Let us now turn our attention to the current control in field effect transistors. There are two types of FETs, the junction field effect transistor JFET and the metal insulator semiconductor transistor MISFET. The MOST is a type of MISFETs where the insulator is silicondioxide and the semiconductor is silicon.

In case of JFET the passage of charge carriers in the channel from the source to drain

is controlled by the gate to source voltage VGS. The gate to source junction is reverse biased such that it widens and this reduces the the depth of the channel. Consequently the drain current decreases and it can be made zero at the cut off VGS.

Since the the gate to source junction is reverse biased the gate current is negligible.

Therefore, the drain current in JFETs is voltage controlled.

The drain current in the MOSFET is controlled by a voltage VGS applied across the MOS capacitor above the channel. Here , one can invert the conduction type of the silicon and the density of mobile charge carriers in the channel by a voltage VGS on the capacitor. The DC gate current is zero since there is no DC conduction of the MOS capacitor. Therefore, the drain current is controlled by VGS.

For all types of field effect transistors one represents the I-V characteristic curves to reflect this fact. where the drain current ID is displayed as a function of VDS at different values of VGS.

In conclusion, the FETs are voltage controlled.

In an FET ideally there is no gate current, just a voltage across gate-source. So there is no flow of electrons into the gate. While in the case of a BJT, the current at the base leads to the collector current "magnification" Ic = beta*Ib. While in case of FET its about Vgs. I guess that is what they mean in terms of "current controlled" and "voltage controlled'. In the KVL analysis of a BJT circuit we directly use Ic = beta*Ib. Does your post intent to bring another insight into this?

When you analyze BJT mathematically then you can see the analysis on the basis of current equations like ic=beta*ib , is and many others and in case of FET the analysis is made in terms of voltages like vgs,vdd ...... The controlling parameter in BJT is current and in FET it is Voltage.

Vivek, as I can see, you assign the property "voltage/current" to the device input and your viewpoint is: there is input current - the device is current controlled; there is input voltage but there is no input current - the device is voltage controlled. But in the case of BJT there is both voltage and current. Then?

Regards, Cyril.

Premanand, is it still possible to control a BJT by a voltage and a FET - by current? I think it is...

Regards, Cyril.

The transistor effect is based on controlling the output current by a direct input current or voltage.The transistor is a three terminal device which is normally connected as two port network by taking one of the terminals of the transistor as common point to the other two terminals e.g., BE as input port and CE as an output port in BJT and GS as input port and DS as an output port in MOST.

Now, we are interested in the transistor effect and we start with the BJT and assume npn transistor. To see the transistor effect, we bias the transistor in the active mode of operation where the BE junction must be forward biased and the CB junction must be reverse biased. As the BE junction is forward biased it injects electrons from the emitter into base, where they cross the base by diffusion and large portion of it reaches the CJ where they will be extracted by the collector junction and reach the collector terminal. A very small part of the injected electrons recombines with holes in the base region. The base current delivers the back injected hole current from the base to emitter and the recombination current in the base region. The transistor is constructed to make the base current IB much smaller than than the collector current IC. AS the collector junction is reverse biased it own current ill be very small. THEREFORE, the collector current is originating from the emitter. The collector current is then completely controlled by the forward biasing of the EJ.

From the transistor theory we have:

IC= IEO(e^(VBE/VT) -1) +ICBO = IEO(e^(VBE/VT). eqn 1- very nonlinear

OR alternatively

IC= B* IB + ICBO = B* IB eqn2- linear

The parameters in the equations have their usual meaning.

we see from equation 1 that the collector current increases exponentially with the emitter to base voltage VBE. While equation 2 shows that the collector current increases linearly with the base current. When we speak about control, we wish that the relation between the controlled parameter and the controlling parameter be linear.

Therefore, it is agreed among the scientists and engineers that the bipolar transistor is current controlled. Therefore in representing the I-V characteristics we represent the collector current as a function of VCE at various values of the base currents and not at different values of VBE.

In my next comment, i will speak about the control action in the field effect transistors.

Let us now turn our attention to the current control in field effect transistors. There are two types of FETs, the junction field effect transistor JFET and the metal insulator semiconductor transistor MISFET. The MOST is a type of MISFETs where the insulator is silicondioxide and the semiconductor is silicon.

In case of JFET the passage of charge carriers in the channel from the source to drain

is controlled by the gate to source voltage VGS. The gate to source junction is reverse biased such that it widens and this reduces the the depth of the channel. Consequently the drain current decreases and it can be made zero at the cut off VGS.

Since the the gate to source junction is reverse biased the gate current is negligible.

Therefore, the drain current in JFETs is voltage controlled.

The drain current in the MOSFET is controlled by a voltage VGS applied across the MOS capacitor above the channel. Here , one can invert the conduction type of the silicon and the density of mobile charge carriers in the channel by a voltage VGS on the capacitor. The DC gate current is zero since there is no DC conduction of the MOS capacitor. Therefore, the drain current is controlled by VGS.

For all types of field effect transistors one represents the I-V characteristic curves to reflect this fact. where the drain current ID is displayed as a function of VDS at different values of VGS.

In conclusion, the FETs are voltage controlled.

As Prof. Cyril has stated the BJT can be represented both as Voltage controlled Current Source or as Current Controlled Current Source (Which is the normal version we are familiar with).

It is based on Two Port Representation of the BJT and not based on Input Source.

The Voltage controlled Current Source may explain the action of the Transistor as a Switch.

Kindly refer the Lecture Notes

Lecture 20 Georgia Tech

Bipolar Junction Transistors (BJT): Part 4 ECE 3040 - Dr. Alan Doolittle

Small Signal BJT Model

given in the link:

http://users.ece.gatech.edu/~alan/ECE3040/Lectures/Lecture20-BJT%20Small%20Signal%20Model.pdf

Dr.P.S.

Is there a linear relation between the voltages across the base of a BJT and the output (either voltage or current) in the collector (assuming CE config)? There is a linear relation between the collector current and base current though.

The characteristics between Vbe and Ic will be almost the same as that of a diode obeying Shockley diode equation.

The different characteristics can be viewed in the link below.

https://coefs.uncc.edu/dlsharer/files/2012/04/C4.pdf

The same input Circuit as given by Prof.Cyril can drive the transistor into all the three regions,cut off with high negative bias of base emitter or saturation region with high Positive biasing or into active region with proper Forward biasing just at or slightly above cut in Voltage of the Base Emitter Diode.

This is achieved by having proper value for Rb. High value drives it into active region and small value and proper biasing makes the transistor work as Switch.

Dr.P.S.

We should not confuse ourselves with the circuit model of the transistor. If one looks only at the physics of the device, a BJT is current controlled while a FET is voltage controlled. The confusion in the circuit picture and in real world applications is that we often use voltage sources (like batteries and opamp outputs) to drive circuits. The voltage applied across the base-emitter junction of the BJT causes a current to flow across the junction and leads to carrier flowing from collector to emitter without recombination without recombination in the base. The carriers flowing through the collector is much more than that flowing through the base and hence, we get a current magnification. The BJT will still work if you are able to force current into the base-emitter junction with no voltage across the base-emitter terminals. The voltage appears because the current is driven through passives such as resistors which is in every material except for superconductors. From this, one can clearly see that a BJT is current controlled.

A similar argument goes for FETs. The voltage on the gate of the FET controls the physics inside the FET. Again, the confusion with current control in MOSFETs is that we can charge the gate capacitance of the FET using a current source. Presence of charge on the gate capacitance result in some voltage across the gate capacitance and that controls the electrostatics in the transistor channel.

Once we can account for these issues, you can easily model the transistor as voltage controlled or current controlled in the circuit of interest. However, at the heart of these devices, BJTs are current controlled and FETs are voltage controlled.

I was absent for a long period - therefore, my late answer.

I cannot agree to the last contribution - as explained below:

1.) For answering the question if the BJT is physically current or voltage controlled it does not help at all to rely on formulas (like Ic=beta*Ib) which are used in practice. They do not say anything about the physical truth.

2.) Quote X. Fong: "The voltage applied across the base-emitter junction of the BJT causes a current to flow across the junction and leads to carrier flowing from collector to emitter..."

Comment: The current across the junction is NOT the physical cause of the collector current. It is just a (small) current that cannot be avoided - however, it is NOT the cause of the collector current.

It is a general physical law that a small quantity never can control DIRECTLY a larger quantity of the same nature.

3.) Quote X. Fong: "The BJT will still work if you are able to force current into the base-emitter junction with no voltage across the base-emitter terminals".

