Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Why is FET voltage controlled and BJT current controlled?

  1. Sep 7, 2009 #1
    A bit awkward question but still it confuses me......

    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.

    Now the question is that current is also generated due to Voltage and still BJT is current controlled and FET voltage controlled.

    Any Ideas?
  2. jcsd
  3. Sep 7, 2009 #2
    A transistor is a non-linear device, and so we use different models to describe its behavior. The current control model is used often because it is approximately linear. Another models like the Eber-Molls model is used to address the voltage control behavior. It is non-linear because of the base-emitter junction acts like an exponential, so different approximations are used to linearize it.
  4. Sep 7, 2009 #3
    They are both voltage controlled, or if you prefer they are both current controlled.

    On bipolar transistors, the voltage varies with the temperature, but the current amplification stays constant.

    On FET transistors, the voltage amplification stays constant, but the current varies with temperature.
  5. Sep 7, 2009 #4
    So it is just our preference as to which model we choose?
  6. Sep 7, 2009 #5


    User Avatar
    Science Advisor
    Gold Member

    Last edited by a moderator: Apr 24, 2017
  7. Sep 7, 2009 #6
    Yes, it depends on the problem you are solving. In general, whenever there is current flowing, there will be a potential difference. So if you increase current flowing in the base, you have to increase the base voltage. If you increase the base voltage in an NPN, and due to constant base-emitter diode voltage drop, a proportional voltage will increase above internal emitter resistance and thus pulling more collector current.
  8. Sep 7, 2009 #7
    Current is not generated "due to voltage". This cannot be overemphasized. Current and voltage are generated in unison. One cannot exist w/o the other. In a p-n junction, such as the b-e junction in a bjt, the change in base-emitter voltage, vbe, cannot take place until after the change in base current, ib. The reason a bjt is called a current controlled device goes back to the Ebers-Moll 1954 paper.

    In the E-M model, the bjt is modeled as a pair of back to back diodes with the base being the common anode. Each p-n junction, the c-b and the b-e, is approximated by Shockley's diode equation

    Id = Is*(exp(Vd/Vt) - 1).

    The b-e and c-b jcns have differing doping densities, so the "Is" in a diode, scaling current, is denoted by "Ies" for the b-e jcn, and "Ics" for the c-b jcn. Hence,

    Ic = alpha_n*Ies*(exp(Vbe/Vt) - 1) - Ics*(exp(Vbc/Vt) - 1).

    Ie = Ies*(exp(Vbe/Vt) - 1) - alpha_i*Ics*(exp(Vbc/Vt) - 1).

    For Ic, the collector current, if alpha_n, the normal mode emitter to collector current gain, is zero, we only get leakage current due to reverse biased c-b jcn. The "-Ics*(exp(Vbc/Vt) - 1), is this current, the 2nd term in the Ic equation. Since Vbc is negative and near 0, we get "Ics*(0 - 1)" or just simply "Ics".

    But the Ebers-Moll model adds a current-controlled current source as follows:

    Ic = alpha_n*Ie. When the emitter injects electrons into the base, rather than recombine in the base region as a diode does, the base is so thin, the electrons are drawn into the collector region due to the strong electric field in the c-b jcn.

    Thus the collector current originates from the emitter current. The same electrons which make up the emitter current travel into the collector and become collector current.

    The alpha factor describes "transistor action". If alpha is 1, we have a perfect device, Ic = Ie, as all electrons survive the trip through the base, and no holes are injected from the base to emitter. We approach this condition by using light doping in the base and heavy doping in the emitter.

    If alpha is 0, there is no transistor action, and we just have 2 back to back diodes. The base current Ib, and base-emitter voltage, Vbe, are still present, but Ic = 0. Without high alpha values, Ic is small and transistor action is poor.

    Ic is controlled by alpha and Ie, but Ib and Vbe are necessary quantities. Since the b-e jcn must be forward biased, Ib and Vbe are absolutely necessary to obtain Ie. Without Ib/Vbe, Ie would be 0, and Ic = 0, regardless of alpha.

    Ib and Vbe are necessary , whereas Ic and alpha are both necessary and sufficient to create Ic. If the base region was very wide, and all electrons recombine in said base, alpha is 0. We have Ib and Vbe, but since alpha = 0, Ic is only leakage current Ics. We have Ib and Vbe but no transistor action as alpha = 0. Ic is NOT controlled by Ib or Vbe. Ic is controlled by Ie and alpha.

