I would go with B, since one of the things one needs to do with any feedback system is to make sure the phase sign/margins are correct.
By explicitly stating a negative sign as the desired outcome, there may be the tendency to skip this check?
OK - the 'S' node should show a '+', not an 'x' :)
Prefer ?
Not knowing what's inside both blocks - even if they hold the same - no way to decide upon. Could be that one of the blocks holds an inverter :)
OK - only slightly more serious, assuming none - or both - of the boxes holds an inverting stage:
Regards
if the feedback is negative, the block diagram should be for negative, and so on
It is interesting to see how many combinations are possible between the signs (kinds) of these three system components - summer/subtractor, noninverting/inverting amplifier, noninverting/inverting feedback network. In addition, in the case of a subtractor, there are two combinations (Xi - XZ or XZ - Xi).
Some of them will lead to a negative feedback, and others - to a positive feedback. Of course, not all of them would be feasible and useful...
Not a lot of difference between the two conventions. I'd say that A implies negative feedback is being used, but this is not necessarily the case. And similarly B implies that positive feedback is being considered, at least, but again, not necessarily the case.
So I'm totally with Cyril and Vasile on this.
There are too much anknown, so no preferance by this time.
For the sake of stability A is the mostly used.The aim of the system should be made clear before discussion.
For my opinion, we only should use the convention B.
Why? Because for negative feedback (desired usually) we require that the sum of all negative signs within the feedback loop must not be even. As a simple example, one of the blocks may have a negative sign. In this case we require a positive sign at the summing node. More than that, oscillators need positive signal feedback (but negative DC feedback).
Why should we restrict ourself to a special case only ?
Simple example: Phase-lead integrator with an inverter in the feedback path which is connected to the non-inv. input terminal of the opamp. There will be no problem if the denominator of the closed-loop function is (1-Aol*Hf) with Hf being the feedback factor (negative in this example). However, students will be confused if they start with (1+Aol*Hf) and do not realize that the inverter sign is already included.
Quote U. Dreher: If I want to build something non-oscillating, I'd prefer the negative feedback.
I rather think, in this case we are forced to use negative feedback. But this was not the question. We can have negative feedback also for a "+" sign at the summing node (when there is a phase inversion within the feedback loop).
For me either is ok.
When somebody else writes the paper, you do not have control on that!!
"When somebody else writes the paper, you do not have control on that!!"
But you have to be attentive and to check the chosen convention.
Perhaps it would be interesting to consider how and why the mysterious S component in the pictures above is implemented in the most popular op-amp circuits with negative feedback...
Generally speaking, it is a summer made act as a subtractor since the negative feedabck system needs the difference X-Y (VIN-VOUT) to be applied at the amp input. Kirchhoff's laws give us two possible ways to realize this subtraction:
https://en.wikibooks.org/wiki/Circuit_Idea/Series_Voltage_Summer
https://en.wikibooks.org/wiki/Circuit_Idea/Parallel_Voltage_Summer
Now, to make the discussion more lively, I will illustrate the two approaches above by an analogy from our everyday life... and maybe this will be the most incredible story about basic op-amp circuits you have ever read... But first, I want to warn you that this is just an innocent joke that should not be taken very seriously... it is just a humor:)))
A well-known situation in life is that when someone achieves something more than us, our mental equilibrium is disturbed:) ... and, simply put, we begin to feel envy. There are two ways to regain our peace of mind (two manners of envy):
Now let's see this behavior in humble op-amp circuits:
So the same op-amp is "good" in non-inverting circuits, and "bad" - in inverting circuits:)
And what about the following approach?
R2 is in the middle picture unmarked.
Josef - what is the question? The last two schemes are classical ones. And the first is a kind of combined positive/negative feedback - also used sometimes (for reducing the loop gain).
Yes Lutz
but hardly you can see today all the options together to realize that all mean "the same".
Josef
Vasil
I prefer triangles in drawing operational amplifiers too. But I did want to show general context only.