Comment: This is nothing else than a claim without any justification (by the way: a claim without any physical background - thus, it is without any sense).

4.) Quote X. Fong: "From this, one can clearly see that a BJT is current controlled".

Comment: No, it is NOT. Are you able to explain the function of a current mirror or the effect of a stabilizing emitter resistor in a BJT amplifier using current control?

LvW

First, I thank Prof. von Wangenheim for his comments and important points which is critical in this discussion.

Yes, it is important to keep in mind the first point Prof. von Wangenheim mentioned. The equations used for BJTs and MOSFETs are meant as a model for the physical effects rather than to explain the physical effect itself, as Prof. von Wangenheim mentioned in 1). It seems my answer to the question was not clear in explaining my understanding of the BJT. Let me now elaborate more clearly on points 2) & 3), which then justifies 4).

On point 2), most textbooks on fundamental semiconductor physics and devices will describe the the physics of a pnp-BJT purely from a carrier transport perspective. It will then be helpful in this discussion to refer to Ch. 10 of Semiconductor Device Fundamentals by R. Pierret. In the description of BJT physics, the base-emitter current itself is critical to the function of the BJT. The sequence of events describing homojunction BJT operation typically goes as:

1) additional majority carriers are injected into the base, putting the base out of equilibrium,

2) minority carriers (in reference to base) are diffuse in from emitter (since emitter is much more highly doped than collector), which try to recombine with the holes.

3) since the base is very thin and narrow, the minority carriers are swept out of the base through the collector (due to the electric field from reverse biasing the junction) before they can recombine with the holes.

On face value, one can accept then that BJTs are current operated. However, if can also be argued that voltage also can result in a similar sequence of events: the forward biasing of the base emitter junction may be achieved through voltage control, which controls the relative Fermi levels in the emitter and the base. I believe this was the point raised by Prof. von Wangenheim in 4).

It is then helpful to think of a different device that has the same structure and consider its operation instead. This brings me to point 3), where Prof. von Wangenheim argued that injection of current into the base without forward biasing the base-emitter junction is physically impossible, since it looks as if I am not injecting energy into the system to bring it out of equilibrium. To clarify my point, I am not suggesting that no energy is injected but rather the energy injected is not in the form of electrical energy. To help illustrate this point about energy injection, it is helpful to look at the phototransistor, which is widely used in optocouplers and many other optoelectronic circuits.

Except for dimensions and materials, the structure of a phototransistor is essentially identical to a BJT. The sequence of events in its operation is exactly the same as what I have described above. The difference is in the way minority carriers are injected into the base. In the case of phototransistors, photons injected into the base may cause generation of minority carriers. Thus, phototransistors are typically made from direct bandgap semiconductors. The injection of light energy into the base gets converted to electrical energy through transistor action. Note that in the case of a phototransistor, there is no requirement for the base-emitter to be forward biased using a voltage source. The phototransistor current is proportional to the light intensity incident on the base, which is a quantity describing the number of photons incident on the base per unit time. Thus, we can think of the photons as a current, since each photon can only generate one minority carrier in the base (absorption of photons is a quantum process).

To quickly summarize, my claim here is that the physics of transistor action in BJTs is current controlled since it is initiated by the injection of minority carriers into the base (resulting in the sequence of events 1-3 as stated above). I then support my claim by looking at two devices (the BJT and the phototransistor) where the structure is essentially identical and only the physics of carrier injection into the base is different. The physics of transistor action in both devices are essentially identical. I hope my explanation above clarifies the points put forth by Prof. von Wangenheim.

Kelvin Fong

A BJT is a current controlled device because its output characteristics are determined by the input current.

A FET is voltage controlled device because its output characteristics are determined by the Field which depends on Voltage applied.

Hello Mr. Kelvin Fong,

thank you for your long and detailed answer. I think, the RG forum is not the right place for going further into details of the carrier movements within the BJT.

However, if the BJT is really controlled by the base current, it should be possible to explain the principle of all transistor circuits based on this "assumption".

1.) Is this possible in case of the simple two-BJT-current mirror?

2.) How do you explain Barry Gilbert`s translinear circuits?

3.) And what about Shockley`s well-known formula? Is it still valid?

Regards

Lutz vW

PS: "A BJT is a current controlled device because its output characteristics are determined by the input current." (Quote Farid Kadri).

That means: It is current-controlled because it is current-controlled. Aha!

I have to disagree that it should be possible to explain all transistor circuit operation based on this assumption. My believe is that the circuit picture is also what causes the confusion in the first place. The reason is that in circuits, current and voltages are tightly coupled. One wouldn't be able to explain how a phototransistor works just by looking at the circuit, since there is no "source" for the light. In all the applications mentioned by Prof. von Wangenheim, it is easier to understand the operation using voltages because it is also easier to model the circuit as such. In all our models, we can always introduce a transducer mathematically to reduce the problem to something easier to solve but that does not always represent the physics happening within the device. My opinion is that the classification of BJTs and MOSFETs should depend on the physics going on within the device rather than on the circuit operations.

"The reason is that in circuits, current and voltages are tightly coupled."

Yes, agreed - of course. And that is the reason for different descriptions/explanations in the several textbooks.

" My opinion is that the classification of BJTs and MOSFETs should depend on the physics going on within the device rather than on the circuit operations."

...rather than? For my opinion, it should always be possible to explain circuit operations based on the "physics going on in the device". It may be another question if this approach always is the best and simplest way - however, it should be possible. Otherwise, some doubts may arise...

PS1: What about Shockley`s equation?

PS2: What about the general and natural rule that a small physical quantity never can control (dominate over) a larger quantity of the same nature?

Dear Lutz,

I am interested in your rule that "a small physical quantity never can control (dominate over) a larger quantity of the same nature" but I have a problem with understanding it. Is this a law of energy conservation? Is it mean that a current cannot control another current by means of a current-controlled resistance?

Regards, Cyril

Prof. von Wangenheim,

Shockley's diode equation relates the current in a pn-junction to the voltage applied across it. Are you then proposing that the BJT be modeled as two tightly coupled pn-junctions? I think your point is that the base-emitter current is voltage controlled and hence, the BJT should be considered voltage controlled. But like in the example of the photodiode (which may be considered as a very poor BJT) I put forth, it isn't the only way of achieving transistor action since the transistor action depends on putting the base-emitter junction out of equilibrium. So maybe it is more satisfying to say that the BJT is both voltage and current controlled depending on the application?

It would also be good if you elaborate more on PS2 because that statement, in its current form, goes against the idea that we can have gain > 1, which to the best of my knowledge, is achievable.

Kelvin

Cyril,

I have expected such a question. I must confess, it was - more or less - a provocation.

I have "created" this rule. (I remember: You like provocative questions/statements).

However, I am convinced of this rule since IMO it sounds logical - unless, you have one or more counter examples.

Lutz

Important remark: I refer to DIRECT control only.

"Direct" means "without the mediation of a controlled resistance"?

Prof. von Wangenheim,

It may be more enlightening if you provide an example of what you consider as indirect control (perhaps a simple resistive inverter).

If I may make an attempt in putting that in perspective with respect to the BJT: the "gain" between collector current and base current is an indirect control and not direct control. The collector current comes simply because the in-build electric field due to doping between base and collector sweeps the carriers out of the base before they can recombine. The field is there regardless of the presence of carriers in the base. However, one still needs carriers to enter the base so that current can flow to the collector, and that is the "input" we control. Now, you might ask then why would there be so much more current since there seems to be more work done than energy put into the system. The answer to that lies in the fact that 1) there is thermal energy from the surroundings to excite carriers, and 2) we have a power source that drives the current around the circuit through emitter and collector. 1) is obvious and one of the reasons why an emitter resistance is needed to limit the gain of the class A amp to prevent thermal runaway. Unfortunately, flow of electrons in the diffusive limit may get obstructed and cause electrons to dissipate energy in the form of heat. The heat generated excites more carriers in the emitter, which in turn increases the amount of carriers that flow through the BJT. In that sense, current control in BJTs do not violate that "rule" since it isn't direct control.

Kelvin

Hi Kelvin, nice discussion - although I know it is an old discussion (as mentioned at the beginning of this topic by Prof. Cyril Mechkov).

I can follow your statements, however I have some questions

1.) "The collector current comes simply because the in-build electric field due to doping between base and collector sweeps the carriers out of the base before they can recombine."

Question: You speak about "the carriers".

Where are they coming from ? What is the physical reason for their existence?

What quantity determines the amount of these carriers?

2.) "However, one still needs carriers to enter the base so that current can flow to the collector."

Question: ....so that current can flow to the collector? Can you explain why a base current is a precondition for the existence of a collector current?

