Hi Cyril, shall I take over again the part of an „Advocatus diaboli“?.

OK - let`s go.

I must confess that I cannot follow each of your initial statements.

[Quote:“ Then, a few days ago, I thought, since there are "current steering" why not also "voltage steering"? To help answer this question, ....]

What is the question? If voltage steering is possible? (I remember, for example, how to control currents in tubes and transistors,..). Or do you refer to another "steering" phenomenon? Is there a difference between "steering" and "control"?

[Quote:The two variable resistors are nonlinear and opposite of the input quantity: in the current steering arrangement, they are constant-voltage; in the voltage steering - constant-current. As they are incorrectly connected (the constant-voltage - in parallel, the constant-current - in series), they interact ("tangle") and vigorously change (crossfade) their instant resistances in opposite directions when one of them tries to change its resistance.]

What means „incorrectly connected“?

[Quote:In the attached picture below, you can see both the dual steering techniques implemented in the famous differential amplifier....The current steering is between the two upper legs (T1-T3 and T2-T4) of the long-tailed pair; the voltage "steering" is between the two transistors (T2 and T4) of the right leg.]

Let`s take, for example, the „voltage steering“ case. Are T2 and T4 ,respectively, the two resistors in series, which change their values in „opposite directions“? I cannot visualize this property.

Regarding the „conflict“ situations:

1.) I think the voltage „conflict“ could be better demonstrated if you divide the common emitter resistor in two parallel resistors - without a common node. Now you have two non-ideal voltage sources. And the so called „conflict“ arises by connecting both emitters again. However, as I have mentioned earlier, it is no real problem because both sources have a finite source resistance. This is a classical network problem for beginner students (several sources and resistors in a common network).

2.) In the circuit given by you, there seems to be a real current conflict (currents through T1 and T2 can be different - currents through T3 and T4 are equal.) However, for normal operation there is another transistor connected to the collector of T2, which serves a load that can carry the current difference (transfer from symmetrical to unsymmetrical operation). Thus - also this „conflict“ is solved.

These are my comments.

Regards

Lutz

Lutz, frankly, I was hoping for a more positive and constructive comment concerning the basic idea behind the phenomenon of "current steering"... perhaps because this idea is deeply connected to me personally I am so sensitive... For the first time I met this phenomenon in 1984 when I used it to create a series of LED indicators (a point - see the picture below, line and plane version)... but just now (a few days ago) I managed to realize fully and presented it in such a neat, symmetric and dual way... as though these various combinations of dual sources (current - voltage), elements (constant-voltage - constant-current) and structures (parallel - series) show us the harmony in nature, particularly in circuitry... In my opinion, to see the harmony in the particular circuit solutions and then to generalize it into fundamental principles, is no less an achievement than inventing them, and although it does not bear the glory of their inventors, the pleasure is the same...

Now about your concrete objections...

[Quote:“ Then, a few days ago, I thought, since there are "current steering" why not also "voltage steering"? To help answer this question, ....]

Yes Lutz, the question is exactly about if there is such thing as a "voltage steering" or it is purely a figment of my imagination:-) My opinion is that there is such a phenomenon and it is exactly what is called "dynamic load". Only, the term "steering" is a little confusing here because the voltage is not a flow-like quantity and it seems strange to steer it. Maybe "crossfading" is more appropriate for it? Of course, we may use the traditional "dynamic load" or "active load" (see the picture below) but there is no symmetry with the dual current steering.

Now about the meaning of „incorrectly connected“...

The most elementary electrical circuit consists of two components - a source and a load connected to each other. There are two kinds of electrical sources and non-linear elements (loads) - constant-voltage and constant-current. There is no problem (a "conflict"), if we connect heterogeneous components (a constant voltage source to a constant-current source/resistor or a constant-current source to a constant-voltage source/resistor) since the one element sets the current while the other sets the voltage; the two elements do not oppose and even mutually "help". This arrangement is widely used in cascode circuits (in the picture below, the "voltage source" Q2 sets the common voltage while the "current source" Q1 sets the common current).

Of course, the most conventional connection without problems well known from the electricity, is a "voltage source to a resistor" or a "current source to a resistor".

Cyril, I am sorry that I was not able to deliver "more positive and constructive comments concerning the basic idea behind the phenomenon of "current steering".

Perhaps I would do if I could understand how this phenomenon is realized in the differential amplifier circuit.

Don`t overestimate my knowledge regarding "current steering" - I only know that this principle can be applied in D-A converters.

But where can it be found in the diff. amp?

And where can I see the "voltage steering" principle as indicated by you ?

(Remember my question: Are T2 and T4 ,respectively, the two resistors in series, which change their values in „opposite directions“?).

Perhaps it would be a good idea if you would give here your general explanation for "current steering" resp. "voltage steering" ? I am sure, this would help a lot.

Thank you and regards

Lutz

PS: Excuse me, I am getting older and older... but I didn`t understand which elements are "incorrectly" connected. Why incorrectly? What is the meaning of "correct" in this regard?

Maybe, in a special question about the long-tailed pair where to consider the differential mode of operation? And another question about a long-tailed pair with dynamic collector load?

So Lutz, imagine what happens if we connect a voltage source to another voltage source with different voltage or to a constant-voltage nonlinear resistor (e.g., a Zener diode) with a different voltage threshold - each of them will try to set its desired voltage; as a result, an unlimited current will flow between them. If we connect two constant-voltage nonlinear resistors in parallel, one of them will take all the current...

Quote: "Maybe, in a special question about the long-tailed pair where to consider the differential mode of operation?"

I have asked this question because you have used the long-tailed pair in your very first posting. Therefore: Is a separate question necessary?

Cyril - regarding your last contribution: Is this an example for current steering? I am a bit helpless.

Hi Lutz,

IMO the long-tailed pair deserves a special attention; a lot of interesting question can be asked about it (I have noted that you don't like to enlarge the subject with additional questions).

Frankly speaking, I am also not sure what exactly "current steering" is (how it is defined); I just use this term to designate initially the group of these similar circuit phenomena and then to discuss them freely. My idea is to evoke the reader's (student's) creativity in the field of the circuitry.

Maybe you are right about my last insertion - precisely speaking, a true current steering can be observed only in a system of three elements (a current source driving two constant-voltage elements in parallel). In this arrangement, the common current is held almost constant while the partial currents flowing through the two legs of the current divider crossfade; this gives an impression of steering (diverting, movement).

So, it seems we have to give up the 2-element arrangement (a voltage source connected to a constant-voltage resistor or a voltage source connected to another voltage source) where the current only "fades in" (increases) or "fades out" (decreases) but not "crossfades" (is steered). But we may keep the combination of a constant-voltage resistor (e.g., Zener diode) connected to a voltage source (e.g., a charging battery or a capacitor) when this network is supplied by a current source. This arrangement is often use to limit the voltage across the rechargable batery; this is also the case of the input biasing circuit discussed in the question below:

https://www.researchgate.net/post/What_do_coupling_capacitors_really_do_in_AC_amplifiers_Are_they_rechargable_batteries_conveying_voltage_variations_or_diverting_biasing_currents?