Perhaps it succeeded at least a bit :-)
Josef - one interesting application of the combined feedback scheme (first drawing) is the case where b=1 and ß=0. For simple resistive feedback this gives a unity-gain amplifier (independent on a or alpha) - however with variable loop gain (depending on the selection of a resp. alpha). This is a good and stable solution in case the opamp is not unity-gain stable.
A similar result is obtained for the case a=0 and alpha=1 (closed-loop unity gain with selectable loop gain) and simple resistive feedback for ß.
This is a very nice advice, but what about noise gain?
No surprise, of course. Everything in electronics (and also in our privat life) is a trade-off. That means - we have to pay for using a non-unity-gain stable opamp in unity-gain applications: Reduced loop gain is connected with increased noise gain.
Hello to all:
I like to extend Josefs question as follows - and I ask everybody:
Are you able - without having to think about it for a while - to give a circuit example for "series-shunt-feedback" and/or "shunt-shunt-feedback"?
I must confess - no, I am not.
For this reason, I strongly prefer and use another terminology: "x-controlled y-feedback" with x and y being "voltage" or "current".
Yes Lutz,
Simple examples are:
- noninverting opamp amplifier: series - shunt negative feedback, or serial to input voltage feedback which increases the input open loop impedance and decreases the amplifier output open loop impedance.
- inverting opamp amplifier: shunt - shunt negative feedback, or parallel to input voltage feedback which decreases the amplifier input pen loop impedance and decreases also the amplifier output open loop impedance.
MHO this discussion is a nonsense! Why?
Because anyone can use any preferable block diagram, and there are no problems. Problem rises when the block diagram should be explained correctly.
Dobri
Yes Dobry
maybe you are right.
But I have met already with the works where the author
mishandled working with "different references."
And was "pretty hodgepodge" - for example with the stability criterion (Nyquist).
Josef
Hi Dobromir,
you may be surprised, but I did know the answer to my questions.
However, in my questions the emphasis was on "without having to think about it for a while".
May I add that I don`t think the discussion would be "nonsense"?
In contrary, I think it is of great importance - in particular for education of students - to use words and terms which are - if possible - self-explaining, unambiguous and do not cause possible misunderstandings.
Don`t you agree that "voltage-controlled voltage feedback" is more clear than "series-shunt feedback"?
As another example, during many years of teaching electronics I have learned how many misunderstandings (and wrong interpretations) the term "current gain" (for the ratio Ic/Ib) has caused. I suppose, you know what I mean.
Dear Lutz,
I support your initiative and consider this issue important (at least to me it made me really interested and continues to interest me).
But I still want to say that "series-shunt feedback" has a substantial advantage over "voltage-controlled voltage feedback" - it shows the structure... the way of connection (series or parallel) between the two sources (input and output)...
In my opinion, some hybrid classification, in which the first word shows the way of connection (serial/parallel), and the second - the output quantity that is taken (voltage/current), would be the most precise...
Examples:
In this classification, the first word actually defines the type of the summer S (serial or parallel), and the second word - the electrical quantity (voltage or current) that is taken from the output and possibly (pre)converted to voltage.
(see my observations about the nature of voltage summers/subtractors on Page 2)
Dear Cyril - thank you for your comments and suggestions, which consist of a combination of the two basic conventions (your term: "hybrid" classification):
Example: Instead of "Series-Shunt feedback" resp. "Voltage-Controlled Voltage feedback" you propose "Series-Voltage feedback" (circuit example: Non-inverting opamp).
My comment (question) : Is it really an advantage if we explicitely mention the "way of connection"? I rather think, two voltages can be added only if they act in series, and two currents can be added only in a common node. So - the way how to add (substract) the feedback signal and the input signal is clear anyway.
As another example - lets take the inverting opamp-based amplifier. I remember that many students had some problems to verify that there is "parallel" combination of two signals at the inverting opamp terminal. However, it was clear that two currents meet at this node.
Of course, each of the three alternatives is correct, but it was Josefs question "What...do you prefer" - and I gave my personal view only.