    We control Ic with Ie.

    With FETs, the gate current is necessary as is the gate to source voltage. But controlling gate current is too hard. The gate current, Ig, needed to produce a specific Id value, drain current, varies with frequency as well as temp and bias current value. Also, a FET presents a very high g-s impedance. A current source at the g-s terminals would charge the non-linear capacitance to the point where punch-through occurs. A voltage source at the g-s terminals charges this capacitance to a safe value. Vgs is used to control Id. A FET does not lend itself to current control. Voltage control must be used for driving FEs. Both are necessary, one is controlled directly, the other is indirect.

    The bjt is controlled via Ie, with Ib and Vbe being indirect. Sometimes, Ib is used as controlling current, such as saturated switch applications. The base is overdriven to fully saturate the collector-emitter. We never drive the b-e terminal voltage directly, but rther indirectly.

    The FET uses direct voltage control, with Ig indirectly defined. A voltage source is connected from g-s, and the resistances and capacitances determine Ig.

    Is this helpful?

  9. Sep 8, 2009 #8
    i think so. thanks a ton.
  10. Sep 10, 2009 #9
    Someting I don't understand is, since Ies is such a small number, how does the Ebers Moll model accurately describe the emitter current? I mean, it's possible to have transistors with many milliamperes or amps flowing through them, and I don't see how you get those numbers when you're multiplying everything by 10^-16 or some such.

    Edit: Also, if you have a chance - how would the above equations be different when the transistor is not in the active region, but in saturation with both the base to emitter and base to collector diode forward biased?
    Last edited: Sep 10, 2009
  11. Sep 10, 2009 #10
    Good questions. Regarding the Ies value of 1e-16 to 1e-11, here it is. For a signal device, if Ies = 1e-15 (1.0 femtoamp), with Vbe = 0.65V, and T (temperature = 298 K (25 C), then Vt = nkT/q, where n = 1.0 in this region of operation, k = Boltzmann's constant = 1.3806e-23 joule/Kelvin, & q = 1.602e-19 C (charge on 1 electron), so that Vt = 0.0257 V, or 25.7 mV.

    So if alpha = 0.99 typ, then Ic = 97.19 uA. If Vbe = 0.60V, then Ic = 13.87 uA. For a heftier device, if Ies = 1e-12 (1.0 pA), then at Vbe = 0.60V, Ic = 13.87 mA, etc.

    The Ies values given in the fA & pA range are accurate in the operating region of uA and mA of collector current. The reason is as follows. The thermal voltage, Vt = nkT/q, is proportional to absolute temp, and "n", a factor which varies as current changes.

    When Ic is in the uA to mA range, n is very close to 1. But at low values of current, below a uA, n increases, eventually reaching 2 in the nA region. This is explained in semiconductor physics. In this region, the E-M equations are not as accurate.

    The exponential/logarithmic relation between I & V in a p-n junction, holds for about 5, or maybe 6 decades of current for a bjt, and up to around 10 decades for certain diodes. With a bjt there are 2 phenomena affecting the relation.

    The first was just mentioned, the variation of "n" at low current values. Another is the c-b junction leakage current. Let's say that Ics is 1.0 nA. Even though the b-e jcn value of leakage, Ies, is 1.0 pA, the Ic minimum value is 1.0 nA. Even though the b-e junction could be at 0 V, there is always the leakage from c to b due to Vcb & Ics. This spoils the log-linear relation between Ic & Vbe, as well as Ie & Vbe. The E-M equations account for the c-b leakage Ics, with the 2nd term. This is why the log-linear function for I vs. V does not extend down into the pA or fA region.

    But, if we hold the Ic value in the uA to mA region, or even amps for a power device, using an Ies value in the fA to pA range, produces good results. The slope of the I-V curve is linear from let's say 100 nA to 10 mA of Ic, or 5 decades of current. If Ies is 1.0 fA, that is way below the straight line region of the I-V curve. Below 100 nA of Ic, the curve flattens out. It is not log-linear way down to 1.0 fA at all. So why can we use the 1.0 fA value if the Ic-Vbe graph does not linearly extend down that low?