Regards, Cyril

Now more about what is correct and what incorrect when connecting two nonlinear elements...

A constant-voltage nonlinear resistor (e.g., a Zener diode) is designed to work in the vertical part of its IV curve (as a voltage stabilizer). So to be "fair" (correct) to him, we must provide him with decent working conditions so that the operating point is located on this vertical part. When connecting another constant-voltage nonlinear resistor (with different threshold) to it, our element, changing its instant resistance, tries to keep up its threshold voltage and, if the difference is significant, it "saturates". So, we are not "fair" to him:-)

So, "to be correct", we should not connect the constant-voltage elements in parallel; we have to connect them in series. Contrary, we should not connect the constant-current elements in series; we have to connect them in parallel.

[Quote: IMO the long-tailed pair deserves a special attention; a lot of interesting question can be asked about it (I have noted that you don't like to enlarge the subject with additional questions).]

Cyril - no, just the opposite is true. I was asking some questions regarding the diff. pair - and your answer was:"Maybe, in a special question about the long-tailed pair ..."

Thus, it was simply a misunderstanding between us.

More than that, I suggest to discuss the current/voltage steering phenomena based on one single example only: The diff. amplifier. This avoids confusion - in particular because it seems that there is not yet a clear definition, is it?

And therefore, I still have the question: Where are the two resistors in series, which change their values in opposite directions (as mentioned by you in your first posting)“? Sorry for my insisting on getting an answer but this is necessary for my understanding.

(Remark: Because there are no other replies from other collegues - It seems that I am not the only one who has problems to understand your intentions?).

Lutz

Lutz, I have (and gradually realized with your and other assistance) the ambitious idea to uncover the truth of basic electrical and electronic circuits through these questions and answers. This is my last web initiative after Circuit fantasia (2002), Wikipedia (2006) and Wikibooks (2007). These questions arise spontaneously, but I systematize them then in the form of datasets. Here is a list of links to my RG questions so far:

https://www.researchgate.net/publication/236896163_List_of_links_to_my_ResearchGate_questions_and_answers

I also systematize written by me into "essays" dedicated to basic circuit ideas. Here are the first two:

https://www.researchgate.net/publication/236899399_Virtual_ground_(my_RG_questions_and_answers)

https://www.researchgate.net/publication/236899906_Negative_impedance_(my_RG_questions_and_answers)

I wrote this explanation to clarify that for me, these questions are more than questions. That's why I try to distribute the best way circuit problems in separate but logically connected questions...

Data List of links to my ResearchGate questions and answers

Data Virtual ground (my Q&A)

In this connection, I intend to ask separate questions about the differential pair, dynamic load, cascode circuits, current mirror and many other great circuit solutions. But as you ask me here about the differential pair with dynamic load, let's begin discussing this clever idea first here. We may briefly disclose its secret if we think in the following "step-by-step building" way (imagine we have to reinvent it now, 80 years after his real invention).

-----------------------------------

1. To build a high-gain differential (long-tailed) pair, we need both current and voltage steering arrangement.

2. The current steering will help us to reject the common-mode input signal and to "tolerate" the differential mode; I can explain why if needed. To realize it, we should connect two constant-voltage elements (the transistors T1 and T2, acting as interacting emitter followers in the differential mode) in parallel... simply, we should connect the two transistors in parallel.

3. The voltage "steering" will help us to obtain an extremely high gain from the right output (the collector of T2); I can explain why as well. To realize it, we should connect two constant-current elements (the transistors T1 and T2, acting as interacting voltage-controlled current sources in the differential mode) in series... simply, we should connect the two transistors in series.

4. But... we have already connected them in parallel! We have already used the left transistor T1 by connecting it in parallel to T2; so we cannot also connect it in series to T2. What do we do then? How do we connect T1 simultaneously in parallel and in series to T2?

5. If we are smart enough, we can come up with the most incredible idea in the circuitry:-) - to "clone" the transistor T1 to T4 and then to connect the "clone" T4 in series to T2 as needed! Thus we have two identical transistors - an "original" and a "copy", that are connected in dual ways - in parallel and in series to T2, and play dual roles - constant-voltage (voltage-to-voltage converter) and constant-current (voltage-to-curent converter).

6. It only remains to find a way to "clone" the transistor T1. But what does it mean? It means that the T4 base-emitter voltage has to be equal to the T1 base-emitter voltage. Then the T4 collector current will be equal to the T1 collector current (we suppose the two transistors have equal transconductances); in addition, this "copy" current has to have an opposite direction compared with the "original" (to exit instead to enter T4 collector). In other words, the problem is to invert T1 current and try to "blow" into T2 collector.

7. If we are inventive enought, we can guess to pass the T1 output collector current through the collector of a an intermediate transistor (T3) with a parallel-parallel (voltage-voltage) negative feedback between the T3 collector and emitter; as a result, it adjusts its "output" base-emitter voltage to be relevant to the "input" collector current. Then we apply this "input" voltage to the base-emitter junction of T4; as a result, its collector current becomes equal to the T1 collector current... and the problem is solved.

8. But ... Eureka!!! We have actually invented another legendary circuit solution - the current mirror! Yes, no lie - T3 and T4 form a BJT current mirror! Overall, we have invented the famous differential pair with dynamic load!

-----------------------------------

Lutz, this really seems to be the most incredible story about the differential pair ever concocted... Frankly, I fabricated it a few hours ago as an answer to your revelation, " Where are the two resistors in series, which change their values in opposite directions (as mentioned by you in your first posting)? Sorry for my insisting on getting an answer but this is necessary for my understanding." Thank you for the inspiration, Lutz!

How wonderful would be if only Barrie Gilbert (Where do little circuits come from?) joins this discussion and corroborates or rejects our speculations!

Regards, Cyril

Maybe, we have to say a few words about the operation of the voltage "steering" right leg (T2 and T4).

Imagine a differential signal applied: Vin1 increases while Vin2 decreases >>> T1 collector-emitter resistance vigorously decreases while T2 resistance - increases (a resistance "steering" between T1 and T2) >>> T1 collector current increases while T2 current - decreases (current steering between T1 and T2) >>> T4 collector-emitter resistance vigorously decreases while T2 resistance - increases (a resistance "steering" between T4 and T2) >>> T4 collector-emitter voltage vigorously decreases while T2 voltage - increases (voltage "steering" between T4 and T2) >>> the output voltage Vout vigorously increases.

Hello Cyril,

I must confess, it is not a simple task to read your last text and to follow all the explanations. This needs some time - and I cannot comment at the moment.

However, it seems I have learned now what - in the context of your explanations - "steering" means: Working in opposite directions at the same time. Correct?

by the way: Yes - I know Barrie Gilbert`s "little circuits" and I like his way to explain things. In particular, I like his introduction to translinear circuits.