Lutz,
As you know, the way of subtraction at the input defines key features of circuits with negative feedback (especially regarding the input resistance).
Regarding the "parallel voltage summer", really it is built on the base of a true parallel current summer (according KCL). By adding the resistors R1 and R2, it is converted into a "parallel" voltage summer. The resistors allow to connect voltage sources "in parallel"... and it is suitable for us to think of this composed resistive circuit as of a parallel voltage summer.
Cyril, yes, no doubt about it - from the "formal" point of view. And this explains the term "parallel" - but my doubt was (and still is) if this term is a good, clear and self-explaining term.
On the other hand, in your example do we really add two "voltages"? Or do we add two currents in a common node thereby producing a NEW voltage at this node (which, of course, can be split into two voltages - but smaller than each of the two driving voltages?
So - in contrast to the voltage feedback case (non-inverter) , we do not add (substract) the feedback voltage to the input voltage but only to a part of the input voltage.
Hmm... I do not know if you will believe me Lutz... but I have been thinking on those two issues anywhere from the middle 80's... and I am still not sure if I found the right answers...
Cyril - of course, I believe you! The same applies to me - because both of us are/were engaged in teaching electronics. However, we should not forget that we are speaking about "conventions" (Josefs term) and our personal feelings/views only. Therefore, there is no necessity for us to agree to each other. No problem at all.
Thank you, Lutz. I would like to take a moment to tell the participants in RG discussions that tomorrow, I go for 10 days on vacation and will follow occasionally the RG discussions.
Regards,
Cyril
Cyril - have some nice days. Best wishes for your vacation.
Lutz
Cyril,
""" In non-inverting circuits, the op-amp feels a "good envy" """
""" In inverting circuits, the op-amp feels a "bad envy" """
Such unusual "humor" !
It begins to explain your great imagination .
Buena Suerte ! while you are gone away.
the apprentice, Glen
Thank You Cyril.
I "connect positive feedback" and I wish for all a nice summer.
Josef
Hi everybody! I am already here and slowly going back to the RG reality:) I only wonder why nice things end so fast:)
Cyril,
Welcome back to this side of the "reality" fence.
There are 'nice things' on both sides of the fence.
Looking at the photo,
I see a hot coffee cup on the table,
and a cold water bottle in your hand.
mmm ... curious combination.
What is that word on the yellow umbrella ?
... I am guessing "Sterling Cider" ... "Best on Ice"
and some palm-leaves nearby .
Must have been one great vacation !!!
You may have been a very long way from "reality" !!!
Hi Cyril, where did you spend the vacation time? Looks nice. I think you enjoyed the time.
Welcome back
Lutz
Thank you for your interest ... but there is no mystery in my stay in this small resort town situated in the Rose Valley... and famous for its mineral water. In addition, it attracts young people with its amusement places where they play a lot of dancing:)
https://en.wikipedia.org/wiki/Pavel_Banya
https://www.google.bg/search?q=pavel+banya&espv=2&biw=1366&bih=643&source=lnms&tbm=isch&sa=X&ved=0ahUKEwiclOG0qMzNAhWDiRoKHRS7C3AQ_AUIBigB&dpr=1
Glen, my favorite drink is MX******* (the Greek spirit) but here, to set a good role model for young people, I keep a bottle of mineral water:)))
Now a little more seriosly:)
I think it would be interesting to consider the connection between the two block diagrams and the Black's formula.
Hi Vasile, thank you for inspiration.
"If we have G(s)=2, then this system stable? Why?"
In real world it must not be G(s)=2!
Every system has limited frequency band - this mean omegaMAX. Thus it must be
G(s)=2. omegaMAX/(s+ omegaMAX)
and it gets an another sense.
See the abstract of
https://www.researchgate.net/publication/281495081_POVINNY_POL_ZESILOVACE_PRI_ANALYZE_STABILITY
Conference Paper " POVINNÝ " PÓL ZESILOVAČE PŘI ANALÝZE STABILITY
"If we have G(s)=2, then this system stable? Why?"