Lutz

Prof. von Wangenheim,

Yes, it is interesting to have this discussion from several different perspectives - at the circuit level and at the device level and I thank all who posted in this thread for their valuable insights.

I have already mentioned the different possibilities of getting the carriers into the base: 1) a voltage source to forward bias the base-emitter junction, and 2) a source of photons to excite the carriers in the base made from direct bandgap materials. But the cause of the intrinsic property (transistor action) in BJTs is due to the carriers. There can be several ways of getting the carriers in there but this is an extrinsic property of the device. My concern is that once we say a BJT is voltage-controlled, it becomes difficult to justify that a phototransistor is not voltage-controlled (since their structures are identical).

The physical explanation for 2) is rather long so instead, I refer you to texts that discuss the physics of phototransistors.

There is another way that I look at this, and it's a very simple minded way that first/second year electrical engineering students may approach this: if we forward-bias the base-emitter junction, ideally we cannot control the current using voltage. Once the base-emitter (BE) voltage exceeds the threshold voltage, the BE current increases exponentially with voltage (as Shockley's work already shows) and hence, from a circuit standpoint, the BE voltage doesn't change. It might change a little but for all intensive purposes, it stays constant due to the exponential dependence. One way we can improve the input voltage range to give us a better "control" is to introduce an input resistor, as Prof. Mechkov mentioned at the top of the thread. But this resistor is external to the transistor and so we shouldn't consider it as part of the BJT. If the BE voltage stays constant, how do we then justify that the BJT is voltage-controlled?

Kelvin

Kelvin, thank you for the valuable thoughts. Your contribution is extremely useful for the discussion. It seems there are no simple things in this world. So, in this connection, it would be even useful to discuss why we prefer to deal with voltage than current in circuitry-:)

IMO the base resistor enlarges the input voltage region but decreases the overall circuit gain. Other alternatives are just to keep the input voltage small (within hundreds millivolts as it is in the common-emitter and common-base amplifying stages) or to apply a negative feedback - series (the emitter follower) or parallel (the input part of the current mirror or the "transdiode" log converter).

So, to control a BJT directly with a (small) voltage, is a reliable arrangement. The only problem is how to hold the input operating point on the steep part of the base-emitter IV curve.

Regards, Cyril

Hi Kelvin,

I must confess that I am not able to understand the following part of your answer:

"Once the base-emitter (BE) voltage exceeds the threshold voltage, the BE current increases exponentially with voltage (as Shockley's work already shows) and hence, from a circuit standpoint, the BE voltage doesn't change. It might change a little but for all intensive purposes, it stays constant due to the exponential dependence."

The BE voltage remains constant - and the current increases exponentially?

To me this sounds contradictory. Did I misunderstand something?

Lutz

Prof. von Wangenheim,

Sorry for not being clear the last time round. When I made that comment, it was in reference to what is being taught to undergraduates in their first class dealing with active circuits. Usually the ideal diode is first taught as an ideal device where it's I-V characteristic is 0 < Vth and vertical at Vth (off when voltage is below threshold and like a voltage source at Vth, but only prodiving forward current). The simplification is taught as an approximation to Shockley's diode equation. In essence, as long as the bias condition is such that the pn-junction is ON, any amount of forward current can flow.

Also, I want to point out that device people usually consider only the base of BJT and gate of FET as inputs (at least that is the trend I notice). People more involved with circuits usually look at the circuit topology first before labeling terminals as input and output.

Kelvin

Kelvin, thank you for clarifying this point - however, I don`t think this will help us to solve our "problem".

Concentrating on the main subject (the BJT without any external elements like resistors) I like to point out the following:

1.) It is the thickness of the insulation layer (and it`s variation) between the emitter and the base region that is the primary source for Ic variation.

Do you agree to this statement?

2.) Which electrical quantity is able to influence the thickness of the insulation layer?

For my opinion, obviously a BE voltage change only.

3.) Of course, this BE voltage variation is connected with a corresponding base current change. But this is an (unwanted) byproduct only and has no controlling function.

4.) Because each voltage (and it`s variation) is connected with charge separation we could also say that the BJT is a charge-controlled device.

__________________

Final question: Can you agree to all 4 points? In case you cannot agree, please give corresponding explanations. Thank you.

Regards to you.

Lutz vW

Prof. von Wangenheim,

Thank you for bringing up these points. Firstly, let me say that I disagree with some of the questions you raised (not that the answer to those questions are 'no' but that they should not be raised in the first place) and I will state my reasons after I respond to the four points. Because the points are closely related, it is difficult for me to comment on each one independently of the others, but I will try my best to to it.

I refer you to page 380 of Semiconductor Device Fundamentals by R. Pierret (I have the International Edition printed in 1996) for the reasoning behind my responses.

1) I assume by Ic variation you mean the the range of Ic given the input, and not due to uncontrollable sources such as process variations. Thickness variation is the unwanted effect in BJTs. Answer continued in explanation for response to 2).

2) Yes. This is correct. But my reply to 1) shows why this point is irrelevant. The reason why heterojunction bipolar transistors were "invented" was to get around this problem, since base width modulation in homojunction BJTs is difficult to solve by just emitter-base-collector doping ratios.

3) Yes and no. Yes because the base current "degrades" the effectiveness of the BJT. Ideally, we want the BE current to be entirely dominated by the current due to majority carriers in the emitter so that Ie = -Ic. And no because both carriers contribute to current in the real BE junctions, and one can equivalently connect a current source to provide the base current, which results in a corresponding voltage drop across the BE junction. The relation between the current and voltage drop is governed by Shockley's equation, as you already mentioned. The real unwanted effect is thickness modulation that you mentioned.

4) From a circuit standpoint, this is correct.

As I mentioned at the beginning of this reply, I think it is not right to argue that the control in BJT is due to the variation of the depletion regions between BE and between BC. It determines the quality of the BJT but is not the reason for transistor action. The concept of modulating current through control of channel depletion is more closely related to JFETs, and MOSFETs in some ways.

- Of course, in MOSFETs we are interested in only when the channel close to the oxide-channel interface gets inverted but it will get depleted first before being inverted.

Phototransistors were invented because inventors understood two concepts:

1) BJT is driven by flowing carriers due to drift and diffusion mechanisms controlled by carrier population in the respective terminals, and

2) the carrier population can be manipulated easily by photons in direct bandgap materials.

The reason I keep bringing the phototransistor example up is that if indeed the BJT is a voltage controlled device, one will then also have to explain how phototransistors work by voltage-control.

Kelvin

Hi Kelvin,

I think my answer can be relatively short.

* You are referring to a textbook. I would have no problems referring to other books/documents which support my opinion. Did you already see the original document from W. Shockley (US Patent Office, 1951)?

To 1) Yes, of course. Since we are speaking about Ic control I do not refer to any uncertainties or unwanted effects due process variations.

To 2) Irrelevant? The thickness variation of the insulation layer (depletion zone) is the REASON for Ic variation, is it not? What do you think makes Ic to vary with a controlling signal? Remember how a pn diode works and why Id=f(Vd).

To 3) The "thickness variation" is an "unwanted effect"??? (see also 2).

Regarding Photo-transistor:

I think, this device can be seen as a "normal" BJT with a Photodiode in parallel to the CB path, OK?

Thus, there is a path between C and B similar to a biasing resistor Rb in some conventional BJT circuits. Thus, the phototransistor behaves like a simple BJT biased by a voltage, which is determined by Vcc (collector node) and Rb.

This is my explanation how phototransistors work by voltage-control.

Lutz vW

I often get the question we are commenting, this is how I answer :

In a linear, static behavior,

If it is current controlled, then the output is directly linked to an input current (in a BJT we have : IC = beta x IB)

If it is voltage controlled, then the output is linked to an input voltage (in a MOSFET we have ID= gm x VGS)

with beta and gm parameter solely related to the component (not external resistors or else)

Prof. von Wangenheim,

I understand your explanation of phototransistor operation but what you have proposed is a mathematical model. The physics is directly related to quantum processes which is easily modeled using currents and hence, does not require a mathematical transducer to convert to voltage. This is exactly the confusion between circuits and device that I spoke of when dealing this subject.

On the topic of pn-junction Id=f(Vd). It can be explained directly using statistics of carriers moving over the conduction/valence band barrier. As Vd changes, the number of carriers that have enough energy over the barrier changes exponentially and hence, Id ~ exp(Vd). This reasoning also explains the temperature dependence of Id.

Kelvin

OK Kelvin - it seems we come to an end of this discussion.