Hi Lutz!

During the weekends, I continue revealing the secrets of the famous current steering phenomenon... Oh my God, how I love those weekends when I can, undisturbed, unleash my imagination ...Today, after I made wonderful sunbathing in the university park near me and recorded my insights on my favorite voice recorder, I relaxed under vine in the yard before my study with a glass of my favorite drink (mint with Sprite) and menthol cigarette (in reaction to modern European trends:-)

I decided to learn more about the famous Barrie Gilbert`s "little circuits". Once again, I became convinced that he is a true circuit genius... but I must admit that it was difficult to understand his writings. I thought I knew decent English, but now I'm not so sure about that ... because there were too many unknown and strange words and phrases in his article...

Still, I was amazed once again by his brilliant thoughts and even invited him to join our interesting discussion. Of course, I had not hoped that he would do it, or would even respond to my email as, as a rule, famous people do not reply to unsolicited emails...

So it seems we need to exert our own imagination and thinking to solve alone the mystery of this phenomenon...

Hi Cyril - nice picture.

Regarding the book in your hand - I am sure you like, in particular, the first story about usage of a barometer. I think - this is exactly the way of thinking you prefer, correct?

I can tell you that very often I have asked the students the same question as in this story: How to use a barometer to determine the height of a house?

And the result was always the same as described in this story.

As we can see (and know from our practice): Experience is not all. We must be cautious that we do not stay within self-build fences. I hope you understand what I mean.

Have a nice weekend.

Lutz

Hi Lutz,

Frankly speaking, I have read only a few stories from this book but now I will read more of them... Most impressed me the story of "True Analog Circuit Design" by Tom Hornak with his visualization techniques since I use the same approach when thinking. The next story deeply impressed me is the Sheingold's "Op-amps and Their Characteristics". Although its name is not so interesting, it contains valuable thoughts about the nature of the negative feedback systems (particularly, about the "inverse of a function" - page 373).

I have not yet read the story about "how to use a barometer to determine the height of a house" but I suppose the so called "pressostat" widely used in washing mashines is based on the same idea - to measure the water level by measuring the water pressure; I will read it...

Lutz, I understand your last hint. But I believe you will agree with me that there are so many "standard" thinking people in this world that's worth at least one of them to think "outside the box"... and if there is such a man - to be encouraged, not punished for this rare ability...

Regards, Cyril

Really nice and surprising for me story, Lutz... it made me laugh:-) It would be better that all this is just a joke, but unfortunately there is some truth in it...

Cyril - perhaps you have another idea how to use the barometer ?

Nice "provocation" Lutz:-) Yes, I have... the opposite (to have no idea) is simply impossible:-)

My idea is to measure the water pressure in the water supply network with a manometer (with an opened tap) - the higher the height of the building, the less would be the pressure ... I have an idea even to measure it without any manometer or a barometer - can you guess how? The atached picture can help you-:)

As yesterday, I relax under the vine, after a hard "brainstorming" in the park ... the only difference is that now my drink is a cocktail of Coke and Metaxa and the cigarette is pink super slim Glamour (unfortunately no longer in the market :-) While cycling in the park, I have collected and voice recorded a lot of speculations about current steering, long-tailed pair, common-emitter stage, etc...

Best regards, Cyril

I have another idea - to measure my blood pressure at the first floor, then to climb the building with the lift and to measure my pressure again on the top floor... and to calculate the height. The advantage is that blood pressure monitors are more available than barometers...

Yes - but my initial question was how to determine the heights of the building with the help of a barometer (alternatives to the methods as mentioned in the story).

What about an acoustic principle caused by the crash when the falling barometer meets the ground floor?

Perhaps a spectral analysis of the terrifying sound prodused by the crashing barometer:-)?

Here is another "barometric" but now a fully electrical way of measuring the building height - it is based on using an electrical "manometer" for measuring the electrical "pressure". Can you guess how? The atached picture may help you-:)

OK - assuming that the barometer provides any ohmic resistance RL between two arbitrary points - why not measure and compare (on the roof of the building) two currents:

1) the current i1 through the device (i1=V/RL

2) the current i2 after the barometer is connected with two long (lossy, resistance Rw) wires down to the ground (i2=V/(RL + Rw))

Aha ... "barometer" = ohmmeter = length meter? But can we use the existing network instead of the additional wires?

Lutz, we have relaxed well in the style of Synectics and brainstorming, but maybe it's time to get back to the main question, "What is 'current steering'?"

In my opinion, we should learn a lesson from this story with the barometer, to listen to the message of the curious student and try to make the subject more interesting for yong people. Since we will be a good team to solve non-standard problems, and others may join us, I suggest you both make a fun "game" called "simulated current steering":-) The purpose of the game is to imitate (emulate, simulate, imitate...) manually the operation of the two transistors of a long-tailed pair.

The arrangement is extremly simple - a constant voltage source Vcc supplies through a steady resistor Re two variable resistors (rheostats) Rc1 and Rc2 connected in parallel. You probably guess that Vcc represents the power supply, Re - the common emitter resistor, Rc1 and Rc2 - the collector-emitter resistances of the transistors T1 and T2. If you prefer, connect Re to the positive supply rail and Rc1||Rc2 to the ground (n-p-n BJT); let's first consider a single-supplied "long-tailed pair".

In this "game" we are "actors" so let's allocate the roles. If you do not mind, I may be a T1, and you - T2? Also, we need to decide what to do to successfully mimic the T1 and T2. In fact, we (transistors) can do only one thing and nothing else - to change ourselves current resistances Rce. But how to do this?

IMO we have fist to mimic the common mode and next - the differential mode. So, we have first to set some voltage across Rc1||Rc2 by simultaneously changing their resistances in the same direction, and then - to begin vigorously changing them in opposite directions so that to steer the current from the one leg to the other.

Our task is somewhat complicated, because I think we have to keep the voltage across the two parallel variable resistors constant (?) until we change their resistances in the opposite directions. So, the first conclusion can be:

THE CURRENT STEERING IS PERFORMED AT A CONSTANT VOLTAGE DROP ACROSS TWO VARIABLE RESISTORS IN PARALLEL.

Thus it seems all the voltages (Vcc, Ve and Vc), the common current Ie and the equivalent resistance Rc are constant; only the separate resistances Rc1 and Rc2, respectively the partial currents Ic1 and Ic2 crossfade.

If Rc1||Rc2 = Rc, we have to keep up Rc/(Rc + Re) = const (think of the two resistors as a voltage divider Rc,Re) or simply Rc = const. So, another conclusion is:

THE CURRENT STEERING IS PERFORMED AT A CONSTANT EQUIVALENT RESISTANCE OF THE TWO VARIABLE RESISTORS IN PARALLEL.

Clearly, we must include measuring instruments - a voltmeter in parallel to the two rheostats to control the voltage drop across the current divider and an ammeter in series to each of them to monitor the current.