My comment: This is a simple statement (without any proof).
Why should it be stable?
Because Blacks formula gives a finite value? That`s by far not sufficient.
The closed-loop poles cannot tell us anything because the system does not work in its linear range. Why not? Because we excite the noninverting system G - and the result is negative (phase inversion). This is a clear indication that the whole system does not work in a linear mode.
Yet what is the role of the sign in the denominator of the Black's formula? I think it depends on the type of block S in the block diagram - a summer or subtractor; respectively - an inverting or non-inverting amplifier.
Yes - of course.
And - as I have mentioned several times: I strongly vote for a summer (because it covers the general case).
I continue deeply thinking about the role of the block S...
According to me, the most correctly is to say that its role is comparing the input and output quantity by subtracting them that is implemented by adding their opposite values.
So, this is a comparator implemented by a subtractor that is made by a summer.
https://www.researchgate.net/post/What_actually_is_a_negative_feedback_system_What_and_how_does_it_actually_do_How_is_it_implemented_Is_it_really_a_negative_feedback_or_something_else
But I have another, more extravagant speculation about the role of the block S - it can be thought as of an input point for introducing the "disturbing" input quantity:
https://www.researchgate.net/post/What_is_the_simplest_negative_feedback_system_What_is_its_structure_How_does_it_operate_How_do_we_implement_simplest_negative_feedback_circuits
https://www.researchgate.net/post/What_are_actually_the_input_and_output_of_a_negative_feedback_system_Can_we_consider_the_input_quantity_as_a_disturbance_and_the_output_quantity-as_a_reaction
Cyril - may I ask you: WHY do you think that it would be "most correctly is to say that its role is comparing the input and output quantity by subtracting them" ?
There are many examples in electronics and control systems which need to add the feedback signal in the block "S" (second figure from Josef) because there is already a phase inversion within the feedback path. That is the reason I prefer to say: For negative feedback we need a negative loop gain - independent on the location of signal inversion.
I somehow slowly come back to reality ... maybe the mineral water, draft beer, dances and beatiful women have softened my brain:))) Perhaps the salt water of the Black Sea would help me? I may try it later... but now let's try to concentrate on the topic...
It may seem trivial but it can be helpful to clarify what is a summer and what - subtractor: It seems both the summer and subtrctor are the same summing circuit but:
Cyril - I suppose, nobody will object your explanation.
However, does that mean that you still prefer the subtractor? As a consequence, we are forced - in case of positive feedback - to require a second inverter within the feedback loop. Does this make sense? I am really not convinced that it would be (quote) "most correct to say that its role is comparing the input and output quantity by subtracting them" .
Do you think that the scheme in Josefs Fig. 2 would be incorrect?
Dear Lutz,
I am just thinking on the "philosophy" of this block diagram trying to figure out the meaning of each component. I remember how, in the distant 60's, when I was a student in an electrotechnical school (a class of Control systems), my teacher drew this block diagram on the black board without any explanations... but I needed them... and now I am trying to fill that gap...
So, what is the role of the component S?
OK - I see.
My simple philosophiy is as follows: At the summing node (sic!) two signals are added. That`s all !. In some cases (negative feedback) we have in the feedback path a signal inversion - and in some cases not (positive feedback, oscillators). There are even cases with positive and negative feedback at the same time (some opamp circuits)!
In my opinion, to figure out the role of S, we should initially exclude it from this block diagram... and simply connect the output of the inverting amplifier to its input ( I mean a negative feedback)... As a result, this possibly simplest negative feedback (control) system will reach the equilibrium - the zero value at the amp input... and will keep it...
So, in the simplest case, the NFB system keeps the zero at its input without any summer or subtractor... and it has only one input quantity that is actually its output quantity...
Then we (the input quantity) come and decide to somehow influence this well balanced system. We have two choices - either to add or to subtract our quantity. To do it, we need another input... i.e., we should include a summing/subtracting circuit at the amp input...