Of course, I agree to the second part of your last answer. And because the BJT has a pn junction between B and E - similar to a pn diode - we have identified the control mechanism:

Diode: Id=f(Vd)

Transistor: Ic=f(Vbe).

Thank you for participation.

Lutz vW

Right, Prof. von Wangenheim. I am glad we can agree on this point. But I think we have not touched on the topic of transistor action of BJT. We have only concluded that to get carriers into the base, we have the option of either using voltage or option of using photons. Are you then saying that the transistor action is due to voltage?

If temperature is taken into account... what self heating will produce in a voltage driven BJT (VBE is kept constant by a stable voltage source) ?

Is the output of a current driven BJT more stable when the base is driven by a constant current ?

Hi Dominique,

Sorry but I am not understanding your question on the temperature part. The self heating from the emitter current may result in more carriers getting injected into the base. The increased current heats the device more and provides a positive temperature feedback and could result in thermal runaway, even at constant voltage.

Back on the topic of the thread, the diode equation, Id=f(Vd), only provides a relation between current and voltage, not the control mechanism. I can very well rewrite Vd=g(Id) where g=f^-1. To illustrate what I mean, if I apply a voltage source across a diode, Id is the current the source must supply; if I apply a current source to pump through the diode, Vd must be the voltage across the diode. From a circuit view, the source I use is the control. But the non-linearity within the device did not originate from the source I use.

Kelvin

The answer to the initial question could be : in theory and at constant temperature a BJT can be driven either by a voltage or a by current, both being linked by the equation of the diode.

But in practical circuits, temperature is never fully controlled, and thermal runaway is likely to occur when a VBE voltage is used to control the current of a BJT, unless some kind of feedback is used to stabilize the operating point. If one uses a current source to control the BJT current, the output current is much more stable, less sensitive to variations of the input and temperature. This is why some people say : the BJT is current driven.

"The answer to the initial question could be : in theory and at constant temperature a BJT can be driven either by a voltage or a by current, both being linked by the equation of the diode."

The initial question was if the BJT from the physical point of view is volzage or current controlled. Here - the physics allows one answer only.

This has to be distinguished from the technical question of control.

Both the answers from Profs. von Wangenheim and Bergogne are good. Perhaps the better term to use, as Prof. Bergogne suggests, is "voltage driven" vs "current driven", which seems more indicative of the underlying theory.

Yes, very good comment from X. Fong.

I would like to return the discussion to the circuit domain with a question that adds arguments in favour of the voltage controlled transistor. Why in some cases they connect a resistor (R2 in the figure below) in parallel to the base-emitter junction?

Regards, Cyril

Some days ago, I have mentioned a similar argument in favour for voltage control:

How can you explain the effect of a stabilizing emitter resistor based on the (false) assumption of current control? I got no answer. And the same applies to the function of a two-transistor current mirror. No answer.

Finally, a question to all who still believe the collector current of the BJT would be current-controlled:

How do you explain the tanh transfer curve of the classical differential amplifier made from BJT`s (long-tailed pair)?

This is an answer to Cyril.

In the schematic shown, if we add a third resistor, R3, between the collector and a positive voltage source, E, the circuit is surely aimed at working in the switching mode. That is, the transistor is either saturated (VCE close to zero and the collector current is close to E/R3) or non conducting (VCE = E, no collector current).

R1 is used to determine the base current, assuming a voltage source, Ectrl, is used to turn the transistor on. R2 diverts some current, VBE/R2, Here is your ANSWER : R2 is used to accelerate the turn-off of the transistor. Why ? R2 can be made smaller than R1, so when the control voltage Ectrl, returns to zero, the electrical charges stored in the base are provided with a low resistance path, R2//R1, accelerating the recombinations inside the transistor, therefore, accelerating the turn-off. As R2 diverts some base current, R2 must be chosen to have a negligible effect of the current set by R1.

If the Ectrl is capable of sinking a current, it is then more efficient to use a diode D + resistor R circuit placed in parallel to R1, in that case, R can be made as small as needed.

Some people also place a resistor between the control electrodes of any transistor for safety : in the event of an open circuit on the control side, this resistor insured a non-conducting state of the transistor.

To Lutz, if I may call you so.

Many good questions in your answer!

The Current Mirror case. As you have guessed, I am a believer of current control. And I have given some of the reasons earlier. In a current mirror, two transistors share the same VBE and therefore forcing a current in the first collector will set an identical current in the second collector, hence the name "mirror". So we can conclude that the second transistor is voltage controlled. If you try to build a current mirror, without cheating with additional resistors, you will find that to get a perfect mirror the transistors must have identical Base-Emiter junctions and also, that the temperature of both transistors must be identical. This is achieved in integrated circuits and widely used.

It is possible to drive a BJT by applying a fixed voltage to the Base-Emiter junction, the current mirror is a good example, but many constraints have to be solved.

To Dominique (if I may call you so),

thank you for responding.

my understanding of your contribution is as follows:

You are a "believer" of BJT current control - however, you mention some arguments in favour of voltage control, right?

BTW: Per accident I have found a short statement regarding BJT control from Winfield Hill (co-author of "The Art of Electronics"), see here:

http://cr4.globalspec.com/comment/720033/Re-Voltage-vs-Current.

Regards

Lutz vW

Yes, I 'm a believer and I also try to have a scientific thinking, this is why I dare making conclusions that do not serve my belief. I will think over your question and over the reference you give. I out for a break now, see you next week.

"I also try to have a scientific thinking, this is why I dare making conclusions that do not serve my belief."

I appreciate such an attitude - and I try to do the same.

Thank you

Lutz

Dear Lutz,

This morning, I pondered over the role of the humble R2 resistor connected across the base-emitter junction. I don't know why but, every time when I see such an "elegant simplicity" - a resistor connected in the base, in the emitter, in the collector, between the base and emitter, between the base and collector, etc., I am deeply impressed-:) It is maybe because this is the possibly extremest simplicity (only one element) but the phenomenon behind it is not so simple as it looks...

Yes Cyril - indeed.

And the situation is even more complex as it seems because there is not only one task R2 has to fulfill:

* D. Bergogne has mentioned one for switching applications : "As R2 diverts some base current, R2 must be chosen to have a negligible effect of the current set by R1".

* However, in case of operation as a linear amplifier, R2 has to provide the necessary BE voltage - and its value must be be chosen as a trade-off between some conflicting requirements.

Lutz, I continue the story about R2... I would like first to make a reservation that I consider here the last situation depicted by Dominique - "...in the event of an open circuit on the control side, this resistor insured a non-conducting state of the transistor"...

So far we argueed if the BJ transistor is voltage- or current controlled. And we finally concluded that it can be controlled either by voltage or current. But note that in both the cases we supposed to work in the useful vertical part of the input IV curve where it is switched on or it is in active mode. But what about the useless horizontal part of the IV curve? It is important when drive the n-p-n transistor in the figure above by another but p-n-p transistor (the Dominique's last situation). The p-n-p transistor actually drives the n-p-n transistor by current; it acts as a current source. For example, when it cuts off, it tries to cut off the n-p-n transistor by decreasing its base current... and this is the problem...

In the horizontal part, the transistor has an extremely high input resistance; so it should be controlled by voltage (figuratively speaking, the BJT behaves here as a FET-:) In contrast, in the vertical part it has an extremely low input resistance and should be controlled by the current. But note, in the useful vertical section we vary the input quantity along the whole region while in the useless horizontal section we are interested only to stay in one point - the origin of the coordinate system; we want just to cut off reliably the base-emitter junction.

The resistor R2, acting as a current-to-voltage converter, solves this problem - it enables us to control the n-p-n resistor entirely by current (along the whole IV curve) as follows. In the useless horizontal section, the resistor converts the current into voltage. In this region, the transistor has high input resistance (behaves as a FET) and it is controlled by voltage as it would "like". When we approach the bend of the IV curve, the transistor changes sharply its input behavior - decreases its base-emitter resistance and shunts the resistor R2. Now it would "like" to be controlled by current and we (the p-n-p transistor) drive it by current along the whole vertical section...

These considerations are valid as well for the measurement arrangements used in the semiconductor devices laboratory when measuring the diode IV curve since it has the same nonlinear (horizontal + vertical) IV curve. If we decide to drive the diode only by current, we have to shunt it by a resistor (R2) converting the current into voltage during the horizontal section. Thus we will obtain enough points to plot the curve in this part. Of course, there is a trade-off solution - to drive the diode simply by voltage through a resistor...