Now let's catch the two rheostat sliders and start acting ... Are you ready? START! :-)

Regards, Cyril

The current equals I=dQ/dt (the charge flow per time unit), which means the current flows and that's why you can steer it. Voltage on the other hand is defined as:

1. Uab=A/q where A is the energy required to move the charge q from point a to point b;

or

2. Uab=φa-φb where φa and φb are the potentials of points a and b.

That's why the term "voltage steering" doesn't have any meaning from a physical point of view.

So, the current is a flow-like quantity that can be steered between the two legs of a "complementary-variable" current divider while the voltage is a differential quantity that can be crossfaded between the two legs of such a voltage divider?

Hello Cyril - I am afraid that - at least - the following requirement is impossible to meet:

"THE CURRENT STEERING IS PERFORMED AT A CONSTANT EQUIVALENT RESISTANCE OF THE TWO VARIABLE RESISTORS IN PARALLEL.".

If both resistors are changed in opposite directions, the resulting parallel resistance never can be constant. Thus, I think it is "problematic" to model the transistors (acting as current sources) as two simple resistors.

Maybe it is only in a limited region? And fter that, the resistors saturate?

Also, take into account that the two resistances not only differentially vary; at the same time, they vary in a common-mode (in the same directions) to keep the total resistance and the voltage constant...

Note also the two resistors act as dynamic resistors, not simply as bare varying resistors...

Nevertheless we want it or not, the fact remains - the voltage drop across the common emitter resistor, respectively across the two resistors in parallel, is kept constant by the mechanism of the series negative feedback (the common emitter degeneration).

Will someone else join the game to help me? I am already too tired since I move both the sliders simultaneously:-)

Well if you want to model (simulate) the work of a transistor all you need is to define it's input/output characteristics. After that everything comes down to the specific scheme, which could be analysed with the Kirchhoff's laws.

If we go back to your initial image (scheme) there could be no such thing as "potential conflict". Write the second Kirchhoff's law for the following two contours:

1. Vcc-T3-T1-Re

2. Vcc-T4-T2-Re

2nd Kirchhoff's law:

1. Vcc=Ut3+Ut1+Ure

2. Vcc=Ut4+Ut2+Ure

This means that Ut3+Ut1=Ut4+Ut2

The voltage drop on T3 and T1 is the same as the voltage drop on T4 and T1 and they both act as voltage dividers. The way the transistors divide the voltage depends entirely on their input and output characteristics as well as their operating points (which might be slightly influenced by Uin1 and Uin2).

Boris, thank you for the comment, although it deviates little from the present theme of "current steering".

Saying "voltage conflict", I would like to present in a more figurative and colorful way (in contrast to the conventional textbook explanations) the fact that we have two devices simultaneously "wanting" to apply different voltages to the same point. From the electrical point of view, this situation is inadmissible. As a result, they react to this incorrect situation by changing their instant resistance and, respectively, the current flowing through themselves.

So, as you have showed above, the voltages across the two legs are equal (it is obvious as the two legs are connected in parallel; so you could have saved yourself the above analysis) but the resistances and the currents are different... and they represent the reactions of the two voltage sources (constant-voltage elements) to the "conflict"... and we convert these current reactions to voltage ones by connecting the collector resistors acting as current-to-voltage converters... and use them as output voltages...

About your last comment below:

"The voltage drop on T3 and T1 is the same as the voltage drop on T4 and T1 and they both act as voltage dividers. The way the transistors divide the voltage depends entirely on their input and output characteristics as well as their operating points (which might be slightly influenced by Uin1 and Uin2)."

It is interesting to me and I would like to ask you a few questions: "What are these voltage dividers? What are these transistor characteristics? How are set the operating points? Why might they 'be slightly influenced by Uin1 and Uin2'?"

Regards, Cyril

Cyril, the following remark is related again to you "model" that contains resistors only:

For my opinion this model cannot "work" - which means: it cannot reflect the reality with two transistors - because these resistors have linear V-I charcteristics.

This is in contrast to the BJT behaviour (non-linear V-I characteristic).

Yes... but if we move appropriately their sliders (change their instant resistances), we can mimic any kind of non-linear resistors. Remember that the nonlinear resistor is just a dynamic linear resistor:

https://www.researchgate.net/post/What_is_linear_and_what_is_nonlinear_Is_a_linear_resistor_static_and_a_nonlinear_dynamic_Is_a_nonlinear_resistor_a_dynamic_linear_resistor?

Okay now i understand why you refer to it as a conflict although i think the classical explanation with the V-I characteristics and the change in the operating point is better.

I mentioned the voltage divider T2/T4 because you referenced it in your initial post as a voltage "steering" which is an incorrect term in my opinion. By transistor characteristics i meant the V-I characteristics of the transistor. And the operating point - well it determines the resistances of the transistor.

Uin1 and Uin2 might change the operating point (aka the resistance) because they would create an additional voltage drop on Re (positive or negative), yet it is expected that Uin1 and Uin2 are of much lower magnitude than Vcc, so it could be neglected.

Well, Boris... We will have more opportunities to discuss "the voltage divider T2/T4" in the future question about the dynamic load and how to set the operating point - in the future question about the long-tailed pair...

For now, I just want to mention that only the common-mode value of Vin1 and Vin2 affects the operating point (the voltage drop across Re, respectively the common and partial currents); the differential value(s) does not affect it since in this case, the voltage of the common emitter point stays constant (a kind of a virtual ground).

But if we replace the emitter resistor Re with a constant-current element (CCS), the common-mode value of Vin1 and Vin2 will also not affect the operating point. In this case, the operating point is fully set by the emitter biasing current. This current is split and its halfs flow through the input sources to the ground (they have to be galvanic)...

Thank you for the contribution, Cyril

["For now, I just want to mention that only the common-mode value of Vin1 and Vin2 affects the operating point (the voltage drop across Re, respectively the common and partial currents); the differential value(s) does not affect it since in this case, the voltage of the common emitter point stays constant (a kind of a virtual ground)".]

Cyril - did you forget the "normal" usage of the long-tailed pair? I mean, neither the common mode nor the symmetric differential mode (Vin1=-Vin2), but the operation with two arbitrary voltages Vin1 and Vin2. Also in this case, the voltage drop across Re changes.

General (personal) remark: In the mean time I am a bit confused - and I don`t know anymore what we are discussing primarily. Too much information.

Cyril, it would be nice i f you could summarize and give me some help what the problem really is today.

Thank you

Lutz

Yes, Lutz... I agree... You mean the single-input differential amplifier is the "normal" usage of the differential amplifier... but it is named "differential"...

Not necessarily "single input". I mean: Two arbitrary input voltages (neither common mode nor Vin1=-Vin2). In this case (if the CMRR is finite), the voltage across Re changes - caused by the common mode part of both voltages.

Is there a dual "voltage steering"?.....could you please elaborate what is this dual "voltage steering"?

Yes Lutz...yes... a good example of a "mixed" mode (both differential and common). The constant current source in the emitters solves all these problems...