When adding a quantity at the one input, we upset the balance and disturb the system. So it reacts to our intervention by subtracting the same quantity at the other input thus restoring the balance (a zero at the amp input)... and v.v.
It can make this subtraction in two manners...
... but in both cases there is a subtraction between the two quantities.
So my thought is that to have a negative feedback, as you Lutz said, we always have in the feedback path a signal inversion... but we can introduce the external input signal either by summation or subtraction depending on the summing configuration (parallel or serial). Nevertheless, in both cases, we have a subtraction.
Cyril - to the first of the last two posts: This suggestion applies to a special case only (inverting opamp). But Josefs diagram is a general one.
I do not completely understand the contents of your last post. Do we speak about a general block diagram philosophy or about an opamp circuit?
Dear Lutz,
I also do not completely understand the contents of my last post:) ... but I feel happy to ratiocinate on this issue... Now I will try to say what I wanted to say in my last post:)
Maybe I wanted to say that always, in the case of the negative feedback, there is a subtraction... but we can implement it either by subtracting two quantities with equal signs or by summing two quantities with opposite signs.
In my opinion, the second Josef's diagram is too general since it is a mixture of the two different cases (negative and positive feedback). I prefer more concrete diagrams where, at least, the kind (inverting/noninverting) of the amp is shown. The block S can be drawn with "-" inside.
"...but we can implement it either by subtracting two quantities with equal signs or by summing two quantities with opposite signs..."
Yes - certainly, no objections against these mathematical laws.
But again- do we speak about a block diagram representation (Josefs figures) or about "implementations" (your word) ? I think, we should strictly discriminate between these two alternatives.
Of course, there is a need of such an "extremely general" block diagram that comprises both negative and positive feedback, and both parallel and serial summation.
It would be interesting to me to see your opinion about my suggestion to consider the input quantity as of a kind of a "useful disturbance".
Cyril - excuse my stubborness, but now we are again at the beginning. Josef has presented TWO of such extremely general diagrams and his question was: Which of the two do you prefer?
OK - you vote for the first.
Perhaps it helps now to use the first block diagram to explain how a Sallen-Key filter works. This a filter consisting of a fixed-gain non-inverting amplifier with positive feedback. The feedback network is a well-known passice R-C ladder topology. Applying the first block diagram, we are forced to allocate a minus sign to this passive ladder network. For my opinion, this sound a bit "weird", does it not?
In contrary - no such problem of understanding with the second diagram. Therefore my former question: WHY do you prefer this option.? Where are the advantages if compared with the alternative to use a real summing junction??
Indeed, the second diagram is more universal and suitable for such abstract circuits... BTW it reminds me the gyrator.
https://www.researchgate.net/post/Can_we_swap_the_voltage_across_and_current_through_electrical_elements_What_is_the_use_of_such_a_swapping_technique_Can_you_show_some_examples
...abstract circuit? It is one of the most used active filter topologies. And there are other filter structures requiring also positive feedback. And - of course - oscillators!
If so, can you explain what is the role of the op-amp (follower) here?
In active filters, it is the task of amplifiers to provide positive feedback with the aim to enable Q values greater than 0.5. Passive filters allow only Q
I need some more intuitive explanation related to some basic circuit principles. I am not familiar with this circuit but let's try to explain it to some "imaginary students":)
I would use my favorite step-by-step building approach. So, first I will remove the op-amp and the capacitor C1 thus beginning with the humble RC low-pass network (R1+R2, C2). Then I will show the problem - the small slope of the amplitude frequency response. I will ask them, "How do we make it sharper"?
Here, I suppose, some "imaginary student" will suggest to connect another RC low-pass network in series (to cascade)... and I will split the resistance... and connect a second capacitor C2 (initially) to ground. As a result, we obtain a circuit of two cascaded RC low-pass networks - R1C1 and R2C2.
Then I will provoke my students asking them, "But can you make it even more sharper?" And, as they would be probably familiar with Miller theorem and other techniques for creating virtual elements, someone would guess somehow to modify the capacitor C1 by connecting a following voltage source in series ("lifting" the ground at low frequencies).