    The answer is that a line drawn from Ies = 1.0 fA upward extends into the nA, uA, & mA region with good accuracy. From Ic = 100 nA until 10 mA, the lines nearly coincide. Below 100 nA, the actual Ic curve is greater than that predicted by E-M based on Ies = 1.0 fA.

    But we seldom operate a bjt w/ fA or pA of Ic. So the error is not a big deal. Spice models for the bjt utilize Ies values in the fA to pA range with good results.

    The Ies value of 1.0 fA or pA does not represent an actual physical current associated with the device. It is a scaling factor derived from doping density, junction geometry, and other intrinsic material properties. If the I-V curve was log-linear all the way down to the Ies value, then Ies is the forward current when the (exp(Vbe/Vt) - 1) factor is equal to 1. That occurs when exp(Vbe/Vt) = 2, so subtracting 1 leaves 1. Multiply this factor by Ies, and we get Ies. That is the physical meaning of Ies.

    Ies is the current which would flow at Vbe = Vt*ln2 = 17.8 mV, if the device was true log-linear down to this value of I/V.

    As far as operation in the saturated region goes, yes, E-M covers that.

    Let's say that we use Vbe = 0.60V again, w/ Ies = 1.0 pA, but we are in the active region. Since Vbc < 0, the 2nd term in E-M can be ignored. The exp(Vbc/Vt) is nearly 0. For example, if the collector is at 1.0V above the base, Vbc = -1.0V, and exp(Vbc/Vt) = 1.23e-17! This is virtually 0, and exp() - 1 = -1. So, the 2nd term is simply -(-Ics), or Ics, the c-b jcn leakage current.

    Ic is then alpha_n*Ies*(exp(Vbe/Vt) - 1), which equals 13.87 mA. But if we increase the collector resistor so that saturation takes place, we now must deal with that 2nd term.

    Let Vce = 0.10V, the device is fully on hard. Vbc = Vbe - Vce = 0.60V - 0.10V = 0.50V. Let's say that the Ics value is 20 pA. The 2nd term of E-M gives

    Ics*(exp(Vbc/Vt) - 1) = 5.71 mA.

    But the E-M eqn has the 2nd term subtracted from the 1st term. Hence Ic = 13.87 - 5.71 = 8.16 mA.

    The actual Ic value is less in saturation, and E-M predicts the same. So E-M covers both active and saturated modes of operation. In active regions, the 2nd term is not needed, as it is small. But in saturation, we must use both terms.

    Does this make sense?

    Last edited: Sep 11, 2009
  12. Sep 12, 2009 #11
    Thank you! That clarifies things a lot for me. Here's a related follow up question - when the transistor is in saturation and the base to collector junction is forward biased, if the collector is connected to a load resistor will that junction behave in a similar way to how the emitter voltage follows the base voltage in an emitter follower? I had been reading about the problem of "latchup" in op amps, and the reason given was that excessive common-mode voltage could cause one of the transistors in a differential pair to enter saturation, and that the global feedback network could switch from being negative to positive. The only way I can picture this happening is if in saturation the collector follows along with the base, causing the output to become non-inverting rather than inverting.
  13. Sep 13, 2009 #12
    There's so much techno-babble in this thread, and none of it really addresses the OP's quesiton.

    Simply put, a FET is a "voltage-controlled" current source, because the output current is modulated by a the GATE VOLTAGE which changes the electro-static FIELDs inside the channel, hence the name FIELD-EFFECT transistor.

    On the other hand, a BJT is a "current-controlled" source, because the output current is modulated by the amount of BASE CURRENT that is injected from the base terminal. The characteristic "amplification" of current in a BJT is directly related to the amount of electrons that enter from the base...

    This is essentially what he needs to know. What does it even have to do with the Ebers-Moll model?? It's just a way of modelling the device from a circuit perspective...
  14. Sep 13, 2009 #13
    No it is NOT up to our choice... The name FET refers to a specific class of devices where the current modulation is done via a VOLTAGE terminal to affect the FIELD inside the device (FIELD-EFFECT)

    Bipolar Junction Transistors are not FIELD-EFFECT devices, they control the the transistor current by the AMOUNT OF CURRENT injected from the base terminal...

    These are FUNDAMENTALLY different in the device level and HAS NOTHING to do with the particular circuit model that's being used to solve the "circuit" problem...

