A start-up company I worked for in the early 80s invented an innovative technique to "burn" images on 35mm slide film, one pixel at a time.
This device consisted of a custom CRT (Cathode Ray Tube) that produced a very small spot of constant brightness, and optics to image that spot onto film. The electronics positioned the spot in x and y on the CRT face, turned on the spot to expose one pixel on the film for a precise amount of time, turned off the spot and moved to the next pixel. This was done for 4000 spots horizontally, and 2000 rows.
The image was built up through a red color filter and then repeated through blue and green color filters, to produce an 8-megabit image with 24-bit color resolution. This quality level is about where consumer digital cameras are now and is high enough to be above the resolution of 35 mm color film. And this was at a time when computer monitors were just beginning to have color at all.
We built an early prototype to test the concepts. It filled a rolling bench allowing us to roll it into a dark room for imaging, since the imaging path wasn’t light tight.
One side comment about an exciting event during this stage of the development: At one point I managed to turn on the CRT beam while the x-y deflection wasn’t working. Before I caught my mistake, the beam had burned through the phosphor on the tube face and fully half-way through the glass. It was a close call. I don’t know for sure what would have happened if the glass had been penetrated, but sudden releasing of the vacuum in a CRT has been known to cause it to explode. Not to mention the fireworks that would have resulted from the high voltage then applied to the CRT with the CRT full of air.
Back to the story: We hooked the prototype up to a PC, ran some test code, and it took 3-4 hours to build up an image. When we developed the film, the results weren’t great, but they were good enough to indicate we were on the right track. The management was pleased, forgot about Murphy’s law, and, based on our early success, began making commitments to display the product at an upcoming trade show.
When we took the next step and sped up the deflection circuitry, the images were terrible. They were full of diagonal light and dark stripes. While we tried to blame the quality on the film technician who had processed the film, we couldn’t get away with it. The problem was ours.
We implemented some changes in the electronics and tried again. No good. Fixing this problem was going to be a slow process. There was no image to look at on the CRT, just a slowly moving spot. The only way to look at the image was to expose color film, and then spend an hour or so developing it. It was a long slow, frustrating process to try everything we could think of:
Was the high voltage power supply modulating the spot brightness? Turn off its modulation and expose a frame. Develop the film. Cuss. Same stripes of light and dark.
Was the low voltage power supply modulating the analog deflection? Create an experiment and expose the film. Cuss. Same stripes of light and dark.
Was the PC somehow interfering with the unit? Build a very long cable to the PC, move the PC out of the room, expose more film. Cuss. Same stripes of light and dark.
We became fairly sure that the patterns were beats against 60Hz interference, because when we changed the deflection rate, the width and orientation of the stripes changed. But we could not find an electrical path for the interference.
Of course, all during this process, the trade show was drawing nearer—as was the management looking over our shoulders.
Eventually, and in desperation, we moved the whole table out into the main lab closed all the shades, waited until after dark, turned out the lights and made some images: Almost no pattern of stripes.
When we moved the bench back to its original position, the patterns returned. We went home, late, tired, and confused.
In the morning, someone suggested magnetic interference with the CRT beam itself. We bought some mu metal, which is used for magnetic shielding. We formed a cone of the mu metal and inserted the CRT totally into the cone. When we made an image with the bench in the original location, it was near perfect. It tuned out that the CRT was so sensitive that its beam was being deflected by the line voltage wiring running in the wall next to the test bench.
We managed to get the first full unit finished and making images by working all night before the trade show. It had two levels of magnetic protection, a steel housing, and a mu metal cone fitted to the CRT. It weighed 50 pounds, and it made great images.
I personally delivered it across country by air, set it up in the show booth, and went to the hotel to sleep.
Afternote: The product was a technical success. We halved the cost of the competitive product, and extended its time between calibrations from a month to multiple years. However, the market wasn’t there, even at the lower price point, and the company was soon bought by a larger company, and our lab was closed.
Charles Glorioso has a BSEE from Purdue and an MSEE from Illinois Institute of Technology. He has over 40 years experience in electronics design and management for industrial and consumer products.