I have been trying to figure out some things in electronics lately...i think today i hit on a big one that mentally closed a loop for me in a way.

it relates to op amp circuits and transfer functions.

i dont remember if the variables within transfer functions are all circuit dependent parameters...i know that frequency is always involved...and i remember a lot of sin() cos() integration and derivatives.

but i guess the question is this...

if i can describe the behavior of a system with LaPlace/Fourier transformable equations, can i not simply break down the results and reverse engineer a function that way?

is that how app notes guys come up with their circuits?


???

 

I can't speak for hard core designers but I have a bit of experience with it.
Yes you can somewhat work from a transfer function but it's a bit tricky. You know what a pole or zero looks like so you can put them in... but I've never actually managed to do that.
I have however, simply thought of the basic behaviour.
Capacitors are a short in high frequency, inductors in low frequency.
Decide what kind of current path you want at what frequency and you can sprinkle some elements down.

For example, a PI controller is a very high DC gain which drops off at higher frequencies but stays non-zero.

Op amp circuit gain is Zf / Zin
For high gain you want Zf high and Zin low.
For low gain you want Zf low and Zin high.

To get a high DC gain you make Zf huge at DC - a capacitor. Just make Zin a resistor for now, Rin.
Coincidentally at high frequency the capacitor will be low impedance, lowering your gain.
Practically you don't want a capacitor by itself in the feedback (due to offset currents and such) so you add a large parallel resistor.
That adds a DC path, lowering Zf. Your gain is no longer infinite, but oh well.
You also don't want the gain to go to zero at high frequency, so add a resistor in series with the capacitor. Then your high frequency gain is just Rf / Rin.

For more complicated circuits I think you'd start off with a basic idea, put it down, then find faults and fix them, or add features.

 

That does make sense. I just have never made that connection until today for some reason - between transfer functions and circuit based characteristics like frequency and impedance...not sure why - its kind of a "DUH - you idiot" moment, to be honest

 

i dont know if self-bumping is outlawed here, but i am dying to know if this is true or not.

have pity on me mods.

 

App note guys are aliens who bleed silicon. They sleep on a pile of paper as a pillow and their drool dries into patterns that we read as app notes. Thats how its done.

I think.

 

App note guys are aliens who bleed silicon. They sleep on a pile of paper as a pillow and their drool dries into patterns that we read as app notes. Thats how its done.

I think.

Hm.

Interesting.

I should kidnap one, make him drool into a beaker and sell it on ebay to private engineering firms.

 

Tried that... they dont last long.. and boy do they STINK...

 

your favorite app note engineer...lol

couldnt resist.


we are not all nerds!

honest.