The special 30th Anniversary Issue of Gear Technology magazine represents a lot of extra work on the part of the staff. I enjoyed contributing to it and thought I’d add a favorite story to the same issue’s “30 Years of Calculation” article.
As KISSsoft AG’s Dr. Stefan Beermann notes, companies used to write their own software and run it on huge mainframe computers. I am sure the total computing power of these monsters is laughable by today’s standards, but in the 1970s it was the weapon of choice in gear design. At Falk (where loyal readers of this Blog know I was happily employed for a number of years) there was a team of really smart people charged with developing and improving the gear rating program.
Unfortunately, a lesser team (I should know, I was on it) was assigned to use the program. We were not very well informed on what was going on inside the black box, and so we merely concentrated on getting our cards punched accurately. That’s right — our group of two-fingered hunt-and-peckers was expected to type its own punch cards with the program input screens, keep them in proper order, and submit them at the end of every day for overnight computation.
I had four sizes of gearbox to design, and maybe 30 total ratios-per-size, with four or five catalog speeds. It was a lot of problem sets.
My work area was a desk and drawing board amidst stacks of punch card boxes. Just one incorrectly punched or out-of-place card got you a bunch of error messages — and no data to analyze.
We spent a lot of time checking our cards and coming up with schemes to keep them in order —schemes that were easily thwarted by the courier dropping the box of cards, or by “hanging chads (see 2000 U.S. Presidential Election recount)” interfering with data reading. Plus, we had target ratings from marketing that were unrealistic. On my four units I had probably twenty or thirty targets that I wasn’t sure I could ever eliminate.
Then one day — out of the blue — I had a problem rating go away. I couldn’t remember making any changes to that set and double-checked the punch card. Sure enough — I had transposed a digit on the pinion outside diameter and got a big strength increase.
Off to the “family recipe book” I went, trying to determine why a thirty-thousandth-of-an-inch-increase in the outside diameter would result in such a rating bump. It was the first I had ever heard of long and short addendum gear geometry. I therefore decided to conduct a few experiments on my other problem ratios — without authorization and right under the collective noses of my completely unaware superiors.
When, a few weeks later, the team presented its interim results to the bosses, I was the only guy without “deficient rating points.” Naturally a full-scale inquisition ensued and my “secret” was dissected by those really smart people in programming. After slapping my hands for messing with the old family recipe, they admitted the deviant gears would work properly and that the improved ratings were legit. A few transposed digits inadvertently opened the door to strength balancing on carburized, hardened, and ground gears.
Indeed, it is often better to be lucky than smart. But do you know what’s best?
Being lucky and smart.