GenRad specialized in both light and sound. I’ve been featuring sound for a long time, now let’s do light.
Polistra and friends are processing food, using a GenRad color comparator to check the quality of tomatoes.
The Comparator was simple to use. You’d place the object on top of the viewport, then turn the filter switch at upper left to Red, Green, Blue, and White. You’d check the meter reading for each setting, and compare it with your standard and tolerance for the tested item. We still use the same four variables in computer graphics as RGBI.
It’s easy to see how this process could have been speeded up or automated. Sample all four colors at once, record the four readings at once, sound a buzzer if the tested item is outside of the tolerance band, or automatically kick it off the conveyor into the recycle bin.
The four pots on the right are solely for calibration, always a primary task for GenRad. Conditions would inevitably differ between sessions. The ambient light could change, the AC mains could have a different voltage, and the photocell and bulb would change with age.
= = = = =
The innards of the comparator were typical GenRad, emphasizing Calibration above all, and using some neat mechanical gadgetry. Showing the innards schematically from above. The bulb sends light upward through the filter onto the sample. The filter switch moves the filter sideways to place four different colors in the light beam. The light reflects from the tested item toward a photocell.
Tech note: The reflection obviously shouldn’t pass through the filter again. GenRad’s diagram didn’t show how this was done, so I’ve imagined a donut-shaped group of photocells letting the filtered source beam pass through on the upward trip, and sensing only the raw unfiltered reflection from the item on the downward trip.
For calibration, each pot lets a different amount of current from the photocell into the meter. The filter switch chooses one of the pots at the same time when it chooses one of the color transparencies. To calibrate before a measurement session, you’d place a standard white object on the viewport. GenRad suggested a piece of magnesium. You’d set each of the pots to max out the meter when the switch was including this pot. A white object would then be baselined as 100% on all four colors, even if its actual measured intensity isn’t 100%.
After setting the pots for the white calibrating item, you’d start checking the real testable products. During this process you wouldn’t touch the pots, just switch to the four colors and read the meter.
Reference footnote: Genrad probably had a manual for this, but it’s not available
at IETlabs, so I’ve used a photo from the 1936 catalog, plus a description in an article at Radio News, Aug 1943.
Sidenotes: I’ve screwed around with color before. The Maico Chromalyzer was a speech analyzer with color, and the Clavecin Oculaire was a 1750s color display for a harpsichord.
Polistra's Mill 