Dinesh, I have elaborated it enough in the overview under the question.

Lutz, my idea here is to reveal the most general idea behind the current steering arrangement and to see how it is implemented in practice.

Well, Cyril - I understand.

However, for my opinion this is the second step only ("to reveal the most general idea..."). As a first step, we need a clear DEFINITION what you mean when using this term "current steering". May be I am already a bit slow in thinking and understanding, but for my opinion such a clear definition is still pending - in spite of the various answers and contributions. Don`t hesitate to blame me if I am wrong.

Well Lutz, I mean what a very ordinary human being imagined - something like railroad switch enabling railway trains to be guided from one track to another. I have not read a definition of current steering (I don't like definitions at all as they remind me of something like religious dogma) but here is what I think about it.

IMO this arrangement should include first some kind of a current source producing the initial current. It supplies a network consisting of two elements connected in parallel (the "railroad switch") forming a current divider. As the current source can be imperfect (V + R), the voltage drop across this current divider should be kept constant so that not to affect the input current when they switch. This simply means the two elements have to be constant-voltage nonlinear resistors with almost equal voltage thresholds that slightly vary when controlled by the input quantities. This formulation describes the situation very well when we connect two diode elements (e.g., LEDs) with different threshold voltage in parallel but there are some problems in the case of the long-tailed pair...

An example of a current steering definition: Current steering is a technique where we switch (steer) the input current from one to the other leg of a current divider without affecting the input current.

Cyril - thank you. You do not like definitions? I think, they are absolutely necessary in order to avoid misunderstandings during written communication. Very often I have experienced such misunderstandings connected with the three terms "loop gain", "open-loop gain" and "closed-loop gain".

Back to "current steering".

[Quote:" Crrent steering is a technique where we switch (steer) the input current from one to the other leg of a current divider without affecting the input current."].

Fine - this can be easily visualized using an IDEAL current source and two parallel ohmic resistors. No problems in understanding.

However, this definition does NOT include the conditions you have mentioned before (constant voltage drop, constant-voltage non-linear resistors, voltage thresholds).

Thus my question: Are these non-linear properties necessary for current steering?

Lutz, I do not like definitions as they predirect my thinking to the "right" direction thus limiting my imagination. I have not forgotten that we have what to talk about loop gain in the topic about the ubiquitous negative feedback...

Very good question you have asked about the current steering, Lutz... the problem is that I have not yet completely realized it... I have only some genius "shots in the dark":-) And this is perfectly normal ... why else would I ask this question? Oh my God Lutz, how easy is it to not go into the phenomena, but to show ready truths on boring PP presentations... and how difficult it is when you want to know the whole truth...

In my opinion, the problem is complicated by the fact that we want to not only redirect (steer) the current from one branch to the other, but to use the partial currents for powering loads... but these loads have resistances (Rc1, Rc2) and will affect the partial currents... In the long-tailed pair, this problem is solved by connecting transistors as current-stable variable resistors ("current sources") in the branches (legs). They are, as Barrie Gilbert has shown, linear voltage-to-current converters; so the current is "linearly steered" between the legs (the partial currents depend linearly on the respective input voltages).

But nevertheless, the voltage drop across the two branches in parallel is again maintained constant. The reason of that is the series negative feedback (the common emitter degeneration). Another explanation, in the case of an emitter resistor, can be that the emitter current (the sum of the two collector currents) is kept constant; so the voltage drop across it is constant and, as a result, the voltage across the two legs is constant... but IMO it is poor...

Let's continue demystifying the mechanism of the current steering observed in the long-tailed pair...

Imagine, at the common mode, the two input voltages are simultaneously increased so, by the mechanism of the series negative feedback, the common emitter point is "lifted" to some "height". Then, at the differential mode, we begin wiggling the two input voltages in opposite directions. As a result, the common emitter point stays "immovable"; we can think of it as a kind of a "virtual ground". So, the two transistor stages act as "transistor stages with virtually grounded emitters".

If you (the reader of this text) try to explain this situation to human beings (students), you can do it in a figurative and colorful manner to stamp your explanation (and maybe a piece of you) on their memory. So present the long-tailed pair as a circuit of three resistors - two variable "pull-up" resistors (T1 and T2, temporarily neglect the collector resistors) connected to the positive supply rail and one steady pull-down resistor (Re) connected to the negative rail (the ground)... or present this combination as a kind of a "double voltage divider" with two separate upper legs and one common lower leg... or, even better, present it as a mechanical system consisting of three springs - two pull-up and one pull-down.

First, we have to "bias" this system of three stretched elements (if you prefer, stretch them in a form of a star:-) For this purpose, the pull-down element (the lower spring) begins slacking the common joint upward until it reaches the desired "quiescent point" position (close to the zero level, in the case of a "split supply").

At common mode (two positive input voltages), the two pull-up elements become stronger and begin simultaneously pulling the common joint upward while the pull-down element resists them pulling it downward. As a result, they lift the common point to some height and the three elements (springs) remain stretched (an equilibrium).

At differential mode (one increasing and the other decreasing positive input voltage), the pull-down element does not do anything; the one pull-up element becomes even stronger and begin pulling the common joint upward while the other pull-up element becomes slack to the same extent and looses the common point. As a result, it stays immovable... as though it is fixed to the real ground... it is a virtual ground...

Let's now return to the electrical domain and see again the impressive fact that, at the differential mode, the two transistors mutually keep up their emitter voltage(s) constant... they mutually "virtually ground" their emitters... and act as "transistor stages with virtually grounded emitters".... as two voltage-controlled current sources driving respective current-to-voltage converters (the collector resistors). This pair is "lifted" to the level of the common input voltage and it is kept up at this "height" by the mechanism of the series negative feedback...

Of course, all these explanations belong also to the future question about the long-tailed pair...

Regards, Cyril

Cyril, I suppose, that you cannot agree to the following - however, I am afraid that we move around in a circle - arriving again at the starting point.

Let me explain:

Your last contribution above describes the classical behavior of the long-tailed pair for the differential mode of operation as can be found in each textbook - without the necessity to use models ("non-linear constant-voltage resistors") or other artificial visualization techniques like "current-or voltage steering". That´s what confuses me a bit. Do you understand?

Lutz, we needed by these considerations... they are not redundant... we will use them when considering the long-tailed pair...

Well Lutz, now I will try to answer your persistent question...

Thinking constantly about the phenomenon of current steering and its specific implementation in the long-tailed pair, I began to realize deeper basic ideas behind it. Here are my latest insights into the circumstances surrounding the invention of this legendary configuration (I will present them in my favorite manner - by building and reinventing step-by-step the circuit):

1. We need to switch (divert, steer, move...) the input initial current between two branches connected in parallel and to drive loads with the partial complementary currents.

