[Subscribe to Daily Digest] |
I suppose the situation in the US is a little more extreme then... the levels of college debts, the incomes of profs etc...
True, IBM + Bell labs did significant fundamental research, at least on the engineering side (much signal processing work came out of Bell labs!)... Up here, Nortel and others fed money back to university research programs (great way to train new employees)... one of the cooler private investments was from RIM founder Mike Lazaridis, starting an institute to fund theoretical physics out of his own pocket... so I'm not saying it doesn't happen, but most private funding is targeted.
wrt "design by test", I mean nominal design that effectively leaves things to figure out in "the lab" that don't need to be... I'll give a simple "hypothetical" example: A mech eng designs a box with a recessed lid. Orders a sample or 2. Lids fit. Great. Supply management orders a few thousand from multiple vendors for production. Many lids don't fit even though they meet spec (so can't RMA). Scrap. why? different batch related tolerance stackups starting at the centerline of the flat head screws wrt thru holes. The original sample testing only proved the lids fit in that 1 case (in fact, the sample was a complete waste of time + money imho, yeah, it's a box, so what? what does the "test" prove?)... tolerance stackup was improperly calculated from the get go (or more often, not calculated at all due to laziness: nominal design). If you do it right, you can order a million boxes + lids from different suppliers + never have a problem. They will always fit or they get RMA'd. No need to test (though it will be "tested" on the line anyway, so better get it right!)... of course the same applies in all disciplines of engineering, even the more sophisticated stuff. Engineering is the application of known science to make new stuff... Things that are not (efficiently) calculable often benefit from lab work, but the engineer doing so had better be damn careful they don't fall into the design by test traps! Anything that can be calculated should be + any experimentation should be very cautious wrt the impact of potential uncertainties. When process tolerances are unknown, it is critical to figure them out (and hopefully thereby find ways to tighten them) vs ignoring them + design by test...
(yeah, that one's a pet peeve... sorry for the rant! my point was that industry is far from being immune from apathy...)
James...
posted by 216.59.242...
, Wed, 13 Jul 2011 11:20:06
, Tue, 12 Jul 2011 09:33:51
, Sat, 9 Jul 2011 13:06:57
, Tue, 12 Jul 2011 10:40:16, Tue, 12 Jul 2011 11:27:51, Fri, 8 Jul 2011 14:20:12
, Fri, 8 Jul 2011 09:26:41
No Site Registration is Required to Post - Site Membership is optional (Member Features List), but helps to keep the site online
for all Saabers. If the site helps you, please consider helping the site by becoming a member.
 |
 |
 |
 |
 |
