Cute amplifier - bootstrapped

You misunderstand what 'no signal' meant. It doesn't mean I think they put out a weak signal and you shouldn't have snipped the rest, which described the alleged 'simulation'.

The current source generates 'the signal', however weak or wonderful one may think that is. The point was that at 40kHz the cap shorts this signal to ground leaving only 2% of however weak or wonderful the original might have been.

I consider a 98% loss, of whatever the original signal was, to be essentially 'no signal'.

Seems flipper's "simple sanity" check failed. Hand-waving and use of amateur-level shortcuts about the voltage across a capacitor will get you in trouble every time.

flipper RUDELY attacked my supposition that a capacitive transducer is properly modeled by a current source in parallel with a capacitor.

Seems I've been vindicated by a paper I found, authored by no less than Marshall Leach >:-}

I'll E-mail the URL link to any "engineer" who asks.

My main task now is to break the anonymity shell of "flipper" and give "flipper" all the exposure he deserves ;-) ...Jim Thompson

Your _main_task_ is to swat anybody who doesn't treat you with the respect you feel you deserve? I feel so let down--here I thought you were a dispassionate seeker after Engineering Truth. ;)

Cheers

Phil Hobbs

Only when they rudely attack. Questioning is not a problem. Shouting that I'm wrong is unacceptable and is, indeed, subject to "swatting". flipper shouldn't be employed.

I am. Is my analysis of this circuit wrong? If you think I'm wrong, please point out where and prove it... no hand-waving BS about bad models... prove it.

Or go back and read about, "An eye for an eye..."

Instead of sticking your nose into postings late in the game, why don't you trace back all the way to the beginnings...

Subject: Re: Cute amplifier - bootstrapped Date: Sun, 13 Jan 2013 13:48:15 -0700 Message-ID:

"I pondered that schematic for some time before realizing what it was.

It's some amateur's attempt at making a band-pass filter (quasi-gyrator), but it has horrible input impedance...

Looking like 10nF in series with 16K at low frequencies,

and -223nF at 194KHz ;-)

The shunt feedback directly to the base, driven thru the 10nF capacitor was the clue."

A totally correct statement in ALL respects.

Then I was attacked. Seems to be the fun thing to do around here. Except, almost always, the attackers are inadequately experienced amateurs.

I AM correct on this. But I was repeatedly slammed by absurd statements... even by a _supposedly-renowned_ PhD >:-} ...Jim Thompson

I thought ET was like big time wrestling... |-P

(Oh... wrong "ET") tee hee hee

noy having examined the circuit in detail 98% seems quite high, what's the resitance of the base-emitter junction?

The norton equivalent models uses the available (short-circuit) current as the input while the thevenin model uses the open circuit voltage.

The current that is "lost" doesn't cost anything, it's just not realised.

I don't see how that follows. obviously the current needs to be scaled with the frequency if you want to match a fixed-voltage thevenin equivalent.

if the model really shows 98% of the current flowing through the capacitor a lower resistance Q1 would seem advantageous.

Bootstraps in general fall into the category of application-specific tweaks rather than general-purpose devices. My main interest is SNR, as you know, because SNR and stability are the basic limitation in measurements--anything else I can fix up afterwards if I have to, in the easier parts of the circuit later in the signal chain. Bootstraps are mostly there to fix the frequency response, with little or no change to the SNR. (You can get a bit of SNR improvement from a good bootstrap in some instances, but generally only 3 dB or thereabouts, unless the second stage is really horrible.)

I don't usually care very much what the reactive part of the input impedance is *per se*, or what the circuit would do with different input conditions from the ones it was designed for.

Negative conductance, that I care about. SNR and bias stability too. But real bootstraps, as opposed to toys, are quite subtle, and not easy to model analytically or numerically--there's a huge difference in performance between A_V = 0.995 and 0.9995 (both quite reachable figures in many instances), but good luck telling them apart reliably with a SPICE or hybrid-pi model.

Cheers

Phil Hobbs

Marshall Leach says you're wrong.

All this hand-waving isn't engineering. If either of you could do loop and nodal analysis you'd know what was happening. ...Jim Thompson

And the applicability to the thread at hand is....?

And you threw some more baloney. Spice can easily find that difference... you just need to set your .OPTIONS properly and sacrifice time for accuracy. ...Jim Thompson

Shouldn't that say "the current coupled by the capacitor" as opposed to flowing through?

Read "coupled" as 'shunted to ground'.

Maybe that will help you see it.

And, as usual, you provide not one shred of evidence to support the claim. Like explaining why a 'capacitive transducer', standing alone and independent of any amplifier, would have a declining output with increasing frequency.

False. In fact my first comment visa vie the 'current source' model was simply "I'm not convinced that a current source is a proper simulation of a capacitive transducer."

But rather than support the model you've been hurling "punk," "village idiots," "stupid," and other pleasantries at not only me but the entire newsgroup as a whole and after the 'dismissal from class' I simply responded with your own words. After all, plagiarism is the sincerest form of flattery.

What's so 'secret' about your 'vindication'?

_YOU_ are pinging me about being on-topic? This is a thread about bootstraps, last time I checked.

And the point I'm making is not the accuracy of the ODE solver, it's that you can't rely on simulations to correspond to reality at the 1E-3 level. As you know quite well.

Cheers

Phil Hobbs

Nope. It's a thread about accuracy in analysis... which I was challenged on.

I've already proved to anyone knowledgeable in loop/nodal analysis that the "bootstrap" is actually a low-grade gyrator and the load impedance presented to the first stage is inductive... which everyone but you seems to want to deny. You want to call it a "bootstrap"... which I'm beginning to think is a bad use of words to describe just a feedback network.

I don't know that. But if you're referring to LTspice set to use the default solver, you're right

...Jim Thompson

Neeerp! You've been going on and on and on about losing that current.

You're just bloviating, because you can't do loop and nodal analysis, which every sophomore student in EE is supposed to know.

If you could you understand how the circuit works. But you can't You're a fraud. (Is that name-calling? Not if it's true... and it is.)

Do your own surfing.....

...Jim Thompson

Ok, I'll bite, why is there no such thing as a "true" differentiator? The op-amp circuit above is a pretty good example of a "true" differentiator in my opinion.

I was referring to *your* comment that "Any AC coupled amp is a differentiator". That is not correct, period.

Rick

My dear GOD! You guys are all what, 12 years old???

I bet you and Larkin get together at night and watch SuperTroopers!

"Bite it, rook!"

Rick

A true diff would need infinite bandwidth.

It is by Jim's standards for a differentiator.

Jim is 107, I think.

In my 2N5089 model it comes to about 4.4k.

By saying 'lost' I'm making a frequency response comparison such as between 100 Hz and 40 kHz, the later being the surmised operating point. Now, in practice the operating point is the operating point but JT is drawing bandwidth, and other, conclusions based on the output response vs frequency, but that's dependent on the input response.

I don't see any 'scaling' or other 'function' driving his current source, nor has he ever said there is anything, and the curves he plotted indicate it's just a plain ole fixed AC current source.

That would just shift the frequency up.

If the models are intended to be 'equivalent' then we have a problem because they produce wildly different results. If they're not intended to be 'equivalent' then we have the question of which, if either, model the transducer the circuit was designed for.