发信人: madman (NaN), 信区: EE
标 题: Re: 我怎么觉得DEVICE对学CIRCUIT DESIGN的人没什么用处。
发信站: The unknown SPACE (Wed May 15 13:43:37 2002), 转信
Don't have as much experience dealing with complicated systems.
Even with simple circuits, one with experience and insight can see the key
points beforehand and design sepcial simulations to verify the thinking.
Otherwise, maybe thunsands simulations later, one representative case comes
by chance; how effecient.
Less experienced people may run simulations and think hard if the results
make sense; and new simulations can be designed. Dimensions, orders of
magnitude, asymptotica cases can be telling.
Saw a leafet on the Physics Department Computer Lab, Peking Univ.; it was
quoted from someone, about "what really matters with computation is the
analysis, not numbers". Another way of saying it is "garbage in, garbage
【 在 stealth (顺顺当当) 的大作中提到: 】
: I fully agree with you, after having worked with full-custom analog circuit
: design for my Ph.D. over last three years.
: Intuition is important, and experience is more important. You can only play
: math on very few transistors. The math would become very complicated in e.g.
: five-transistor amplifier , even you use the simpliest equation for each
: transistor. Of course, it never hurts to have solid math, which will help you
: understand the simulation result. But from my experience, if you can only
: learn one kind of math, I would recommend you learn numerical analysis,
: instead of device-physics-math. Having solid knowledge in numerical analysis
: would help you a lot when you design/simulate large circuit. I often have to
: deal analog circuit with 10k+ transistor, and I was naive enough before to
: believe everything from the simulation.
: If you also fabricate and test the chip you design, experience is very
: important. That's is my biggest lesson dealing with large analog circuits. You
: have a good idea, you design well, you simulate well, you verify and simulate
: the final layout very well, there are still good chances that the chip won't
: work as you expected when you test it.
: 【 在 madman (NaN) 的大作中提到: 】
: : Yeah. I believe if one really understands something, one can describe
: : it in two or three sentences in English. One who struggles with pages of
: : math does not know a thing; and those formulas can be forgotten easily.
: : In real work, what is important is good intuition, simple calculation, and
: : the idea of the orders of magnitude. Seldom is there time to play with
: : math.
※ 来源:．The unknown SPACE bbs.mit.edu．[FROM: 188.8.131.52]