specs that end up butchered by reality?

[coppercone2]

What are some examples of specs that are butchered by reality?

I noticed that with heat sinks.. optimal ones are pretty damn weak to impact (like on chassis exterior). (500 simulations later you realize its like a dandelion)

Did you ever spend a long time trying to max some kind of equation out then you found the implementation is ridiclous, fragile, etc? and then that the solution is basically a difficult to explain hack job? (why did this take so much time??)

[SiliconWizard]

It's better than reality ending up butchered by specs...

[Mr. Scram]

It's better than reality ending up butchered by specs...

Who says it isn't?

[coppercone2]

god

[SiliconWizard]

Not sure who that is.

Anyway, I guess examples of specs that are butchered by reality: many Li-ion batteries of dubious origin.

An example of reality butchered by specs? Maybe the Boeing 737 MAX's MCAS?

[coppercone2]

no no, I meant something along the lines of where you do alot of optimizations to your design and it turns out for some weird reason its all useless and that thing you thought of when the project first began as being too simple is the correct answer (after you found some design options)

not necessarily negligent or complicated, just odd

[T3sl4co1l]

Did you ever spend a long time trying to max some kind of equation out then you found the implementation is ridiclous, fragile, etc? and then that the solution is basically a difficult to explain hack job? (why did this take so much time??)

It is a poor craftsman that blames his tools.

Or a good anecdote from the great Richard P. Feynman: he was with some colleagues discussing computer science research.  The subject was a learning and optimizing system, which was programmed to find any way necessary to achieve its end goal.  Well, the first thing it found was to simply set the end goal memory address to the goal value.  Okay, well, that's not helpful.  So they added conditionals prohibiting direct "hacks" like that.  While the CS guys found it a bit annoying, Feynman found it amusing; they literally told it to find any solution, and it dutifully did!

To explain better:
So you're optimizing a problem with respect to a certain number of variables, but disappointed when the solution apparently unexpectedly ends up suboptimal in other, previously undefined variables?  So you really weren't solving the problem you thought you were; you should've specified it properly in the first place!

Writing a comprehensive spec, of course, is half the challenge of engineering.  Failing to do so, is not much different from any other bug or omission in ones' work, just part of the job.

Tim

[coppercone2]

not tools, im not saying its not manufacturer-able, just later you look at it and some other detail comes to mind and you are laughing because the 'looks right' solution was the best one after all\

the idea that its something random that comes up. which makes it funny.

[Berni]

I like to call that "over-optimizing"

In quite a lot of cases when you really really optimize a design for one specific spec you will be in trouble. What you end up with tends to be one of more of the flowing:
1) Ridiculously unpractical to build: Oh so maybe using 100 opamps in parallel might be a bit expensive... but the noise is nice.
2) Needing a very strict use case to work: Look it works great... tho if my warm breath touches it the output drifts off towards the supply rail faster than you can say oh cap.
3) Sucks at other specs: Look my amplifier can output 2kW with that tiny heatsink... but the THD-N is 60%

Its not that these "over optimized" designs are bad designs. But by the time you reach them the applications for the design become so niche that you might not be able to justify it. But a good design is a design that hits ALL of the relevant specs simultaneously (Even if the said specs might not be so impressive on there own). That's what engineering is about.

[SiliconWizard]

Yeah, you can call that "over-engineering" as well.
But if your over-engineering eventually fails to survive reality, then as T3sl4co1l said, it was probably ill-specified and/or unproperly modeled.
If it doesn't fail, but could be done with a simpler solution - well it's exactly what engineering is all about. Finding the simplest solution, but not simpler than that, is the holy graal of engineering. Some have spent a lifetime finding the sweet spot.

[Mr. Scram]

god

I never said such a thing.

[Mr. Scram]

not tools, im not saying its not manufacturer-able, just later you look at it and some other detail comes to mind and you are laughing because the 'looks right' solution was the best one after all\

the idea that its something random that comes up. which makes it funny.

Considering how many posts you've made about the actual intent of this thread I'd say it's one example.

[coppercone2]

 

[dom0]

specs that end up butchered by reality

Lots of power MOSFETs.