It was my story about the humble resistor connected between the base and the emitter-:)

Regards, Cyril

Hi,

I think

Semiconductor devices works on voltage or current .The method of connection of source with control resistance is important.Bipolar junction transistor is controlled by controlling base spread resistance.This base resistor can be controlled by excitation at base. If we apply voltage source in series with source resistance(external) to the base.The base current is responsible create a drop of VBE =0.7 (Si BJT) at base emitter junction. Since this is cause of device to work hence it can be referred as current controlled device.

But in case of FET usually we connect voltage source VGG at input or gate terminal (in CS configuration) which is always connected in parallel with RG which is very high value (Mega ohms). This cause to have Vgs drop at FET gate source terminal to work FET. Hence it is called Voltage controlled device.

I wl suggest to try simulating simple DC biasing circuit with BJT and FET you wl certainly notice difference ou sources and its control to device.

Thnaks and regards

Ohh - what a long discussion about such a basic question.

But it could be also caused by a misunderstanding.

Let me explain: I see a fundamental difference between the questions

(a) how the BJT internally works (how the Ic is CONTROLLED physically), and

(b) how external parts (resistors) DRIVE the base of the BJT.

And that leads to my claim: We have not a choice at all how to CONTROL the BJT either by voltage or current. It is simply the physical reality that determines this property.

However, of course we can use a very large base resistor Rb in conjunction with a relatively large voltage with the aim to mimic a current source. In this case, the classic formulation applies: We inject a base current into the base and the whole circuit can be called "current driven".

BUT: This does not influence physical laws, which means IMO that the BJT still works voltage controlled. In this case, the resistor Rb forms - together with the base resistance - a simple voltage divider that creates the necessary base voltage Vbe that physically CONTROLS the current Ic.

To Cyril:

I have some problems to understand some parts of your last contribution.

What do you mean with "vertical" and "horizontal" parts of the input IV curve?

In particular the following sentences needs some explanation:

(1) "In the horizontal part, the transistor has an extremely high input resistance".

(2) "in the vertical part it has an extremely low input resistance and should be controlled by the current".

______________

The input resistance depends on the selected operating point on the Ib=f(Vbe) curve and normally has maximum values of 10...20 kohms. (extremely high?).

Thank you and regards

Lutz vW

Lutz, we can roughly present (approximate) the p-n IV curve by two sections - an almost horizontal (slanting), starting from the origin, and an almost vertical (steep). Within the first part, the transistor has high input resistance while, within the second one, it has low (differential) resistance. Imagine we want to see what a particular diode IV curve is (a classic experiment in the laboratory of semiconductor devices). Then, to easily obtain sufficient number of points (to vary calmly the input quantity): in the first region, we should apply a voltage across and measure the current through the diode; in the second region, we should pass a current through and measure the voltage across the diode. "Geometrically speaking", this means the curve of the input source to be perpendicular to the IV curve in the particular section.

Thank you for the distinguishing between the two viewpoints - "physical" and "circuit"; it was necessary.

Regards, Cyril

OK, Cyril.

It was not clear to me (I was a bit confused) that you spoke about measurement techniques. Initially, I thought you would consider the BJt as voltage-controlled for very low base currents as well as current-controlled for larger currents ("vertical part").

To make my point clear:

The physics of the BJT requires a particular voltage to influence/change the width of the depletion area (insulation layer) and, thus, the collector current - independent on measurement techniques and external circuitries that create this BE voltage (even in case the base seems to be "current-driven"). I use quotation marks because, in reality, we drive the base not with a current but with a voltage source. But this voltage source is connected to the base node via a resistor that is considered as large if compared with the BE resistance (normally, in the low kohm range).

Regards, LvW

... And this is the problem of this current control since, to cut off reliably the base-emitter junction, we have to apply low (zero) voltage...

....and where is the problem you expect?

(By the way: I am pedantic and, therefore, claim that this is a current-drive rather than a current control).

The problem of the BJT in the "horizontal" part of its IV curve is the same as the problem of a FET when trying to zero its gate voltage by increasing the resistance connected between the gate and Vdd...

My students have sometimes a similar problem when trying to change the voltage across a voltmeter (an open circuit) by a variable resistor (a rheostat) connected between the voltmeter and the power supply... or the voltage of an op-amp input by changing the resistance connected to this input-:)

"Current drive" characterizes the external "current-driving circuit" when "current control" characterizes the physical nature of the internal base->collector control?

Regards, Cyril

Cyril - yes that is my understanding of the difference between "drive" and "control".

Regarding the "problem" in case of very low currents Ic and Ib, respectively:

In this case, I never would recommend a "current drive" biasing scheme.

I think, the good old voltage divider at the base node will do the job much better.

Lutz

Maybe it is well to name the "horizontal" and "vertical" parts accordingly "constant current" and "constant voltage", "high resistive" and "low resistive", or "the first" and "the second"? I suggest this since frequently there is a need to talk about some of them.

I must confess that - up to now - I never have seen the necessity to allocate different names to "both parts" of the characteristic curve Ib=f(Vbe). To me, there are not two "parts", but it is rather an exponential function, which for larger currents is linearized due to an ohmic resistance.

You may remember that the terms "horizontal" resp. "vertical" have caused some confusion on my side. And the same applies to the terms "constant current" and "constant voltage", respectively. To me, this is an over-simplification of the real characteristic, which will cause further confusions.

In which cases you feel the need to divide this curve in "two parts"?

Above I needed it to explain the role of R2... Really, "constant current" is a little misleading since the curve begins from the origin; actually, it is a curve of a high ohmic resistance...

http://www.newworldencyclopedia.org/entry/diode#Semiconductor_diodes

First, sorry for this long answer.

Thank-you Lutz for the reference to the book,

The Art Of Electronics - 2nd Edition

Publication Date: July 28, 1989 | ISBN-10: 0521370957 | ISBN-13: 978-0521370950

may I advise Cyril to have a look at this book, now he has gained much insight for the topic :)

and Chapter 2- Transistors , with my comments,

The physics of the BJT implies that to keep a current flow in the collector, electrons must flow continuously thought the base, hence, my belief : the BJT is current controlled.

However

That means that in the real world, the current amplification property ( little current input => large current output) of the BJT is not practical to use, at least in an open loop circuit (no feedback).

But Unless there is some king of stabilization effect (or feedback) in the circuit, like in a current mirror.

Finally

also

Therefore, for analog circuit analysis, the BJT is a voltage driven device (but the physics inside relies on a continuous flow of electron through the base --a current-- to control the collector current.)

For switching circuits analysis, the current drive approach is more practical. Then modeling the switching BJT is much more complicated.

In conclusion: thank you for the opportunity to look-back on BJTs and try to explain basics in a simple way. In the past years I have been using transistors as switches and I had forgotten the analog analysis point of view.

Hello Prof. D. Bergogne,

here are two of your sentences that sound contradictory:

"The physics of the BJT implies that to keep a current flow in the collector, electrons must flow continuously thought the base, hence, my belief : the BJT is current controlled."

"Therefore, for analog circuit analysis, the BJT is a voltage driven device".

Please, can you clarify this?

Thank you and regards

Lutz vW

To Cyril

Looking for arguments about BJTs I found this document:

http://www.eecs.berkeley.edu/~hu/Chenming-Hu_ch8.pdf

witch is a good complement to the Art of electronics on BJTs.

Prof. D. Bergogne,

thank you very much for pointing our attention to this very detailed and valuable document. Isn`t it funny and surprising that so many years after inventing the BJT it seems to be still necessary to "fight" for the physical truth (voltage vs. current controlled)?

Lutz vW

Remark:

For me it is the first time to see a publication that - with clear words - states the truth about the base current (page 297, 3rd line):

"IB is an undesirable but inevitable side effect of producing IC by forward biasing the BE junction."

To Lutz

I am preparing a direct answer to your request for clarification, which I will publish later.

Why "fighting" over the BJT today, for "truth" ?

Well, I think this question addresses the human being. I think that the "truth" is a personal thought about the perception of the world one's have, and the more people share your truth the more it is... true. Of course, science is a great help in providing tools and methods to verify if "things are true".

In this world of intense access to information, human beings need to talk ("fight") to get some reality and be reassured.

Talking about me and this discussion on BJTs, I am not a specialist of semiconductor devices but I knew enough to get on the discussion, then I got interested by the exchange, and carried by the controversial aspects... Finally, I got to read about BJTs, the last article being: http://spectrum.ieee.org/semiconductors/devices/how-europe-missed-the-transistor .

To Prof. D. Bergogne:

(Quote:"Why "fighting" over the BJT today, for "truth" ?")

I fully agree with your opinion that "human beings need to talk (fight) to get some reality".

