Math and electrical desgin

That's because you aren't very good at explaining things. Feynman used to say that anything he couldn't explain to an undergraduate in ten minutes was something he didn't actually understand.

I have successfully explained all the things in your examples to complete science-illiterates. It's possible, but you need to find the right entry point.

CH

Designing something cool with a few smart people at a whiteboard is the most fun you can have standing up.

Cheers

Phil Hobbs

Yup. That's part of the way I maintain ascendency over various whippersnappers. ;)

Cheers

Phil Hobbs

By definition, anything that only comes with practice is *not* instinct.

I think you mean intuition. Intuition comes with practice. It's just a cognitive habit, and habits come with practice.

Students are taught to think of electrons flowing inside wires. Is it any wonder they have trouble with RF, where they have to visualise everything as a field, with wires to guide the field?

Clifford Heath

----------------------- Flyguy wrote:

mpedance to someone w/o math; it just won't make any sense. Or how a capaci tor can conduct current. Or how a transformer works. Or how RF signal trans mission works. Had a discussion once with an FAA person on why radar can pi ck up non-metallic objects like birds - it pretty much ended when I mention ed "dielectric constant." It really becomes indispensable when you start de signing control systems or advanced signal analysis.

So your quoting him is absurd.

A wise man once said " everything in science has an easily understood expla nation - only problem being, that explanation is not true.

...... Phil

Instinct is super useful for generating ideas. We come up with some scheme by instinct, but then test it by math. The math involved is super familiar--what's the noise floor, the bandwidth, the settling time, and so on. It's the familiarity that makes that seem like it's the same as design instinct, but it isn't.

Cheers

Phil Hobbs

(Who does a lot of math because he's fortunate enough to do a lot of stuff where the instinct hasn't developed yet)

I think I could do that, purely visually with no math.

Those can be explained pretty well without math.

He also said that he had to visualize something before he could start to apply equations to it. Einstein said similar things: visualize a concept first.

I haven't solved a simultaneous equation or used matrix math since I was forced to do that in college.

I did once take an integral, about 25 years ago, to compute a mosfet dynamic power dissipation, but I already knew the answer well enough from simple arithmetic. Spice would have been more accurate than either.

Did you program the FFT yourself from first principles, or did you look up the code somewhere?

They are so impressed and stunned that they don't check our numbers.

Especially in such large projects it is nearly impossible to use a 100% top down approach. There are too many unknowns. It is very common to design a portion of a large design, construct and test it before continuing with t he integration. This is common with such components as engines in the spac e program.

Yes, I didn't say top down is never used. I simply said it is not always u sed 100% and is not always the best way. It is very common to separate out a portion of a design and investigate that portion before designing it int o the whole. There is no point in trying to design the system surrounding and depending on a component if there are questions about the component tha t can only be resolved by further design or even construction and test of t hat component.

I don't expect this conversation to get anywhere with you. We have disagre ed on things many times you often refuse to even try to understand what I s ay. Your reply this time doesn't offer any promise of a change to that pat tern.

A Discrete Fourier Transform is easy to program from first principles, as soon as you know what it actually is. FFT not so much, but having done a DFT at least gives you the intuition about how an FFT is possible.

An FFT is not so hard. You just have to get your head wrapped around the permutations. I can't imaging being the guy who first came up with it. Li kely recognized something in the sequence of sines related to something els e in math he was familiar with. That's what the FFT is about, decomposing the DFT sines into operations that can be shared among all the harmonics.

Don't recall the name, but I think this was before it was very useful given the state of computing, so I don't know that the guy was looking for a way to speed up the DFT. I suppose a wikipedia search would answer that quest ion.

g.

ph.

e.

e permutations. I can't imaging being the guy who first came up with it. Likely recognized something in the sequence of sines related to something e lse in math he was familiar with. That's what the FFT is about, decomposin g the DFT sines into operations that can be shared among all the harmonics.

en the state of computing, so I don't know that the guy was looking for a w ay to speed up the DFT. I suppose a wikipedia search would answer that que stion.

I should have searched first. Tukey and Cooley specifically were looking f or a way to speed up calculations. I was thinking this was invented in the 1950's, but it was 1965 when T and C published their paper having reinvent ed what Gauss discovered in 1805. Great minds think alike.

Anyway, in '65 computers were prevalent but sorely lacking in speed so the FFT was a big improvement.

You need top-down and bottom-up simultaneously.

Bottom-up allows you to reuse known working components, processes and concepts.

Top-down on its own can lead to requiring impossible components.

Who was it who said "the best result of mathematics is that you don't have to use it all the time".

The fundamental maths leads to the mental short cuts we use everyday in visualising circuits.

calculate how good it _could_ be, (b) figure out how to get there, and (c) if that's a win, proceed, but if not, think up something better and goto(a) .

as a lot of lore and old fashioned crank-turning.

Amen, George H.

Any explanation of how a transformer works that doesn't involve the transformer equations

V1= L1.dI1/dt + M.dI2/dt

V2= M.dI1/dt + L2.dI2/dt

turns out to be eventually misleading. It took me years to get onto that, and it's definitely a case where math is essential.

John Larkin doesn't seem to have got there yet.

g.

ph.

e.

e permutations. I can't imaging being the guy who first came up with it. Likely recognized something in the sequence of sines related to something e lse in math he was familiar with. That's what the FFT is about, decomposin g the DFT sines into operations that can be shared among all the harmonics.

en the state of computing, so I don't know that the guy was looking for a w ay to speed up the DFT. I suppose a wikipedia search would answer that que stion.

Danielson and Lanczos published in 1942, Cooley-Tukey didn't publish until

Lanczos is famous for a lot of stuff like that - I've known about the FFT s tory for many years now. I cited his 1957 book on "Applied Analysis" in my Ph.D. thesis.