Hello Josef, I think there are two basic ways to learn subjects like circuit theory:

a) To build the structure of knowledge beginning from the fundament like building a house. Of course this means starting with things which in itself are not very exciting. In school, we would start with Kirchhoff's laws and probably Ohm's law.

b) To be driven by a desire, for example to build a radio. One can start with a DIY kit, and learn at least how to solder. Then comes a highly important branching point: You can build the next kit and so on (and learn nothing but soldering), or you can be driven by the desire to investigate and to understand: Wouldn't the radio work better if you would use lower resistors and have higher currents?

Formerly, teaching in school followed method a): Entering the classroom, the teacher would say: "Today we will define the concepts of current and voltage ..." Now since didactics isn't an exact science like circuit theory, you can sell wild ideas as paradigmas. So one or two decades ago, someone thought that method b) is much more exciting than method a) (which is true), and "action-oriented" teaching was born (which is questionable).

Now entering the classroom, the teacher says (not literally but effectively): "We are driven by the desire to build ... (something)". This is only half a lie because indeed the students will rather build something than learn Kirchhoff. But in many cases there are two things lacking:

The desire to investigate and to understand, on part of the students. And the time to design and to build some applications *and* to thoroughly cover the theory behind it. So, while the ideal student would discover that she or he needs a resistor at a certain point, and that one must know Ohm's law in order to determine which resistor ... and so on, many students are satisfied with somehow getting a working schematic (e. g., via Google).

Additionally, as a student I was often not sure whether a certain formula or derivation was just an example or very basic. I think with action-oriented teaching this uncertainty is even higher. (Perhaps Ohm's law is valid only in radios?)

Paradoxically, most things I learned I learned by method b) (outside school or university) but in my opinion method a) is more suitable for people lacking a deep desire to understand the subjects at hand, or if time is short.

Having learned EE starting with Ohm and Kirchhoff ('method a'), I later had a lot of fun with 'action'. Coming from 'taskforce' activities - caused by colleagues that may or may not have been taught following 'method b'.

What was slightly different in the method a I experienced: we started with resistor and capacitor DC circuits. After 6 weeks we had a great intermezzo with MOSFETs that could be calculated with only R and C DC theory. This was great fun (and awfully long formulas.) Then came the chokes, AC and so on...

It is not necessarily the method in itself, it has to do a lot with how this method is 'applied'. I kindly recall that professor (long retired). And I am not the only one :)

So 'method c' might apply: make method a more 'digestable' - interrupting the dry matter with some 'fun themes' that can be analyzed with the knowledge acquired at that point.

Yes, it is difficult problem. Just imagine what all you must know for the simplest reciever ("crystal") below :-))

The circuit looks simple, but it isn't. I've always marveled the "analog gurus" like Bob Peace, Jim Williams and alike (many gone now) for their quest to minimize the number of components and maximize the use of each component.

This was driven by 'sheer need', but the results are marvellous.

Napoleon - I cannot completely agree with you.

So I like to ask you: What means "work"? Correct results?

I think this is not sufficient because - even in case of correct results - a wrong way for finding the results can cause conflicts and contradictions.

Here are two examples (and this is - at the same time - a direct answer to Josefs question):

* Many books contain the false statement that the BJT would be a current-controlled device. However, there are many effects which can be explained based on the voltage-control feature only. More than that, sometimes the current-control principle is mentioned, but all examples are based on voltage control formulas. 

* Some authors model the inverse transconductance of a BJT as resistor r=1/gm in the emitter path. I think, this is not correct because the transconductance connects  the input voltage with the output current.  Hence, gm is not identical to a classical two-pole element. In some books you can even read that this resistance r=1/gm would cause a feedback effect (like a stabilizing ohmic feeedback resistor Re). This is simply wrong.

Summary: Josef - I don`t think that in all cases the circuit theory is sufficiently (correct) taught.

Hi Lutz

and what about this model?