2. To do it, we have somehow to regulate the magnitudes of the partial currents through the branches; so we connect transistors (voltage-controlled constant-current resistors) in series to the loads (Rc1 and Rc2). They act as linear voltage-to-current converters (current "sources"); so the current can be linearly steered between the two legs (the partial currents depend linearly on the respective input voltages).

3. If we wanted to apply a voltage across the partial current "sources" (to supply them by a voltage source), there would be no problem. But we want to pass the input current through them (e.g., to supply them by a current source from the side of the collectors), so the problem arises. Just imagine what this means - a current source connected in series to another current source (constant-current element)... what happens is what I call "current conflict". I have tried to discuss this situation (without much success) in the question below:

https://www.researchgate.net/post/Can_we_apply_an_input_current_to_the_collector_of_a_BJT_whose_base_is_held_at_a_constant_voltage_and_to_take_the_collector_voltage_as_an_output?

4. Conversely, there would be no problem if we pass the input current through voltage sources. So we can remedy the problem of the "current conflict" if we make the constant-current elements in the branches behave as constant-voltage ones; this situation - a current source connected in series to a voltage source, is even desired. But how do we do it?

5. The series negative feedback (emitter degeneration) can help us to do this "magic". This simply means to pass the input current through the transistor emitters (to supply them by a current source from the side of the emitters) as we have discussed (with little more success) it here:

https://www.researchgate.net/post/Can_we_reverse_a_BJT_by_passing_the_input_current_through_the_emitter_and_taking_the_base-emitter_voltage_as_an_output_Where_can_we_see_this_idea?

As a result, the transistors will adjust their collector currents to keep the emitter voltage equal to the input common voltage, and the sum of their currents will be equal to the input emitter current.

6. As final conclusions:

- The two elements (transistors) in the branches behave both as constant-current elements (when driven by the input differential signal and when looking from the side of the collectors) and as constant voltage elements (when driven by the emitter input current source and when looking from the side of the emitters)

- We need the two elements connected in parallel to behave as constant-voltage elements (when looking from the side of the input current source) even if they actually are constant-current elements... just because to pass the input current through them

- In some limited cases, the two elements can be constant-voltage ones (e.g., LEDs, Zeners, diodes...) as in the attached picture below.

-----------------------------------

Lutz, this was another incredible story about the differential pair that "cannot be found in each textbook":-) I created it especially for you as a gift for your helpfulness. I hope it will dispel your confusion. Thank you again for the inspiration, Lutz!

Best regards, Cyril

It now remains only to explain why in the latter case, the two elements must be constant-voltage non-linear rather than simple ohmic resistors...

Hello Cyril, it seems that I am the only one who has problems to fully understand your explanations. Thus, your last contribution has created new "persistent" questions (sorry for that).

[„Just imagine what this means - a current source connected in series to another current source (constant-current element)... what happens is what I call "current conflict". I have tried to discuss this situation (without much success) in the question below:

https://www.researchgate.net/post/.......“]

My question: Was the conflict solved? And how?

[„4. Conversely, there would be no problem if we pass the input current through voltage sources. So we can remedy the problem of the "current conflict" if we make the constant-current elements in the branches behave as constant-voltage ones; this situation - a current source connected in series to a voltage source, is even desired. But how do we do it?

5. The series negative feedback (emitter degeneration) can help us to do this "magic". This simply means to pass the input current through the transistor emitters (to supply them by a current source from the side of the emitters)“]

My question: Where is a voltage source? Is the constant-current mirror in the collector path connected to a voltage source? How does this solve the conflict ?

My conclusion: You are right that there is a severe „current conflict“: The current mirror requires two „equal“ currents (forgetting the base current), whereas the main transistors act as current sources with different currents (controlled by the diff. input voltage).

However: We must remember the reason for using such a current mirror in place of ohmic resistances. The „conflict“ was desired - and it was solved by feeding another transistor stage with the „conflicting“ difference current I(diff) .

Thus, the long-tailed pair was no longer a voltage amplifier (with ohmic collector resistances) but a transconductance stage with voltage-in and current-out I(diff).

I think, we must not forget this approach for solving the conflict because it is the only justification for using the current mirror at all! Otherwise it would be without any sense to replace the resistors by the current mirror.

Of course, it would be interesting to examine what happens without such a possibility to allow a current path I(diff) for solving the conflict. Perhaps I will do some corresponding simulations.

Lutz, I will be thoroughly acquainted with your comment, but I just want to clarify that for the purposes of the current steering phenomenon I consider here only the classic long-tailed pair (with collector resistors, not with a current mirror)... and even without collector resistors (only two "pull-up" transistors and one "pull-down" emitter resistor)... And I am thinking how the current coming from the one element can be diverted (steered) between the two other elements connected in parallel...

For this purpose, we have first to "blow" the current into the two parallel elements and only then to begin steering it. My main idea is that we cannot simply pass the current through them if they are constant-current elements since they will not obey this our desire and will try to maintain their own current. We can "push" the current into a constant-voltage element (connect a current source to a voltage one) since, in this case, there is no "discrepancy" between the two heterogeneous sources; they will even mutually help. So we have to see how the constant-current elements are made behave as constant-voltage (from the side of the input emitter current source)...

Lutz, I like your explanations about using a current mirror as a dynamic collector load for the long-tailed pair. Yes, in this case we take the two current outputs (collector currents) of the transistor pair and make them contradict (we provoke a "current conflict" producing a "voltage reaction"). For this purpose, we "copy" the T1 collector current, pass it in series to the T2 collector currents and begin changing them in opposite directions. Kind of trick, huh?

But this is another story about the "dynamic load" great idea where heterogeneous sources mutually oppose... No less interesting is the dual story about cascode amplifiers where homogeneous sources mutually help...

But let us return to the main question - how to "push" the current through the two constant-current elements (transistors). To understand me, I suggest to you to conduct such an experiment (mental, simulated or real) with a long-tailed pair:

Set some equal input voltages Vin1 = Vin2 (e.g., 0 V in the case of a split supply) and then vary the magnitude of the emitter current. The collector currents vary as well (Ic1 + Ic2 = Ie), right?

Now disconnect the input sources from the ground and connect them to the common emitter point (floating sources). Set low enough voltages (about 0.6 V) producing the same effect as the voltages above and vary the magnitude of the emitter current as above. Do the collector currents vary as above? Why?

The difference is that, in the first case, there is a negative feedback that forces the transistors to adjust their base-emitter voltages so that to pass the 1/2 of the common emitter current while, in the second case, there is no negative feedback and the transistors will even react negatively to the input current change ("current conflicts" appear)... That was the point of the previous question:

https://www.researchgate.net/post/Can_we_reverse_a_BJT_by_passing_the_input_current_through_the_emitter_and_taking_the_base-emitter_voltage_as_an_output_Where_can_we_see_this_idea?

Hi Cyril - I am sorry, but I really had the impression you were discussing a long-tailed pair with a current mirror as a dynamic load. Otherwise, I do not understand the source or the reason of a "current conflict" you have mentioned. Can there be a "current conflict" in case of ohmic resistors? Where and why?

