ShopInfoLinks 
Comic Info:
September 16, 2014 - 12:48am
My company does a fair bit of R and D and it's really fun. It's amazing how much work can go into answering a simple question... and sometimes the best answer you can come up with is "maybe...?"
Topic:
Tags:
Support this comic on Patreon!
Comic Comments
-
#795harleyrundle1977905 (not verified) -
#794Madasamadthing -
#793Jam -
#793SwiftAusterity -
#792Ronald Riehn
Comments
Sorry, was the "definitely less than eight" intentional?
Unless we're talking fractional tubes (whoah!), "seven" would do. Unless, unless he meant "definitely no more than eight"?
(What, you thought I was going to complain about "less" vs "fewer"?)
But it might be six!
Worst case might be six? Why bring in eight, then?
Adds a 1.3 Factor of Safety. Depends on how critical the function of the device is. If it's very critical, I'd go with 9 or 10 tubes (3x3 array or 2x5), since that bumps things up a bit more. Less critical, 7 (packs nicely in a hex-pack) since you're not risking as much by pushing down the FS.
If it breaking means people die, more like 20 or so. Or you can go with the Brooklyn Bridge's factor of safety and use 36 tubes, if you're designing it for use by the public for a hundred plus years.
The answer was apparently so interesting, or took so long, that literally half the people changed shirt colors.
Funny; reminded me of this story, where Richard Feynman, working on the Connection Machine, determined the optimal number of buffers to use in routing would be 5, but using differential equations. His analysis was distrusted and they attempted to go with 7 as suggested by a more conventional (for computer architecture), discrete analysis, but the chips with 7 buffers turned out to be too big to manufacture efficiently (for reasons presumably unrelated to either analysis), so they thought they'd rely on Feynman's work and reduced the number of buffers to 5, and the computer working with the chips such manufactured turned out to work OK.
In honor of this event, the US government even created a new agency, the ATF - Always Trust Feynman.
The organization you normally think of as the ATF is actually the BATFE. They just shorten the acronym because the US believes in the Rule of Three, as noted by its obsession with 3 letter acronyms for agencies. (CIA, NSA, FBI, ATF,...)
(In case text strips the humor from this, the only true statement here is that the proper acronym for the ATF should be the BATFE - Bureau of Alcohol, Tobacco, Firearms and Explosives....or for something more pronouncable, the Boat-fae)
Given time, engineers can do quantified qualification really well. If you don't have time to do it right, give the favorite SWAG of a friend of mine, "X, plus or minus 2 dB." (Caveat emptor, only attempt this with quantities that don't normally scale exponentially.)