Josef - yes, I know this kind of "transistor model".

My comments:

1.) It is a mixture between a small-signal model (which should contain only controlled sources and passive parts) and an electrical circuit diagram . This model contains a new "artificial" symbol for the BJT with zero volts between the nodes B and E1. This could cause misinterpretations.

2.)  The current through the element in the emitter path is given with vbe*ge=vbe/re. In case of a real resistive element this current should be equal to the current through the nodes B and E (base current). However, here it is another current (ie~ic). I think - at least, this can confuse the reader.

3.) Where is the advantage of this model? Does it really help to understand the transistor principle?  It is stated that ge would be the "intrinsic emitter conductance". Hence, the reverse would also describe an "intrinsic emitter resistance" - is this really true? More than that, this element can be confused with a REAL intrinsic ohmic path resistances within the body of the BJT. There are some authors (as I have mentioned already) which even see a feedback effect caused by ge=1/re (as a result of this confusion).

4.) Why not use a name which clearly describes the physical function: Relation between input voltage Vbe and output current Ic (transconductance which characterizes the voltage-controlled current source in the classical small-signal model).

5.) Therefore - with respect to my comments above - where are the benefits of the "simplest signal model" of the BJT?

As with all theories, there are bound to be proponents and opponents; until a law is established temporarily but never permanently,

Dear Napoleon, in principle I agree with this sentence. However, in my example (BJT), it is not only a (temporary) theory that the transistor is voltage-controlled but a fact, which can be proven. But - as I have mentioned - we face the funny and surprising situatioin that some authors claim that the BJT would be current-controlled although they present examples supporting the voltage-control properties (without realizing this conflict).

Hi Lutz.

The model that I use allows a "very convenient calculations" and gives us "right solutions".

It's not about the principle of a transistor, but about of its model (linearised)

This model I got "via nullor analysis" and it is "more physical" than "nullor model".

My question is more concerned with whether circuit theory is taught (currently) to a sufficient extent.

Josef

„The model that I use allows a "very convenient calculations" and gives us "right solutions".........

My question is more concerned with whether circuit theory is taught (currently) to a sufficient extent.“

Hi Josef - yes, perhaps my response did not fully meet the subject of your question. On the other hand - is there anybody who knows „how circuit theory is taught“ today (round the world) ? Who could be able to give an answer ?

I think, everybody who is engaged in circuit theory has his own experience that cannot be generalized. And that was the background of my response which was based on an example. And my experience with students concerning the T-model for the BJT (and your model is a special derivative of the T-model) is as follows: Students become confused because of some „unusual“ properties of this model (if compared with models based on hybrid or y-parameters). Let me explain:

(1) The transconductance gm appears as a resistor r=1/gm between the base node and the emitter node.That means: The value of this resistance represents the input resistance as seen from the emitter only (common base configuration). Hence, for common emitter configuration there is a contradiction between the mathematical expression for the correct input resistance and the T-model, which suggests that the input resistance would also be 1/gm.

(2) In case of emitter degeneration (Resistor RE in the emitter path) we have two resistive parts in series: r and RE. Consequently, the denominator of the gain formula contains the sum (r+RE). This expression can be misinterpred because one might think that both parts play the same role (e.g. feedback). But, of course, that is not the case.

(3) The BJT acts as a voltage-controlled current source (voltage-current transducer). Therefore, the understanding of the transistors working principle - together with the derivation and application of corresponding expressions for gain and input/output resistances - is supported and confirmed if the equivalent signal model explicitely contains the parameter which controls the output current: A current source controlled by the transconductance gm.

(4) My conclusion: I don`t like the T-model at all. For my opinion, it does not enable "very convenient calculations", if compared with other classical small-signal models. But - of course - such considerations always represent pure personal views only.

Regards

Lutz

Hi Lutz.

I think that model that I used is better then "T models" - certainly it is more physical.

And this feedback - it is very interesting problem - worth thinking about it.