Refer to one of your earlier posts:

["But we want to pass the input current through them (e.g., to supply them by a current source from the side of the collectors), so the problem arises. Just imagine what this means - a current source connected in series to another current source (constant-current element)... what happens is what I call "current conflict"."]

Finally, I must confess that I feel a bit confused now about the aim of our discussion. I wonder if there is somebody else who can contribute to the subject.

Lutz, I have mentioned the long-tailed pair with a current mirror as a dynamic load and attached a picture about it (below the question) in the beginning only as an example of the dual "voltage steering" ("voltage crossfading"). Then I continued considering the current steering phenomenon as observed in the classic long-tailed pair (as you suggested)...

About your question, "I do not understand the source or the reason of a "current conflict" you have mentioned. Can there be a "current conflict" in case of ohmic resistors?", I can say. No, there are no current conflicts in the legs since current sources (the transistors) supply ohmic resistors (the collector resistors).

I have considered and directed your attention to the network consisting of the two elements connected in series. The upper element consists of two constant-current elements (the two legs) connected in parallel (we can think of them as of one constant-current element); the lower element is the emitter constant-current source. So it seems there are two constant-current elements connected in series that is bad... and it would be really bad if we drove the transistors by floating input sources (the imaginary experiment above). But I showed that due to the series negative feedback (the emitter degeneration) the upper constant-current elements are made behave as constant-voltage elements when looking at them from the side of the emitters. Thus, actually the emitter current source drives two voltage sources (emitter followers) connected in parallel (we may think of them as of one voltage source)... and there is no current conflict... This was a conflict for which I spoke...

Lutz, please do not be surprised that I so strongly consider this circuit trick. I do it since this is extremely important biasing technique (biasing from the side of the emitter) widely used in modern op-amps... old kinds differential amplifiers... emitter-coupled circuits (gates)... analog multipliers (Gilbert's creations)...

["But we want to pass the input current through them (e.g., to supply them by a current source from the side of the collectors), so the problem arises."]

OK - again a misunderstanding. But if your circuit you are speaking about has ohmic resistors in the collector`s path the reason for this misunderstanding was a typo in your circuit description (see the quote above:....from the side of the collectors).

In your last post you mention "two constant-current elements". Are you referring to the main transistors T1 and T2? Constant-current elements? Yes - without any differential input voltage. More than that, if you like to view these transistors as "voltage sources" - I have the following question: Between which two nodes are these voltage sources effective?

We can summarize all about the current steering and long-tailed pair into two general problems:

1. "BLOWING" A CONSTANT CURRENT INTO THE WHOLE PAIR.

2. STEERING THE CURRENT BETWEEN THE TWO LEGS OF THE PAIR.

So far, we have been discussing how to solve the first problem. It's time to consider the second problem...

Lutz, thank you for supporting... until someone else (e.g., Barrie Gilbert:-) helps you:-)

I have said this "But we want to pass the input current through them (e.g., to supply them by a current source from the side of the collectors), so the problem arises." as a hypothetical possibility; that's why I have written this "e.g." in the brackets. But really, if we try to "blow" the current from the positive rail to the collectors, we will provoke "current conflicts".

In the next imaginary experiment, I have suggested another situation leading to a "current conflict" - if we try to sink the current from the emitters and, at the same time, "floating" voltage sources are connected to the base-emitter junctions of the transistors...

Now about your last (very interesting) comment:

"In your last post you mention "two constant-current elements". Are you referring to the main transistors T1 and T2? Constant-current elements? Yes - without any differential input voltage. More than that, if you like to view these transistors as "voltage sources" - I have the following question: Between which two nodes are these voltage sources effective?"

Yes, I mean that the collector-emitter junction of any transistor with constant input voltage applied to its base-emitter junction behaves as a constant-current nonlinear element ("current source", as they usually say); so "the main transistors T1 and T2 are such constant-current elements" (if only their base-emitter junctions were driven by constant input voltages)....

But (in the common mode) their base-emitter junctions are not driven by constant input voltages; they are driven by the differences between the input voltages and the common emitter voltage, i.e., they act as negative feedback voltage followers (emitter followers, as they are usually named)... or as voltage sources...

Now about your question what are these sources and where their voltages are applied... Actually, one constant voltage source (the power supply Vcc) supplies the two voltage source outputs - the two emitters (but they are joined, with some purposet) via the collector resistors and collector-emitter junctions connected in series. So the output voltages of these voltage "sources" appear and are applied across the emitter resistor (or the emitter constant-current element, alias "current source")...

So, another "wisdom" (i.e., my insight:-) is that in this arrangement (long-tailed pair) in total three "sources" are connected in parallel - two voltage sources (T1 and T2) and one current source (the emitter constant-current element). Actually, there is only one voltage source and three non-linear elements - two constant-voltage and one constant-current...

It is extremely interesting to see how these three "sources" interact during the long-tailed operation (at common- and differential mode)...

Well, let's do it...

1. BIASING MODE. The emitter current source varies trying to set the desired biasing current. The two voltage sources cooperate and help the current source; the sum of their currents is equal to the input biasing current.

2. COMMON MODE. The two voltage sources cooperate and "move" side by side the common emitter point (change the common emitter voltage). Тhe emitter current source helps them keeping up a constant biasing current. As the voltage sources do not "strain" at all (they "see" or "think" they "see") an open circuit, the currents through them do not change.

3. DIFFERENTIAL MODE. The two voltage sources oppose each other and mutually neutralize the voltage changes at their outputs (an emitter virtual ground); the emitter current source "watches" passively and does not do anything (there is no reason to do anything). As the voltage sources extremely "strain", the currents through them vigorously change (are steered).

An impressive analogy: think of the emitter biasing current as an "electronic beam" and of the transistor pair as an electrostatic deflection system driven by the two input voltages... and imagine how the "beam" deflects to the left and right...

Good morning, Cyril.

Yes - now I could follow your explanations.

That means in short: The two transistors act as controlled voltage sources - as far as their emitter nodes are concerned. And they act, of course, as controlled current sources regarding their collector nodes, right?

Yes, Lutz... yes... exactly... It would be better if we had previously examined, in a similar way, the basic common-emitter stage with emitter degeneration as we can consider the long-tailed pair as a combination of two such more elementary circuits. Well, let's do it then... I know you do not like new questions...

We can think of it as of a network consisting of three resistors in series - in the ends, two steady ohmic (the emitter resistor Re and the collector resistor Rc) and in the middle, one variable non-linear (the collector-emitter junction of the transistor T). The current flowing through the whole network is the same and serves as a kind of a "transmission line" connecting the voltage drops across the two steady resistors; so Vc/Ve = Rc/Re. This current is set (regulated) by the middle element - the transistor, which sets the magnitude of the current proportionally, according to its transconductance, to its base-emitter voltage. So, if we apply the input voltage across the base-emitter junction, a voltage-controlled current source (VCCS) will drive the network and we can use the voltage drops across Rc and Re as two outputs of this amplifier... and even the voltage drop across the transistor as a third voltage output. This was the natural way of controlling the transistor...

