General question about algebraic circuit analysis

Well yes, but Spice does the tedious stuff ;)

In other words, I reality-check the results, and may calculate some starting values, but the final pass is always done in SW.

My favorite Calculator for the first pass stuff, is this one

and I often paste the eqns and the answers into project text files/ source code.

-jg

Nice. Units, too!

Too bad no complex numbers or functions of complex numbers, matrices, etc. But then there is Octave, so I'm good.

Jon

When doing feedback loops, non-trivial filters, control systems, or any analog electronics that has state-of-the-art performance, you have to think first and simulate/prototype later; not vice versa. I am actually getting paid not for the code/circuit, but for the problem solved.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

I do conceptual stuff symbolically, then back up that up with Spice.

Actually, now that I'm mostly a systems engineer and assumed to be too much of an egghead to be competent shoving electrons around, usually I do the conceptual stuff symbolically, then some hardware guy spends a lot of effort trying to prove me wrong in Spice, often proving me right without my having to lift a finger.

I forgot to mention that one thing that a symbolic transfer function does for me is to give me a deep understanding of the actions of each component. Until you write out the transfer function of the passive network of a Clapp oscillator circuit you won't understand the juggling act of selecting the three (or four) capacitor values for best circuit performance -- without that understanding, any simulation exercise approaches the 10^6 monkeys with typewriters trying to produce Shakespeare.

Yes, but rarely. I find a pretty small class of problems that's so complicated that you can't figure it out by inspection, but still simple enough that you've got a prayer of sitting down and working out a meaningful equation.

At least a few years back I know that those who working above around

I reasonably often work on signal-conditioning front-ends for sensors (e.g. thermocouples) - i.e. mostly DC stability stuff. Tolerancing and temperature drift (and input offset voltages and currents) are the Big Baddies, and I need to ensure that any trimmers have exactly the needed travel and no more - and that every one of (potentially tens of thousands) of these suckers that get built will work, so long as components are within tolerance. So yeah: I tend to write a transfer function (often as a Thevenin node-nibbler) and throw raw horsepower at it and work through all the permutations ;). Monte-Carlo ain't for me...

PS: just recently re-ran one of these which I'd written in C around '87. Back then, it took several minutes on a 286-class machine. Now, on my 3GHz Core 2, the result is there before I can take my finger off the return key...

Steve