I’ve now completed construction of the Gamma prototype. Thanks to Beto Peliks for his help with the MIT water-jet machine, which cut out the sheet metal parts: the covers, bumpers and film gate components. The heatsinks were cut on a CNC Bridgeport mill and the cylindrical parts on a 10” South Bend lathe. The sheet metal parts were bent on an arbor press using a 1” wide 90º punch and die. Threaded inserts for M4 screws were installed in the heatsinks with an M4 capscrew mounted in a “hand tapper” to keep the insert aligned with its hole. (A hand tapper looks like a small drill press with a hand crank instead of a motor.)
Project engineer Allen Armstrong (L), product manager Peter Fichter, and first light with the Kinkajou gamma prototype.
Gamma prototype inverted with bottom plate and film reels removed.
When power was first applied to the LED, illumination level varied unacceptably across the screen image, being bright in the center, and quite dim at the sides. This was traced to a too-large distance from the LED to first condenser lens surface caused by a combination of three effects:
1. The LED was not retained by the printed circuit board, as that’s not available yet, but instead by nylon 4-40 screws, whose heads caused the mock PCB to be too far from the LED. The condenser lens mounts to the PCB, so it was moved also, about 1.5mm..
2. The condenser lens housing was mistakenly machined 1.1mm too long, also moving the first lens surface away, because the dimensions were taken from the surface that was too long.
3. In designing the optics, we were vague in communicating whether the LED-to-first surface distance was measured from the chip or from the surface of the lens included in the LED package.
These errors cause the condenser lens to focus the source image near the film plane rather than in the projector lens, and the image is smaller than the microfilm, so that the edges of the film receive only stray light.
Moving the lens closer has largely solved the observed problems. The light level is now quite uniform across the image, and quite bright. The image is 4 feet wide at 7.5 feet from the projector. There is some “halo” around the text letters, and focus deteriorates near the left edge. Still readable, but not in focus. Optikos have suggested a slight curvature of the film plane, and this will be incorporated in a revised film gate design, necessitated by the optimum focus point found to be 9.8 mm from the first projector lens surface, rather than the predicted 15.2mm.
The film transport system works as hoped. The one-way roller clutches incorporated in each of the two knob shafts prevent the film being “pushed” backwards. The knobs can only pull on the film, action is smooth and the drag springs hold the film from unwinding without causing excessive turning friction.
The focus knob permits fine control of the focus. My first impression is that it will be needed in the final product, because sharp focus really improves the screen image.
I noticed that the square holes in the centers of the two microfilm reels differed in size by 0.3mm, which is enough to cause the loose reel not to be well located. We’ll want to ensure that we get all our reels from one supplier, and size the drive squares to fit.
The prototype will be turned over to Tim McNerney tomorrow, along with the mechanical bits he needs to do his testing of the PCB.Posted by Allen Armstrong at May 22, 2004 10:37 PM