But we prefer (for some reasons) more sophisticated artificial ways to control the transistor. For example, we can make it change its collector (emitter) current so that to keep the voltage drop across one of the resistors equal to the input voltage; then the other drop will be proportional to this voltage drop, accordingly to the input voltage. It is easier to do this trick with the emitter voltage since the output emitter voltage and the input base-emitter voltage are naturally (inherently) connected. So you just need to disconnect the input source from the emitter of the transistor and hook it to the ground (series negative feedback, emitter degeneration). As a result, the constant-current transistor begins behaving as a voltage source keeping the voltage across the emitter resistor constant... it is a current source that is made behave as a voltage source... But it continues doing this job by a current... and we can use this current as an output thus obtaining another but more perfect (due to the negative feedback) voltage-controlled current source... So, if we look at the transistor from the side of the collector, we will see another current source.

BTW with the same success we can make the transistor change its collector current so that to keep the voltage drop across the collector resistor equal to the input voltage; then the emitter voltage will be proportional to the collector voltage drop, accordingly to the input voltage. But as the collector is only an output terminal, maybe we should connect an additional op-amp to compare the two voltages and to drive accordingly the transistor base...

Very interesting transformations... We have taken an initial "output-current" device (not so good)... and, by applying a negative feedback, we have converted it to an "output-voltage" device (the emitter follower)... then, taking its collector current, we have converted it back to an output-current device (good)... and finally, taking the voltage drop across the other resistor, we have obtained an output-voltage device again (not so good)...

Now the most interesting... closely connected to our topic...

We can make the general conclusion that the main purpose of this negative feedback system (common-emitter stage with emitter degeneration) is to keep up the emitter voltage equal to the input voltage... it is a voltage (emitter) follower... a voltage source... and the emitter is the main but unused (internal) output of the circuit... The transistor does this work by regulating the common current. We can consider the emitter resistance as a kind of disturbance and the current (the voltage drop across the collector resistor) as a reaction to this disturbance... and the transistor does its work with less effort, the higher the resistance. So we can make a series of fun experiments with this common-emitter stage by replacing the resistive emitter load by more exotic loads and observing the reaction of the transistor.

If we replace the ohmic emitter resistor with a constant-current non-linear resistor (current "source"), the load becomes extremely "soft" and the transistor will control the internal "output" emitter voltage without any effort (current change)... the voltage drop across Rc will not change at all as well... and the gain will be zero. We exploit this idea in the long-tailed pair when working at common mode.

Conversely, if we replace the ohmic emitter resistor with a constant-voltage non-linear resistor (voltage "source"), the load becomes extremely "hard" and the transistor will exert all the forces (maximum current change) to generate the required emitter voltage... the voltage drop across Rc will vigorously change... and the gain will be maximum. We exploit this idea in the long-tailed pair when working at differential mode. But how do we implement this idea?

Here, we can simply replace the emitter resistor with a Zener diode. But in the case of a long-tailed pair working at differential mode we cannot do it since we want the pair works in a common-mode as well. So we want once the emitter load is "soft" (at common mode) and other times it is "hard". The solution is clever - to "harden" the "soft" constant-current load by connecting another constant-voltage element ("voltage source") in parallel... and this is the other leg of the pair. So, connect another transistor (with or without a collector resistor) and fix (or move in an opposite direction) its input voltage... the common emitter point will "freeze"... and become a virtual ground...

Sir plz tell me where i m publish my journal........how this process will be going on.......i m keenly interest to publish through online

Isha, I can't understand what you want to do - to upload your articles on RG?

Voltage steering is trivial. Any CMOS logic gate does that by connecting the output to one voltage rail or the other depending on the logic state. Current steering diverts the output of a constant current source. ECL logic does this. Many ECL gate designs steer the current between two differential outputs. The advantage of current steering is that it can be very fast.

I agree with you, Orin; CMOS gates are typical stages with dynamic load where the voltage drops across the two transistors crossfade.

It is interesting to explain also why current steering circuits are very fast - maybe because the voltage does not change during the switching... and therefore the stray capacitance does not delay the process?

Another simple example for voltage steering - similar to the CMOS arrangement - is the complementary combination of the two bipolar transistors for an A or AB push-pull amplifier.

Exactly! But maybe the simplest example is a potentiometer with moving slider - one voltage input, two complementary voltage outputs, constant current through it...

Current steering means to change the path of the operating current in analog circuit design.For implementation Gilbert Cell is used.

Gilbert cell is implemented using op amp.

Cyril, did you mean CURRENT MIRROR when you posted this ?

This is the most common way to mirror a current from one point to the other in an IC.

Example : Look inside a 741 OPAMP. You will see 2, 3 of these ...

The idea is : Ic=I0.exp(Vbe/Vt) ... So, if two BJT's have the same VBE, they will have the same Ic. I can use that to MIRROR the current ...

In your circuit above, I am using T3 almost like a diode (i.e., to SAMPLE the Vbe), since I don't really need the third terminal. But, I clearly cannot connect the B and C on the second transistor. This will short everything out !

This sounds great, but, it will be very SENSITIVE to temperature if you are making it out of discrete components, since the manufacturing tolerances and TEMPERATURE (i.e., Vt) will cause even a tiny thermal deviation to create very different Ic (exp is a very temperamental function !). However, it works great inside an IC, since the two transistors are

Hi Tolga! I do not mean the ubiquitous current mirror here; I have asked two questions about it before:

https://www.researchgate.net/post/What_is_the_basic_idea_behind_the_BJT_current_mirror_Is_it_a_reversible_device_can_we_swap_its_input_and_output_and_why

https://www.researchgate.net/post/Can_we_reverse_a_BJT_by_injecting_the_input_current_into_the_collector_and_taking_the_base-emitter_voltage_as_an_output

Thanks for the profound remarks.

Regards, Cyril

To breathe more life into the discussion, here are some fun experiments with LEDs as an example of the current steering technique; I carried out them with my students last week:

http://www.circuit-fantasia.com/ResearchGate/LED_current_steering.AVI

http://www.circuit-fantasia.com/ResearchGate/Lubomir/LED_current_steering_Lubomir.AVI

http://www.circuit-fantasia.com/ResearchGate/LEDTL.AVI

http://www.circuit-fantasia.com/ResearchGate/LEDTL_2.AVI

Enjoy! Cyril

Now I hope you can easily reveal the secret of my invention from 84's - a zero voltage LED indicator consisting of only two resistors (the base resistor can be omitted) , two transistors and, of course, three LEDs (one green and two red).

Here is how, in 2010, my students reinvented step-by-step this circuit on the whiteboard...

After that, they mounted the circuit on a prototyping board and investigated its operation...

http://www.circuit-fantasia.com/ResearchGate/Investigating_the_LED_indicator.MOV