One for the MIT alumni

What is R4 for?

I bet bitrex _doesn't_ have an answer >:-} ...Jim Thompson

I don't! I remember analyzing something like this years ago, but I'm not sure if I drew the circuit correctly from memory. Actually I think it's easiest to analyze as a transimpedance amplifier - base current flow in R3 of Q1 is negligible, so the current from base of Q1 goes into R4, and the drop across R4 generates the output signal current across R5.

I did analyze your negative capacitor filter on paper though, that was fun. It's unstable!

Removing R1 and treating it as a transimpedance amplifier, I get Vo =

-Ii*R4*R5/(R4 + 1/R3) for that structure.

I think perhaps you should brush up on your feedback amplifier principles before suggesting such a circuit as an interview test (or presenting an analysis).

May I suggest Rosenstark's text as offering a practical analysis approach.

Looks like I drew Q2 as a PNP. Embarrassing!

It sounds like Bertrand Russel lived in a world only inhabited by geniuses and fools, which must have been a continually depressing situation.

Worse. Russel described a false dichotomy. Most of the geniuses are fools as well.

Nope. Russell was a realist. The world continues to be the same as he saw it... probably worse. ...Jim Thompson

Yep everything is bad it's all terrible.

Would you like to see my analysis of the negative capacitor filter? I can't guarantee it to be correct, but it was fun!

Yes, I would like to see it. ...Jim Thompson

ation.

People do foolish things when they aren't paying full attention to what the y are doing at some particular instant. The genius is more likely to get di stracted than a regular person, but Brian Josephson is more the exception t han the rule. Ben Franklin and Isaac Newton are better examples of the bree d.

Part 1:

Part 2: to show such a transfer function must always have a pole in the RHP...

Marvy! Now show us your derivation on the -C itself. ...Jim Thompson

If you knew loop and nodal analysis...

If you knew loop and nodal analysis you'd have no difficulty analyzing the active circuit and seeing the problems with such a circuit. ...Jim Thompson

I'm sure I was better at it years ago, but I don't have a lot of need to work on stuff at that low a level, these days. Use it or lose it...

What wasn't intuitively clear (to me at least) was whether the filter transfer function is unstable for _any_ arbitrary values of components, not just R1 = R2 and -C1 = C2. It's fairly straightforward to show that it's impossible to not have a pole in the RHP.

Other interesting question: is there any way to amend the circuit so it is stable for some selection of components, and improves the phase response over the standard 2 pole RC (though perhaps not 30 uD)

A positive real pole will always produce a time resp Message-ID:

that the only "value" of a negative capacitor is in parallel with a slightly larger "real" capacitor, but it saves no net driving current requirements. ...Jim Thompson