So, at the final step, I will disconnect the right end of C1 from ground and connect it to the output of an op-amp follower...
Cyril - I would follow another method for explaining the circuit.
1.) Without any amplifier, we find a second-order lowpass with two RC sections in cascade. (Hence, we do NOT forget the capacitor C1). This passive lowpass has a very broad transition range (between pass and stop band) corresponding to a very small quality factor Q0.5, which has a better ("sharper") transition from the pass band to the stop band. (Q=0.7071 for Butterworth response and Q>0.7071 for the various Chebyshev responses.).
5.) Perhaps the following can help to understand the feedback action:
* For very low frequencies we have less feedback (due to the large impedance of C1 at low frequencies);
* For very high frequencies, we have less feedback because the ampliifiers output signal is heavily damped (low impedance of C2).
* For a certain (selected) frequency in between we have enough feedback (depending on the C1 value as well as on the amplifiers gain) to cause the wanted magnitude enhancement. If the gain is too large, the poles move to the imag. axis and the filter oscillates (for some Sallen-key alternatives this happens for a gain of "3").
I think the two explanations are quite similar...
Here is my final definition of this filtering circuit:
The Sallen-Key filter consists of two cascaded RC low-pass circuits where the latter controls (by means of an amplifier) the ground of the former.
Is there such a definition of this legendary circuit in "thick books":)?
(Only one remark to your explanation: I am not so sure about the role of the positive feedback here because of the very small gain).
I think, instead of "controls the ground" we better should say "provides another input signal for the former" - because C1 is not grounded anymore.
(For your information: I have worked for more than 10 years on the area of active filters - and, under my leadership, many students have prepared their diplomas)
An interesting viewpoint at the first RC circuit - thus it can be considered as a 2-input frequency-dependent summing circuit... like the simpler R1-R2 summing network...
Of course, you should not take seriosly my chain of thoughts above. I wanted only to demonstrate my way of thinking when trying to realize unknown circuits... I hope I was not too tedious:)
"Physical note" to the filter see below (qualitative view only). Exakt detemination of Q see for example
https://www.researchgate.net/publication/281480600_Theory_of_electronic_circuits
Book Theory of electronic circuits
Josef - I must confess that I do not understand the first drawing. For my opinion, primarily the phase is responsible for the peak.
To show this effect we have to add two phases at the upper node of C1: (a) The phase caused by the lowpass action with respect to the input signal and (b) the phase caused by the high-pass action caused by the feedback signal. It can be shown that - in the frequency region of interest - the phase difference is rather small (condition of positive feedback).
Yes Lutz, you are right. My simplification was too large. This situation (with the follower) is actually very interesting. If we have the ideal follower, the voltages at C1 and R2 are identical; the voltage at C2 equals to the output voltage. It enables us very easy construction of a phasor diagram for the chosen output voltage (and for given frequency). We are able to determine all currents and voltages by "graphic way" only - and see phase and amplitude ratios simultaneously.
I tried it and it was interesting :-)
Josef
Josef - interesting aspects! Thanks to your question (which feedback convention...?) we have discussed interesting properties of positive feedback circuits.
Lutz - an example is below. Relationship between input and output voltage is obvious.
Of course we can continue for other omega.
Hi Josef - nice diagram. I will try to understand it and give my comments later, OK?
Thank you.
First comment: The diagram confirms the effect of positive feedback:
1.) The diagram was constructed for one frequency only - the pole frequency wo. In the diagram, this can be verified because the output voltage lags the input voltage by exactly 90 deg. This is in full agreement with filter theory (90 deg phase shift).
2.) The output voltage is larger than the input voltage by a factor of app. 2.5. Again, this is in full agreement with theory because the pole-Q for the given parts values is 2.5. It is a well-known fact that the peaking at the pole frequency is very close to the Qp value.
I must confess - although busy with filters since more than 20 years - that I have never seen such a diagram before. Thank you.
