Q: feedback and input impedance in a emitter follower

"Eeyore" skrev i en meddelelse >

Yes - but instead of writing all these reply's with all that noise you can answer the basic question: WHY is R6 the dominant input impedance?

Peter

Because it's feeding a 'virtual earth' as a result of the negative feedback from R5. That's why the figures for input impedance with and without the feedback network in place are so hugely different.

Basic electronics.

You've ben reading the wrong books or paying too much attention to university professors and their waffling theory who know bugger all about real circuits. I don't even need a data sheet to solve this kind of thing although some simple ( valid ) assumptions need to be made.

It should take no more than 5 minutes on a piece of paper.

It's a lousy configuration for noise figure too btw.

Graham

Where did I say I was doing a *simple* analysis?

The OP asked "...how to do this right and how to calculate the input impedance, gain and so on...".

My purpose was to show one way to "do this right".

He's pointing that to a certain level of approximation your circuit with feedback is similar to an inverting op amp with a 10K feedback resistor and

1K input resistor. The base of the transistor is, to some extent, a "virtual earth". So, as a first approximation, you could expect the input impedance of the circuit to just be the resistance of R6, 1K ohm. And the gain of the circuit to be -R5/R6, or -10.

However, a more accurate analysis shows the gain to be about 10% lower than that, and the input impedance to be about 5 % higher. Your measured values also show about that level of error in the approximate values.

Without feedback, R6 is not so dominant.

Which you totally failed to do since you clearly can't analyse a simple circuit !

And what's the tolerance on those h parameters btw ?

Graham

To 'some extent' ???

Close enough that the answers are in practice swamped by tolerances !

He measured 0.9% in fact.

As for the gain, The Aol of that circuit is ~ 100, so with closed loop feedback targeting a voltage gain of 10 it will indeed be in the region of 9x.

Without feedback it's a totally different circuit !

Graham

The correct formulas require information about the transistor which is usually not known. Approximations commonly used are:

Gain = R3/re

Z = Beta x re in parallel with (R5/Gain)

re is the ac resistance of the emitter and is often estimated by .026/Ie where Ie is the dc emitter current

The Gain and Z above are from the right end of R6

Without doing an accurate analysis, how would one know whether this is true or not? How does your "simple" analysis enable a person to determine the sensitivity of gain and input impedance to transistor parameters?

After doing a number of accurate analyses, a new designer of circuits will eventually gain the experience to know when approximations are good enough.

Accurate analysis has its place.

He says he measured an input impedance of 1.08 Kohm, which is 8% higher than 1 Kohm, not 0.9%.

So you agree with my statement?

!

I *did* analyse it, so how is it that I failed?

The Vishay data sheet only gives tolerances for hie (3 times variation) and HFE (2 times variation) for gain group A.

Given those tolerances, what does your "simple" analysis method say the variation in input impedance and gain would be?

I had no idea when I first responded to your question that this was going to turn into a pissing contest; and I really did think that it was a homework problem because that circuit is generally a real loser. It is typical of academic assignments, where the intent is to teach concepts but not necessarily specific examples of good design.

Anyway, accurate modeling (for this circuit ilk) is usually done with either hybrid parameters or scattering parameters. The hybrid parameters are often not available or are too flaky to be trusted. So, many practicing designers use circuit approximations ... dig back into today's posts and you will see how simple they can be. But, they can be the source of errors approaching

100% ... but those kinds of errors are usually easy to deal with in practical designs ... you know, design for an approximated gain of 2X greater than that will ever be needed, and then add a volume control (or use AGC).

If you want to crunch the numbers it does that just fine.

I reckon I can produce an answer as accurare as you like. Expereience shows it's not vital to do this. Relying on transistor parameters is kinda odd anyway !

I thought he said 1.088 kOhm? That's close enough to 9% for me.

As for accuracy, k is lower case and most units including Ohms start with a capital letter !

feedback

Which one ?

Graham

circuit !

Perhaps you could elaborate on your analysis ?

Minute. That's off the top of my head.

The circuit without feedback will have a more variable input Z with hfe but gain will be almost unaffected.

Graham

circuit !

As I said in an earlier post, and you apparently didn't notice:

"I've posted the analysis over on alt.binaries.schematics.electronics with the subject line "Andersen circuit analysis". Unzip it into a temp folder and double click the index file. It should open in your browser."

gain will

How about sensitivity to the other 3 h parameters? How about some numbers rather than just saying "more variable" and "almost unaffected"?

Show us how to use your simple method and get the change in input impedance and gain as hfe changes from 100 to 110.

it's not

But how would you get the experience without first having done some accurate analyses?

How else would you get an accurate analysis? I guess you're saying that you rely on approximations.

But *you* said: "He measured 0.9% in fact.". You did *not* say 9% until just now.

capital

feedback

The one right above yours, where I said "Without feedback, R6 is not so dominant.".

gain will

What parameters. There are none of interest.

Hiow about you learn to analyse real circuits ?

Graham

So minute as to be negligible.

Sure I can work it out. Who'd want to in reality ?

Transistor parameters of interest here vary by at least +- 50%.

There is therefore no such thing as an accurate analysis hence it's stupid to even bother with one.

The trick is to design parameter sensitivity out as far as possible.

Graham

So you keep saying, but why are you reluctant to share with us how your simple method would solve this problem?

People who want to increase their knowledge base would want to. I'm sure some of the other readers of this thread as well as myself are eager to learn a simpler method to calculate sensitivities to variations in h parameters.

even bother

A worthy goal. To know when the goal has been reached, you must have a method of determining what the parameter sensitivities are. Will you share your method of doing this?

capital

--- Units are only capitalized if they are abbreviated, otherwise they're written in lower case. That is, the correct form is either, say, 1 volt or 1V. 1 Volt is incorrect.

-- John Fields Professional Circuit Designer

The impedance looking directly into the base is proportional to beta, pretty close to

Zin = 26 * B / Ie Ie = emitter current in mA.

which ain't very free, since beta could range from 20 to 1000 for various transistors.

Knowing that, you can calculate the open-loop voltage gain of the transistor. And knowing that, you can calculate the effective impedance looking into the feedback resistor. Now put all the relevant impedances in parallel, add the 1K series, and you're done.

John