Perhaps I have expressed myself not clear enough:

It was my intention to emphasize only the fact that such a discussion ("fight") takes place 60 (sixty!) years after inventing and using the BJT.

In this context, I like to mention that I appreciate your new topic about current sources. I think, it is really important to clarify what this thing called "current source" really is.

Yes, I got your idea : 60 years should be enough for a knowledge to be fully spread.

Lutz, I am not sure what aspects of current sources you want to clarify but I have asked a general question about the "philosophy" of the current source:

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

Regards, Cyril

Dominique, thank you for the advice; I have the "bible" of H&H and really, I use it as an "idea stimulator"...

Thanks Cyril. Don't forget to add some Berkeley mix in it. Practical electronics are based on theoretical stuff and sometimes it helps.

Prof.Lutz von Wangenheim and Prof. D. Bergogne have given good explanation.I learnt from books of well known authors 40 years back that BJT is current device as the control is based on Drift and Diffusion currents of respective carriers and EM model is developed on that . Even present day authors like JocobMillman, Razvi, Sedra and many explain the principle and all calculations are done using formulaes on minority current. MOSFET is the voltage device as control is based on the voltage and field.

It was a good topic as as we struggle to make our students understand this basic funda

To Mr. D. Suratkal:

May ask you a question:

What are you going to teach the students regarding the question that was extensively discussed in this columne: Is the BJT physically voltage or current controlled?

Dear Prof.Lutz, I am happy to be in this discussions. In every application or circuit of BJT, we analyze the currents on the applied voltage and calculate the outputs in terms of voltage. That does not mean that BJT is a voltage device.Even in case of simple Op Amp using BJT we analyze the currents. The confusion still exists (when I take applications of BJT) with many of my students, because basics of Diode, BJTs are not dealt properly. Hence physically, operationally and process tells us that BJT is a current device.

To Mr. D. Suratkal:

Thank you for your answer.

However, may I direct your attention to the following document (from Berkeley University):

http://www.researchgate.net/go.Deref.html?url=http%3A%2F%2Fwww.eecs.berkeley.edu%2F~hu%2FChenming-Hu_ch8.pdf

Among other arguments and justifications it contains the following sentence:

"IB is an undesirable but inevitable side effect of producing IC by forward biasing the BE junction."

Summary: The BJT undoubtly is a physically voltage-controlled element.

However, this applies independent on the practical usage, which often exploits the simple formula Ic=beta*Ib. This formula is helpful but says nothing about the physical truth.

Regards

Lutz vW

Dear prof. The several advantages of MOSFET or FET as a voltage controlled device including the less amount of noise as compared to BJT the attractive option.The popularity of the FET as voltage controlled device will increase and o f course the ease of chip design using MOSFET are another attractive feature. The BJT characteristics clearly shows that the output current is the function of input current Ic is the function of Ib at constant voltage so we take all the current calculations in amplifier. The FET has the property that drain current will change as we change gate voltage so the voltage controlled device.

Dear Mr. A. Khadke,

I don`t know if your reply is an answer to my former contribution - however, may I ask you to physically justify/explain the following sentence?

"The BJT characteristics clearly shows that the output current is the function of input current".

Thank you and regards

Lutz vW

Hi everybody here! I would like to "revive" the discussion (although someone may think that 76 answers and 4555 views are enough:) by asking the questions, "What does 'voltage controlled' and 'current controlled' still mean? Is this a property of the load or the source?" Here are my speculations.

There are three kinds of sources (constant-voltage, constant-current and real) and three kinds of loads - voltage (high-resistive) input, current (low-resistive) input and real (moderate-resistive) input, and they can be connected in the possible combinations below:

constant-voltage source --> voltage-input load

constant-voltage source --> real-input load

real voltage source --> voltage-input load

constant-current source --> current-input load

constant-current source --> real-input load

real current source --> current-input load

The next combinations below are inadmissible since they lead to conflicts:

constant-voltage source --> current-input load

constant-current source --> voltage-input load

Thus, "voltage/current controlled" may mean that a real load is controlled by a constant voltage/current source that establishes the desired (by it) quantity. It is usually implemented by a negative feedback system keeping up the desired quantity. So, in this case, "voltage/current controlled" is a property of the source output; it is sooner "voltage/current controlling". Or, it may mean that a perfect (voltage/current-input) load is controlled by a real voltage/current source. In these situations, the perfect load provides ideal load conditions for the imperfect source so the desired (by the source) quantity is kept again with the same success. So, in this case, "voltage/current controlled" is a property of the load input. But what does "perfect load" ("ideal voltage load" or " ideal current load") mean?

The ideal voltage load has infinite resistance (open circuit) or infinite differential resistance (constant-current dynamic resistance) - an example is the transistor (collector, drain) output. But it is still interesting if we can say "the transistor output is voltage controlled"...

Conversely, the ideal current load has zero resistance (short circuit) or zero differential resistance (constant-voltage dynamic resistance) - an example is a diode (ordinary, Zener, LED, etc.) So, all sort of diodes are voltage-controlled devices...

These were some my thoughts about the issue...

Hi Cyril - to me, it is a good idea to resume this question again.

It was YOU who has opened this thread - and I am not completely sure about your position regarding the main question (BJT - voltage or current-controlled?).

In this context, your last contribution is really helpful because it can help to clarify at first the question: What does it mean to be "controlled"?

For my opinion, we have to distinguish between the two basic cases:

* What kind of control signal?

* What happens physically?

Let me explain these two cases using the BJT example:

* We can inject a current into the base and, thus, change the outout current Ic (and it seems we control the BJT with a current), however

* physically spoken this is NOT true. The BJT is a voltage-controlled device - and in reality the injected current causes a voltage Vbe that controls the thickness of the depletion layer which, in turn, influences the current Ic.

Thus, the BJT is a voltage controlled device - however we can use, of course, a current for this purpose (exploiting the beautiful law of G.S. Ohm)

Hi Lutz! Nice to see you again here! To see again my friends here, in analog electronics area, I have temporarily left the "kingdom" of the "analog digital electronics":) where there is no such a great interest in fundamental ideas (circuit, logic, educational, whatever...)... even in the most fundamental digital idea you can imagine, "How do we implement logic gates?"

I will add new thoughts about this issue after a few hours since now I have to meet my friend (and my former student from the late 80's) Dilian from Sweden - a very interesting person who supports me and gives me strength in my educational circuit endeavors. I hope we will reveal some of the fundamental digital ideas during our reunion...

https://www.researchgate.net/profile/Dilian_Gurov/

Lutz, really analog electronics is more interesting than digital one...

You are absolutely right that we have to distinguish between two basic cases; I would say "between two parts of this arrangement" (source - transistor). It seems, "voltage controlled" and "current controlled" terms refer only to the transistor input (the base-emitter junction) but actually, they make a connection between two points of the arrangement.

In the first case, they make a relation between the source and the transistor input. I have described just this case above and you have described it as "We can inject a current into the base and, thus, change the outout current Ic (and it seems we control the BJT with a current)"

In the second case, they make a relation between the transistor input and its output (the collector current). You have described it thus "physically spoken this is NOT true. The BJT is a voltage-controlled device - and in reality the injected current causes a voltage Vbe that controls the thickness of the depletion layer which, in turn, influences the current Ic."

The first viewpoint consider the input interfacing problems while the second - the internal (physical) relation between the input/output transistor quantities. In the first case we are not interested in the very transistor; we treat it as a short connection (when driving with a high voltage through a resistor) or as a non-linear resistor when driving directly with a small voltage... we think of it as of a current load (zero resistance) or a voltage load (moderate resistance)... we do not interest how it further processes the input quanity...

It is interesting to see how in the circuit of a transimpedance amplifier the voltage-controlled op-amp is artificially made act as a perfect current-controlled device. For this purpose, we have connected a resistor between the op-amp output and inverting input. The op-amp "sucks" current through this resistor from the input source thus creating an illusion of a current-controlled input...

Yes - A really good example. Put the whole circuit into a black box (without the voltage source resistor, of course) - and you have the illusion of a CCVS.

Back to the BJT: Some people might argue if it is important to know whether the BJT is VC or CC. My answer: There are some applications (current mirror, emitter degeneration, translinear circuits...), which, indeed, require this knowledge.

These designs are based on the VC feature.

Thank you, Lutz! I continue reasoning about this issue... and would like to add some more thoughts...

I think we can control the bipolar transistor both by voltage and current just because its input is not a perfect ("ideal") current load (i.e., "a piece of wire" or a constant-voltage non-linear element with an absolutely vertical IV curve). If it was perfect, we couldn't apply and change an input voltage (by a perfect voltage source) to its base-emitter junction...