Josef

Josef - coming back to your main qestion I like to present and discuss the following example:

The jpg-file shows a classical 2nd-order lowpass in multi-feedback topology (MFB). Without the parts R1, C2 and R3 the remaining parts form a classical MILLER integrator. That means: If we inject a signal voltage into to the node A  the opamp output V(out) will be the time integral over V(A) (with a minus sign).

Now the question: Will this be true also for the shown (complete) circuit ? Is V(out) still the time integral over V(A)? Now we use the input V(in) and don`t inject V(A). Can we answer this question without time consuming calculations? Can we find a quick answer based on the laws of the circuit theory? We note that V(A) now is a voltage which is not provided by a signal generator (zero source resistance). Instead, V(A) depends on V(in) as well as on V(out).

If V(out) would be still the integral over V(A) the transfer function V(A)/V(in) would have a bandpass characteristic because V(A) would be the time derivative over V(out) (multiplied by "s" in the frequency domain). 

Background of this example: If the considerations as mentioned above are valid, we could easily derive the bandpass function V(A)/Vin using the known transfer function of the lowpass (Vout/Vin). And the question is: Is this procedure allowed? If yes (no) - why (not)?

(I hope I could clearly describe the problem.)

Lutz,

"The limit states" are clear.

I do not think that Vout is time derivative somvhere - if the OPA is ideal.

Josef - I am afraid you misunderstood something. I did not claim that V(out) would be a time derivative.

I will repeat the corresponding part:

If V(out) would be still the integral over V(A) the transfer function V(A)/V(in) would have a bandpass characteristic because V(A) would be the time derivative over V(out) (multiplied by "s" in the frequency domain).

More than that: Why do you refer to the "limit states" ? They have nothing to do woith my question.

With the aim to find an answer to Josefs question, I have tried to start a discussion on an active filter circuit two days ago. However, up to now - no answer to my question?

So I am afraid, that perhaps I was not able to describe the problem (and my corresponding question) with sufficient precision. (On the other hand - if the description was clear enough the lack of reply could be a sign that Josefs question (title of this discussion) could be answered with „no“).

 OK - so I will give the answer in the following:

Yes - the output voltage V(out) of the opamp is the time integral over the voltage at the node A. That means: It is unimportant if the voltage V(A) results from signal injection (zero source resistance) or is the result of the currents within the surrounding network (including feedback).

This finding is in accordance with „Substitution Theorem of Network Theory“. In this context, I have noticed that there are only very few books in which this theorem is mentioned (although the consequences of the theorem are widely applied without any explanation).

With this in mind, I think the question if “today the circuit theory is sufficiently taught“ can be answered with „not always“.

 Finally - because differentiation is the inverse operation of time integration - V(A) will be the time derivative of V(out). Based on the popular lowpass transfer function V(out)/V(in) which can be found in many textbooks, we easily can find the transfer function V(A)/V(in) without any further calculation by multiplying the lowpass function with (sR4C5). Hence, V(A)/(Vin) is a bandpass function having the same pole location as the lowpass.

Yes Lutz, you are right.

I like to use admittance procedures. When the procedure is correct it will include everything necessary "automatically". See attachment for example.

I links up to my works:

https://www.researchgate.net/publication/281612858_Nullor_nodal_analysis_as_a_result_of_a_nodal_analysis_with_ideal_op_amps

https://www.researchgate.net/publication/281961470_MODERN_INTEGRATED_ELECTRONIC_DEVICES_IN_LINEAR_CIRCUIT_THEORY

Conference Paper Nullor nodal analysis as a result of a nodal analysis with i...

Conference Paper MODERN INTEGRATED ELECTRONIC DEVICES IN LINEAR CIRCUIT THEORY

Any electronic trial circuit or network can be interpreted in terms of integrodifferential equations in time domain.  Since the solution is complicated to understand,. We use transforms.  The elements do not change their nature but in composition with other elements, the outcome is resultant.