Mechanical Test Card Player

The inspiration for this project came from an article about a device Jeremy Jago built years ago. It used an LP record with tiny holes manually drilled around the edge to create a black or white 32-line video signal viewable on a standard monitor. My idea to improve on the device was to replace the drilled LP with a computer-generated pattern printed on copy paper.

I initially planned to model my version of the machine after a floppy drive, where a sensor on a slide attached to a stepper motor reads the disk as it spins. Different tracks are selected by moving the sensor radially along the disk surface. Now, if you were to replace the magnetic head with an optical emitter/detector pair and the magnetic disk with a circle of dots printed on paper, then you would have a simple test card playback device. If only there were some kind of optical version of a floppy drive to build this around. Oh wait, there is!

The core of my device is built around the guts of a computer's CD-ROM drive. The drive already contains a linear tracking sled and a spindle, so all I had to do was remove the original optical components and grind off unneeded plastic until I had a flat surface on which to mount my 3D printed replacement parts. The original optics were not suitable for this project. The spindle in these drives also has a very clever magnetic cap that holds the disc in place even at high speeds, which I kept.

I first tried using my 3D printer to create a miniature optical system consisting of a red LED shining through a slit in a piece of cardstock and a small lens (ironically from another Compact Disc player) to create an oblong spot. However, I was unable to get the projected spot clean enough, no doubt due to my limited knowledge of optics. A circular halo from internal reflections developed in a similar way to the laser modules, and the depth of field of the system was very shallow, requiring an extremely flat disc. I did by chance eventually come across an ordinary laser module that works, although it stuck out of the top of the device rather awkwardly. A quick redesign of the housing and the addition of a 45-degree first surface mirror removed the unnecessary appendage and lengthened the optical path so the threaded lens remained further in the module and was thus, more stable. A rubber O ring secures everything in place. The laser module also eliminated the depth of field issue. In fact, this laser can be focused to a very small spot indeed, approaching the size of the 600dpi toner dots used to create shades of grey on a laser printer. To pick up the reflected light, a generic photodiode in a 5mm LED package is mounted below the sled just above the disc, pointed toward the laser spot.

The test card player connected to a monitor

A red laser module is used to read the pattern from the paper disc, though I really wanted to avoid using a laser for two reasons. First, a laser is not strictly required for this to work, only something that can produce a small enough spot of light to reasonably resolve detail. Second, I initially had a hard time finding a laser pointer or module that could focus to a point no more than a few inches in front of itself. These are usually designed to throw a small spot out to infinity. In every case the threaded lens in the module had to be brought so far out of the housing that either it ran out of threads and fell out, or internal reflections developed that marred the quality of the spot, or both.

Mechanical Test Card Player without disc

Test card player with pattern disc removed

I could have made the system transmissive rather than reflective, and in some ways that may have been easier, but it would mean abandoning the CD drive sled and constructing my own device with arms on both sides of the disc. The sled was much more precise than something I could easily build. I also had some concerns about visual noise caused by the paper fibers in the lighter parts of the image when using a transmissive design (if you've ever held up a piece of paper to light, you'll see what I mean) so I chose to go reflective.

The Pattern

NBTV Forum user smeezekitty (author of FreeNBTV) kindly created some command line software for me that takes a picture, converts it to 32 lines, then forms it into a ring, creating one complete picture per ring. By stacking and resizing patterns in concentric circles using the image manipulation software GIMP, I had something that looks like a strobe disc for a turntable. An early proof of concept using this pattern, a blue laser and a photo panel from my NBTV Mixamp produced promising results.

Originally, I envisioned using a disc the same size as a CD or DVD, and using one of those half-thickness protection discs that comes on a stack of burnable media as support for the paper. Ultimately, I chose eight inches as the disc size which is close to the maximum width of ordinary copy paper to provide the highest resolution and make centering the disc less cumbersome. At 750 rpm, the paper is not flattened enough by the centrifugal force, so I cut an approximately eight-inch diameter, 1.2mm thick circle of acrylic (about the same thickness as a DVD) for it to rest on. Acrylic was chosen not for its transparency, but for its stiffness and resistance to drooping.

The disc is driven by a small DC motor via a belt, not unlike a standard NBTV monitor. There is really no need for precise speed control so long as it remains stable over time, so the motor is driven by a variable current drive derived from the Club's motor control circuit, but without the 4046 IC. In fact, my prototype used the Club's motor control circuit board with the 4046 IC removed. This can be used to produce any frame rate you want, up to the limit of the motor's speed. A strobe disc pattern is included to assist in finding the right speed, though its usefulness varies with modern LED lighting.