In this connection, it is interesting to do such an experiment with the transimpedance amplifer above - to (try to) drive it by a perfect voltage source. This is my favorite question that I ask to my students in the analog electronics course...

@Dr. V. Surducan,

In case you have read all the contributions in this thread you will not be surprised that I am not with you regarding the PHYSICAL control mechanism of a BJT.

(I don`t want to repeat my arguments again, which were posted here on the 4th., 6th, 9th and 13th of May)

In particular, I like to point again to the Berkeley paper (as given from Prof. Bergogne on May 13th.)

* Question: Which electrical quantity influences the thickness of the depletion layer (Vbe or Ib), which in turn controls the current Ic?

I am aware that "ideal things do not exist in electronics".

I think, exactly this is the reason for an unwanted by-product (Berkeley) that is called "base current".

Of course, I agree also that there is no current without voltage and vice versa.

HOWEVER, the main question is whether (as in our case) the current is necessary to provide a control function or if it just flows because it cannot be avoided.

And what about the classical triode valve? Current or voltage?

Do you really consider such a question as "useless"?

Regards

Lutz vW

Since I do work with DC-DC converters, this question always comes up as to which one is better to use ? BJT ? or MOSFET ? as a switching element. This is very related to your question.

BJT's are current-controlled devices. This means that, Ic=hFE*Ib. So, if you stop Ib ... Ic stops. This is a problem when it comes to energy (or power) efficiency, since you do not want to always supply current to control a device, which means that, you always have to burn power to control the device.

MOSFETs, on the other hand, are voltage controlled. Once you turn them ON, they are on forever ... Well, to prevent the noise, a small pull up resistor is fine, which burns minimal power.

So, MOSFET always wins ? NO ! If you look at the input capacitance of the MOSFET, it is typically 5x, 10x higher than a BJT. This means that, you need a very strong gate drive to turn on a MOSFET, whereas, you can drive a BJT with a much smaller current. As an example, BJTs can be turned on with 100mA, and can drive a 10A load (hFE=100), while a MOSFET would need a 2A surge drive for a short period of time, and no sustained current after that once it goes beyond Vthreshold. This implies a necessary GATE DRIVER for a MOSFET, which is something that BJTs do not require up to much higher currents (if at all)..

Let's calculate the total energy we burnt for a 20KHz operation with a duty cycle of 50%:

Assuming the Vbe=1V to turn on such a powerful BJT. Assuming 2A gate drive for the MOSFET, and a 2nF total gate capacitance, and Vthreshold=3V.

CASE 1:) BJT will need the gate drive sustained for the entire ON period, which is

50% * 50us * 100mA * 1V = 2.5uJ

CASE 2:) MOSFET will need enough energy to charge up the TOTAL gate capacitance of 2nF to 3V ... E=1/2CV^2 = 1/2*2E-9 * 3^2 = 9nJ. Also, you possibly need a gate driver for such a high drive, which could burn another 5, 10 nJ. That's still 15-20nJ.

As you see, current controlled devices need more than two orders of magnitude energy to turn ON/OFF the load. This is why BJTs slowly but surely lost their place as a switching element.

I would only add that "there is no voltage without current and current without voltage" is not always valid (if you mean a "voltage drop across" and a "current through" a 2-terminal element)...

A fancy story about analogies making a bridge between the well-known old and the unknown new...

About the relation between voltage (drop) and current... There is a voltage across an element with infinite resistance without current flowing through it, and there is a current through an element with zero resistance without voltage across it. There is both voltage and current across/through elements with some ohmic/negative resistance, and across/through elements with zero/infinite differential resistance...

@ Tolga Soyata

Quote: "BJT's are current-controlled devices. This means that, Ic=hFE*Ib. So, if you stop Ib ... Ic stops. "

I must confess that I am somewhat surprised because of this simple statement. WHY? Because it is a claim without ANY justification or explanation. Just a statement. More than that, it simply neglects all (serious!) sources showing that and why the BJT is PHYSICALLY voltage controlled. The mentioned relation Ic=hfe*Ib is just an equation saying NOTHING about cause and result. (See the impressive contribution from Cyril Mechkov about the transimpedace usage of a voltage controlled opamp).

Everybody knows the role that a voltage of app. 0.6 volts plays in a pn diode.

Is it per accident only that the same voltage plays a similar role in a BJT? No, of course not. In both cases the depletion layer is influenced by this voltage in a same way.

@ Vasile Surducan

Quote:"An engineer which is forced to design something quickly (his project must run from the first revision) will never ask himself this question (even in the Berkeley terms). He will notice that a BJT is more a current than a voltage driving device...".

I cannot fully agree to this. Such an engineer is really a bad engineer.

A good engineer who is designing a circuitry involving BJT`s must always consider BOTH parameters (Vbe as primary and Ib as "by-product").

When the BJT must work in it`s linear region it must be biased with app. 0.65 volts. That is, for example, the only purpose of the resistive divider at the base node.

For a rough design he even can NEGLECT the base current. When he is using an emitter resistance the resulting error is acceptable in many cases.

As I have mentioned already before, some other applications only work because of the voltage control feature of the BJT (current mirror, emitter degeneration using Re, translinear circuits,..). Their operating principle cannot be explained and verified using only Ic=hfe*Ib.

Lutz, here is even a more impressive viewpoint at the ubiquitous transimpedance amplifier. It is based on my favorite Miller theorem:)

https://en.wikipedia.org/wiki/Miller_theorem#Implementation

In the circuit of a transimpedance amplifier, we have connected a virtual resistor with zero resistance in parallel to the voltage-controlled op-amp input. This composed virtual resistor consists of two "resistors" in series - a "positive" ohmic resistor R and a true negative resistor (the op-amp output producing a "mirror" voltage Vout = -Vr = -I.R). Thus, the total resistance of this network is zero... the voltage-controlled op-amp input is shorted by this "piece of wire":)... and behaves as a current-controlled input... Figuratively speaking, all the transimpedance amplifier circuit (including the power supply) is just a "piece of wire":)

Cyril - but this works (as described by you) for an idealized opamp only (gain infinte).

Correct?

@Lutz,

When I say "BJTs are current controlled, and MOSFETs are voltage controlled ?" let's open it a tiny bit. I think MOSFET is pretty clear. So, I will focus on opening up why "BJT is current controlled."

*** Our starting point is the two equations: Ic=I0 * exp(Vbe/Vt) and Ic=hFE*Ib.

*** I can utilize these two equations in two main cases :

*** CASE 1: USING Vbe AS INPUT :::::

As we see, for very small Vbe changes (once you are around, say, the 0.6V - 0.7V bias region)... Ic can be approximated as a linear formula (i.e., small signal model). So, about a biasing point of Vbe, call it Vbe0, the behavior is linear:

Ic ~ Io*(1+Vbe ) ... If I {AC couple through a capacitor} the Vbe as the INPUT VOLTAGE and convert Ic to a voltage through a resistor (Re ), Then, I get the formula : Vout=GAIN*Vin.

So, in this case, the input is a voltage, output is a voltage. I used the BJT as a gain amplifier: This is what I will call the ANALOG use of BJTs. In this case, the transistor is working as a voltage controlled resistor, not as a SWITCH element .

*** CASE 2: USING Ib AS INPUT :::::

Now, when you want to use BJTs as switching devices, things change. Our starting point is precisely the two formulas above again ... But, this time, I do not care about the LINEARITY. I do not care about harmonic distortion. All I care about is an EFFICIENT SWITCHING OPERATION. I want the BJT to work in the most efficient two operating points for switching : a) completely OFF, b) completely ON.

Once the BJT turns on above, say, 0.7V or so, the current is going to increase exponentially from 0.7V, to, say, 0.75V. So, My Ib is the only thing that will LIMIT the amount of Ic I can draw out of this BJT.

This phenomenon is completely different than MOSFETs, where, once the MOSFET turns on, you do not need to keep pumping current into it.

Once the BJT turns on, since, the Vbe is pretty much irrelevant above Vbe>= 0.7V or something, I control what I can draw from Ic through Ib. So, my control parameter is Ib. This is why I call BJT a current-controlled device. This is not something I made up !

Lutz, we often come to this point... and you always ask that question... and I always hint that there is a "magic" way to make the circuit perfect (to set the virtual ground really at zero voltage). Okay, let's see how... this is only my speculation provoking our creative reasoning...