Lutz Yes, it is the consequence of your comments :-)
In such a way we can get the corresponding input voltage for selected omega and the selected output voltage.
Yes Lutz, "omega zero" was really choosed for its very easy "verification".
But I've sketched out approximately other frequencies also.
It's (graphically only) very quickly and it works.
We just must to start "from the output and go to the input" :-)
Josef - unfortunately, you have shown this method a few years too late. Otherwise, I would have used this nice "teaching tool" for improving the students understanding of the working principles of active filters.
Glen,
Sallen-Key is "multiple feedback" also - see "The Bridgman-Brennar multiple-loop feedback biquad" in annex. The filter with inverting OPA needs infinite gain (ideally).
In "classical times" the ultimate gain needed by the Sallen-Key filter
was easily done (with vacuum tubes and transistors). And it was very great advantage :-)
Lutz, I did not use this solution before.
It is the result of our debate.
I add an another solution :-)
Josef
Hi Vasile,
it is only an example of a "normalised filter". In general we are able it all convert easily to any frequency (and values of R or C).
Glen - perhaps you have overlooked that we here didn`t want to discuss the various filter topologies. The only reason the Sallen-Key structure came into the game was to have an example for positive feedback. THIS was the matter we were discussing in this thread. The MFB topology as mentioned by you is based on negative feedback - and, hence, could not support the subject of this discussion.
By the way - there are also NEGATIVE gain Sallen-Key filter structures (less known, but possible).
"Back to the "Thread" => "Feedback Convention" ."
Yes - and Josef has shown a nice and intuitive way to demonstrate how positive feedback can produce an output voltage which is - in the pole frequency region - larger than the signal input, even if the amplifier has unity gain only.
Dear all,
I will be silent for a while (about ten days). I will investigate waves - in the Black sea.
I hope you will discus diligently :-)
Josef
Yes, it is evident that both conventions are used.
But we must be very careful.
For example the Nyquist criterion "for A" and "for B" differs "visuallly".
For details see eg .:
https://www.researchgate.net/publication/281480600_Theory_of_electronic_circuits
Book Theory of electronic circuits
Hi Josef - for a good understanding of your contribution it would be helpful to know which "conventions" you are referring to. For example, I cannot find "case B" in your book. However, from the drawings, I can assume what you mean - however, why do you see the necessity to distinguish between these two cases?
Let me explain my position: From the stabiliy criterion, we know that the loop gain LG must not be "+1" (oscillation condition). Hence, the critical point in the Nyquist plot is "+1" (your case B). Here I have mentioned the loop gain LG, which is the gain of the complete open loop (including ALL sign inversions). Measurement and simulation of the loop gain function LG(s) requires opening the loop at a suitable node. That`s all.
Now - why considering another form of the Nyquist plot (your case A)? Here we have nothing else than a graphical representation of "-LG". This is idenical with the product of all contributing function blocks - however, WITHOUT the sign inversion at the summing node of the corresponding block diagram. But there are cases, where we don`t have such a sign inversion because this takes place anywhere within the feedback loop. This situation can lead to misunderstandings (and that`s what I have experienced with students several times !).
Therefore, I always recommend to use case B only. Which means: Plotting a Nyquist graph for the actual and complete loop gain LG(s) (including all sign inversions within the loop) and use "+1" as the critical point.
Lutz
generally you are right.
But if we work vwith OPAs - I think the case A is useful also :-))
Josef
"But if we work vwith OPAs - I think the case A is useful also :-))"
To me, for opamps with negative feedback the feedback factor is "k=-R1/(R1+R2)" and the loop gain is LG=-k*Aol. This applies, of course, to inverting and non-inverting operations.
An interesting case are circuits with both negative and positive feedback (e.g. NICs)... where there are two Ks... and another subtraction...
Hi Cyril
Yes, this problem we have discussed somewher, but I do not exactly know where it was.
Josef
Yes Cyril. Exactly.
In this case, it is very important to have both signs and to add both k values (with different sign). And for stability, we require that the result is still negative.