The original plan for generating sync pulses was to have a dedicated sync pattern printed on the innermost track of the disc, picked up reflectively by a separate emitter/detector pair. The trouble here is that I wanted to keep the device as electronically simple as possible to emphasize the "mechanical" design. I planned on using Peter Smith's Universal Video Clamp and Sync Inserter (Vol. 35 No.1) which uses an infrared LED shining through holes in a disc to create the sync pulses. When a photodiode detects the light, it drops the video signal to zero volts, creating the pulse. This means that the width and angular location of the holes are important, unlike a monitor which can cope with a simple black/white alternating stripe and a reflective pickup. I didn't want to use a second laser just for sync, and unfortunately the black toner is surprisingly transparent to light shining through the paper, so the next best thing was to physically cut the holes in the form of slots directly into a disc made of oiled stencil board. This also allows the location of the sync pulse to be adjustable by offsetting the paper disc slightly relative to the support disc, in case you want that sort of control for some reason.

The pattern disc (click for larger)

A problem appeared though when I calculated the width of these sync slots. In this location, the width of the slots would be almost as thin as the stencil board itself, so with a little reconfiguration, the sync assembly was relocated to the outermost diameter of the disc. Even there the slots are only 1mm wide. This could have been avoided by using more reasonably sized sync holes to trigger a 555 monostable which would send a proper width pulse to the Sync Inserter, at the expense of additional circuitry.

Sync emitter/detector pair

The downside to moving the sync location is that the image pattern is about 5/8" smaller, and there's now a potential for crosstalk between the sync LED and the image photodiode, especially on the first few tracks. In practice though, crosstalk is not an issue. Shrinking the pattern slightly is no problem with the laser, and using a green visible light LED also sidesteps any issues of IR transparency of the stencil board and 3D printed parts. Placing the LED below the semi-opaque disc pointing upwards also provides a convenient (though unintentional) indicator light when power is applied.

To change out the disc, the original CD-sized disc required the laser sled to be swung up out of the way to access the magnetic hub. Conventional drawer hinges were too loose to be useful, as any movement of the sled while changing tracks would disrupt the delicate timing. Switching to the larger eight-inch disc brought the disc hub out beyond the edge of the sled support, making the lifting mechanism unnecessary.

The machine has a small panel on the side for the power input, composite NBTV video out, a switch to disable the motor, and speed control. There is no power switch. At the time I was having the sync disc cut, my friend with the laser cutter was experimenting with cutting mirrored acrylic. He made a variation of the control panel (that was to originally be made of wood) out of acrylic, and it came out so well I decided to keep it.

Control panel

Results

The images, when viewed on a monitor, are surprisingly crisp. I was impressed with the quality this device produces. The only real visual difference from a test digital waveform played back from an iPod are some slight geometric distortions and a little bit of hunting in the image, likely caused by the lightweight disc and motor. Grayscale values are decent. One other defect that occasionally appears is a vertically oriented bright patch on one side of the frame, caused by slight warping of the paper. This is exacerbated by humidity, even with the back side of the disc laminated to help keep warping under control. An exaggerated version of this can be seen in the last screenshot (at the bottom of this page) where I moved the laser to a blank part of the paper.

For course track selection the laser is simply grabbed and moved with your fingers. For finer tracking adjustments you run your thumb along the worm gear. While slowly changing tracks, the images will crossfade between each other. If the disc is not perfectly centered, the images will "soft wipe", meaning a wiping transition with a very soft boundary will occur from one side of the image to the other, depending on which part of the pattern is the most off center. Sometimes the wipe will begin near the center and progress outwards. It is also possible to display parts of two test cards at once, since I did not put spaces between the tracks.

Probably the most notable benefit of this device over traditional digital playback is the ability to switch test cards without loss of sync, and to set the frame rate to any reasonable number, even on the fly, as long as the adjustment is made slow enough.

Warped paper due to high humidity

Off-monitor screen shots

If I were to try this project again, I would likely change a few things. Jeremy suggested that I could make the project even simpler by eliminating the sync inserter altogether and reading the composite waveform directly from the paper, as he did in a later version of his device. Pure black would represent the sync pulse and dark grey would become black, at the expense of reduced dynamic range in the image. Having the sync included in the waveform would also eliminate the fiddly process of properly lining up the sync disc.

NBTVA member Peter Spies also had the excellent idea of attaching the pattern disc directly to a Nipkow disc monitor, thereby integrating the test card generator directly into the monitor itself. No sync would be necessary. Another variation of the idea would be to add two additional sets of tracks and pickups to create a color system, or a larger disc/smaller pattern could allow several frames per revolution to create short, artistic animations. Like mechanical TV in general, variations on the core idea are almost endless.