As you have noted above, in the classic circuit of a transimpedance amplifier, even when the op-amp gain is extremely high, the voltage of the inverting input is not exactly zero; it has some small value (Vout/K) with the same polarity as the input voltage. This is because the op-amp has not managed to "pull" the summing point enough to the opposite direction... and it has not managed to do it because of the insufficient (not infinite) gain... and, as a result, its output voltage is less than the voltage drop across the resistor... We cannot do it by further gain increase since it has to reach the inifinity... So, the abilities of this approach are depleted... we have to try something else... something clever...

My idea is somehow to make the op-amp increase its output voltage so that it becomes exactly equal to the voltage drop across the resistor. IMO we can do it by introducing a small positive feedback in addition to the existing negative one... and this will bring us to the powerful VNIC idea...

To implement this clever idea, we should connect a voltage divider R1-R2 between the op-amp output and its non-inverting input (i.e., as a positive feedback network) and adjust its transfer ratio so that to zero the virtual ground. At the same time, there is a negative feedback network (R and possibly Ri of the input source) connected between the op-amp output and its inverting input. The negative feedback dominates over the positive one, so the circuit is stable (it works in the linear mode). The small positiwe feedback only compensates the insufficient op-amp gain by adding the needed voltage to the op-amp output voltage...

https://en.wikibooks.org/wiki/Circuit_Idea/Linear_Mode_of_Voltage_Inversion_NIC#Middle_negative_resistance_region

This fancy story can be used by well-meaning teachers as one possible way to reinvent the odd VNIC with the purpose of explaining/understanding the exotic circuit soultion to/by their curious students... I share it for the first time here...

Hi Tolga! I really like your thoughts as a way of thinking and highly appreciate them!

As I can see, when considering the two cases (Vbe or Ib as an input), you look at the transistor input (the base-emitter junction) from the view of the input source (as you said, " So, my control parameter is Ib. This is why I call BJT a current-controlled device."); "my" has a meaning of "input source's". This is the first case of my and Lutz's explanations above. I think this viewpoint (looking at the transistor input from the side of the input source) is most useful from a practical (engineering ) viewpoint. The second viewpoint (looking inside the transistor to see what causes the output collector current) is maybe more useful for the specialists in the area of microelectronic devices...

Your remarks about the big initial input MOSFET current are very interesting, attractive and probably, unsuspected for many people. But I am not sure if we should consider them here in this issue, since implicitly we have assumed to consider only the static mode of the transistor operation (we can with the same success consider the role of the BJT base capacity). Maybe, your sentence "...once the MOSFET turns on, you do not need to keep pumping current into it..." will puzzle some people...

Also, I do not understand why, when considering the current control, you confined only to the switching mode of the BJT...

@Tolga, thank you for your answer, however - again - I must admit that I cannot follow you.

* At first, I do not think that the physical principle of the transistor function can be different and will depend on external conditions .

* Secondly, regarding switching operation of the BJT:

I cannot see that "Ib is the only thing that will LIMIT the amount of Ic I can draw out of this BJT" (end of quote).

Why do you think that Ib sets the limit?

I suppose you are referring to the classical switch configuration with a collector resistor Rc. In this case, the maximum value of Ic depends on Rc - which means on the position of the working line within the Ic=f(Vce) characteristic. When Ic exceeds a certain value the voltage across Vce goes under the saturation limit and the base-collector diode starts conducting. I really cannot see how Ib sets any limit.

Again, your formulation: "you ... need to keep pumping current into it" neglects the influence of the voltage Vbe. There is no "need" to "pump" a current Ib into the device. The current Ib cannot be avoided, unfortunately. That`s all.

Where is you PHYSICAL explanation (based on depletion layer thickness) for the Ib control property?

* In the theoretical case without any collector resistance, we have a similar situation.

The current Ic increases with rising Vbe until the transistor crashes for thermal reasons - but not because the Ib has reached any limit.

Or did I misunderstand any part of your contribution?

Lutz, think of it this way: You could have a Porsche that can go 320 kph. However, nothing is preventing you from using that car for a) grocery shopping, vs. b) a race car.

In my comments above, I am comparing the same exact BJT, but, using it in two different operation modes. Refer to this for my TERMINOLOGY:

http://en.wikipedia.org/wiki/Bipolar_junction_transistor

CASE 1 : BJT IS A LINEAR RESISTOR (i.e., trans-resistor, i.e., TRANSISTOR, where the name came from). In this case, power efficiency is not your primary concern, but, rather, having a nicely linear relationship between Vbe and Vout. Of course, the transistor will have a linear relationship between Vbe and Ic. You will turn that into a linear Vbe vs. Vout by using the Re resistor (or Rc, whatever). The transistor is in Forward-active mode here ... And, the current is an irrelevant discussion, so, I am controlling the output voltage through the input voltage ...

CASE 2 : BJT is a SWITCHING ELEMENT : In this situation, you do not care about the linearity of the input voltage vs. the output voltage. You want the output voltage to change between two extremes, and you want that switch to happen as fast as possible. Because of that, you have a TURN ON VOLTAGE (e.g., 1V) and TURN OFF VOLTAGE (e.g., 0V). This makes the input voltage an irrelevant parameter, in a sense. The BJT is working in either CUT-OFF mode, or Saturation mode. At these two extremes, since the voltage has very minimal control at the switching behavior, another CONSTRAINT pops up: THE MAXIMUM OUTPUT CURRENT YOU CAN DRAW IS LIMITED BY THE MAXIMUM INPUT CURRENT YOU HAVE TO PUMP INTO THE BJT. Although Ic=hFE*Ib, it is an illusion to think that, this is a linear relationship. hFE is heavily dependent on Ic due to the non-linearity of the hFE vs. Ic curve.

So, to get the semantics right, "current controlled" might not be the right terminology, but, the idea is simple : You are using the BJT as a current amplififer in this case, rather than transconductance amplifier as in the first case ...

This operation is very similar to how a current-limited voltage source works. Say, you set the output voltage to 5V, and the current limit to 1A. Up to 1A, the voltage source regulates the output voltage ... Once you reach 1A, at that point, you start working as a current source ... and the voltage doesn't matter, s long as it doesn't exceed 5V.

BJT is identical. Due to the exponential relationship between Vbe and Ic, once you reach 0.7V, 0.8V, etc. after that point, current is all that matters, and I worry about controlling the current ...

The base current (or the voltage or the base resistor) can set (limit) the collector current only in the linear mode when the transistor is not saturated or "crashed". In this mode, the transistor output acts as a constant current source (not so good but still a current source). The collector resistor is not mandatory; its only function is to convert the collector current variations into collector voltage variations...

Lutz, I agree with your assertion, "I do not think the physical principle of the transistor function can be different and will depend on external conditions". They do not depend on the external conditions (the driving circuit part). Figuratively speaking, the input source (circuit) and the transistor can "think" in a different way - the input source can "think" that it controls the transistor by current while the transistor "thinks" that it is controlled by voltage:)

It is interesting to see why, in the switching mode, the collector resistor sets the maximum collector current; I like to consider this situation with my students. For the purposes of understanding, it is useful to think of the transistor as a varying resistor (nonlinear but still a resistor)... and this "rheostat" begins decreasing its static resistance to increase the collector current. I repeat it to my students - "the only thing that a transistor can do in this situation, is to decrease its resistance... nothing else..." Finally, its "resistance" becomes zero and only the collector resistance remains in the circuit... only it can continue controlling the collector current...

But I do not stop here; I continue "provoking" my students by asking the next question, "How can the transistor continue controlling (increasing further) the current?" The answer is, "Only if it contained an internal varying source; if it was a true negative resistor... like the VNIC above":) Of course, the source should be "helping", i.e., connected in the same direction as the power supply when travelling along the output loop (- + --> - +)...

Tolga - I am sorry, but you cannot convince me.

* The relation Ic=hfe*Ib applies under small-signal conditions only (hfe is a small signal differential parameter). Thus, you must not use it at all for explaining BJT operation as a switch (on mode).

* Quote: "THE MAXIMUM OUTPUT CURRENT YOU CAN DRAW IS LIMITED BY THE MAXIMUM INPUT CURRENT YOU HAVE TO PUMP INTO THE BJT."

Here you repeat an assertion - again without any justification. Can you give any reference or any formula or any physical explanation for this claim?

* I ask you again: Which circuit are you speaking about? With Rc or without?

* What about my counter arguments? Sorry to say, but for my opinion you are repeating your arguments - again without any consideration to my answers, questions and counter arguments. Sorry - but this is my impression.

Lutz vW

Quote Cyril: "The collector resistor is not mandatory; its only function is to convert the collector current variations into collector voltage variations..."

Yes - of course, it is not mandatory.

However, in context with the question "which effect limits the current Ic" it is important to know if we have Rc=0 or not.