured to the aluminum clips with 1/2" x 1/2" Scotch Permanent Mounting Squares. It was purposely designed to be lightweight so it could be direct-driven from a VCR transport motor. Due to its flimsiness, though, it's more useful as a visual example than a usable device.

Mirror drum V2 with hand-cut clips.

Attempt 2: My next attempt fixed the problem of flimsiness by replacing the hand-cut .020" support disc with a much sturdier 1/8" aluminum plate. The rest of the design elements remained the same, except, due to the thickness of the plate, the mirror clips were positioned on the same side of the disc instead of one on each side. This one was intended to mount to a VCR head drum turned down into a hub on a lathe and belt-driven by an external motor. I also mistakenly ended up leaving too much space between the mirrors.

Attempt 3: By this point I had access to a waterjet cutter capable of cutting thin aluminum sheets, so I moved the design process fully into the digital realm. This drum was also going to be a 30-line Baird drum, rather than 32-lines like the previous ones. By arraying circles around a center point, an extremely accurate support disc could be made. One problem I ran into was that the waterjet cutter was not set up to use any type of abrasive for cutting, and therefore could only barely handle blasting through even a single .020" sheet. To get around this, I designed several patterns that would be bolted together as layers, effectively increasing the thickness of the support disc. Since the disc was now being machine made, the ability to manually adjust the radial position of the mirrors was (in theory) no longer necessary. One way I took advantage of this shortcut was by creating a cutout of the exact shape of the clip in the center pair of the four total laminations. By "subtracting" the shape of the clip from the inner laminations, a perfect cutout was made to hold the clips firmly in place. Therefore, the mirror clips could once again be mounted directly against each other, without leaving a space in between them equal to the thickness of the support disc.

Shortly after making that disc I realized that there was no reason for the separate mirror clips to be cut at all and they could now be integrated directly into the disc's inner laminations. However, before I got around to redesigning the disc again, I discovered that small variations in both the adhesive squares and the bent clips prevented the mirrors from being perfectly aligned radially (the vertical direction on a vertically scanned image) as I had hoped. With no way to adjust them, I had to abandon the plan to mount them using the cutouts and put them on the outside of the disc for fine adjustments as in the previous versions.

Outer disc laminations (left), inner laminations (right), and mirror clips with razor blade for scale.

Only one problem remained; that of precisely tilting the mirrors to deflect the light spot in the horizontal direction. Dave Gentle relied on the flexibility of the aluminum clips to hold each mirror in a pre-bent position, but for long term stability and precise alignment, a more permanent method was called for. Originally, I planned to slice off a very short length of angled aluminum, drill a hole in each flat face, and thread a set screw through one of the holes. The other hole would secure it to the support disc. The set screw would push against the mirror clip, bending and holding the mirror at the proper angle. Instead, I decided to purchase my first 3D printer and custom design a plastic version which worked perfectly and was much easier to produce. The bottom of the piece has a cutout that exactly matches the shape and depth of the mirror clips so the adjuster won't wobble around when the support bolt is tightened. The hole on the mirror side is just big enough to force a set screw through, and screwing it in and out several times cut threads into the soft plastic. I did have to use a very tiny amount of oil to make it easier to turn and prevent "binding" -where the screw would move in sudden lurches and only with considerable force.

Mirror drum V3 with adjusters. The grey adjuster indicates line 15.

Mirror drum on it's stand. The LED stuck on a piece of scrap aluminum on the lower right shines through a slide projector lens, then onto a mirror which reflects it onto the drum. The metal pins are for locking the drum for alignment.

Detail of the adjusters.

The most difficult part of this project is undoubtedly the alignment procedure. This requires a very solid mount clamped to the table and a way to lock the drum every 12 degrees with practically no play. The fact that it acts as a projection device means that every error is amplified on the screen, so absolute perfection is hard. I built a heavy duty test stand out of wood and chose to mount the drum directly to the shaft of a blower motor of the type used in a car's climate control system. It undoubtedly had the torque and solidity to get the drum up to speed without instability. The light source is a 3W white LED that produces an appropriately sized square spot when coupled with a slide projector lens and another small lens to optically shrink the size of the die. Since the LED die itself is square, no aperture or hole for it to shine through is needed.

I also created a 30-line alignment pattern of 3.5" x 8" (close to the screen size of a Baird-Bush mirror drum receiver) and found it helpful to mark each line number on the pattern and the disc to keep track of which mirror I was adjusting. Two metal shafts pushed through the sync holes into a block behind the drum locked everything in place during adjustment.

After alignment and fine tuning, I was very pleased to get a reasonable image from a test video. It was quite a thrill to see what is normally reserved to a tiny square on a disc monitor thrown onto a white paper screen. The motor was fed from a bench top power supply and manually controlled. However, something within the motor would cause the speed to shift suddenly every few seconds, even with the inertia of the drum, so I am doubting that this particular motor will be suitable if pressed into service, but further tests are needed.

30-line alignment pattern with numbered lines.

Soon after that the trouble started. I noticed that a few mirrors were starting to wander in the horizontal direction. Not by a little, but by five to ten lines! Adjustment of the set screw brought them back into position, but after a few days another mirror would do the same thing. I eventually discovered that the plastic mirror adjusters were developing hairline cracks at the elbow, and starting to bend backwards! This is surprising, because the force on the adjusters is not large, and any centrifugal force should pull the mirrors away from the adjusters, so perhaps when I printed them I left too much empty space on the inside of the part. (Called infill, instead of making a solid part, material, weight and print time can be saved by placing a honeycomb-like structure inside the part at the expense of mechanical strength. The size and thickness of this honeycomb determines the amount of empty space that is left.) This is how the project sits at the time of this writing, until I get around to redesigning and printing a better adjuster. But, the results so far are very promising. There are quite a few ways to make a mirror drum and there are other options I am looking into as well, including direct 3D printing of the entire drum, but one thing these projects have shown me is that mirror drums are finicky devices!