    The names are given to accentuate the PHYSICAL principles of the device, hence in this context, Ebers-Moll and such "terminology" is useless and irrelevant.
  15. Sep 13, 2009 #14
    As a device physicist, most naturally I have strong doubts on what you are saying here.
    Could you please point out that paper you are mentioning showing the point where "current-controlled device" is defined for the first time?

    This is most probably incorrect, because the reason BJT is called a current controlled device has nothing to do with a "back-to-back" diode model created for circuit applications at a time where COMPUTERs were not as powerful as they are today to do the EXACT simulations.

    Nowadays, commercial tools like MEDICI or TAURUS know NOTHING about Ebers-Moll, because it is much stronger (and accurate) to solve the problem from a charge-control point of view... Where drift-difussion equations are solved self-consistently with poisson equation.

    So Ebers-Moll is NOT AT ALL fundamental and it was a mere convenience at the time.

    The real reason why BJT is called a current controlled device is related to the DEVICE PHYSICS as I explained in my previous posts.
  16. Sep 13, 2009 #15
    I had the paper on my old pc. I'll find it this week and post it. It does describe the bjt as back to back diodes, with a current controlled current source. With 2 diodes, 1 forward, 1 reverse biased, E-M describes each diode in terms of Shockley's diode eqn. But since electrons emitted from the emitter, pass through the base region and are collected by the electric field of the reverse biased c-b jcn, E-M described the collector current with

    Ic = alpha_n*Ie.

    The emitter current controls the collector current with alpha being the factor of proportionality. Hence, at the mAcro level, a 1st order approximation, a bjt is a CCCS. This is not the most fundamental model.

    I don't dispute your charge control assertion. I also agree that E-M is a "black box" mAcroscopic view of the bjt. If you wish to get down to thew mIcroscopic view, or "fundamental" if you prefer, then I agree that "charge control" is the best model. Beyond charge control, is the quantum mechanics modeling. QM is over my head. The late Richard Feynman, one of the 20th century's best physicists, said that nobody, including himself, really understands QM. I fully agree. I studied QM in modern physics class in the late 70's, and semiconductor physics 2 class in 2008. I know that I don't understand QM. I accept the published findings of the science community. I don't know enough to dispute them.

    Finally, although the charge control model works best, when we externally drive a bjt, we cannot inject a specific amount of charge into a device. We control the device terminal currents and voltages. The bjt is definitely a charge controlled device, but we use a current source to control the bjt, or a voltage source plus a series resistor. This is current control, or "current drive". Internally, again I agree with your charge control model as the best description of bjt behavior.

    We agree, so there's no need to argue. Thanks and as soon as I find that paper, I'll post it.

  17. Sep 14, 2009 #16
    Techno-babble? That was rude! He asked why the bjt & FET are classified as CCCS & VCCS resp. So, I gave him an historical account of the math and modeling of these devices. I did go deep into details, but one cannot explain device physics in 1 paragraph. My treatise exactly addresses the OP question, how can you suggest otherwise?

    It has everything to do with E-M! E-M was the 1st model attempt at the bjt, and still stands today. It has been modified and improved per Gummel-Poon. Also, for large signal operation and/or switching mode, the charge control model is extensively used. Also. the collector current is modulated by the emitter current, NOT BASE current. Base current is necessary for bjt operation, as is base-emitter voltage, but Ic is controlled by Ie. Every semi maker affirms the same. "Current-controlled" refers to emitter, not base current. You don't know semi phy.

    I was just giving him perspective. I studied EE at the BE & MS level in the 70's. In 2007 I returned to grad school for the Ph.D. I have this semester and next to complete my course work. I've been a practicing EE for 32 yrs. in R & D. I have developed many many pieces of equipment using discrete bjt, diodes, FETs, SCR, triac, IGBT, LED, photodiode, etc. I took semiconductor physics from the physics dept. in the 70's as a senior. As a grad student (MSEE) I took semi phy from the EE dept in '79. As a Ph.D. student, I took semi phy 2 in 2008, fabrication in 2008, and advanced sensors in 2007.

    I was mentored in the 80's by a boss who became an EE just as the bjt hit the market in the 50's. In his drawer of sample parts, he had germanium diodes & bjt parts from the 50's. The part nos. were 1N3, 1N4, 2N3, etc.! He knew transistor physics quite well and I learned much from him.

    If you don't see how my historical treatise based on years of post-graduate studies and mentoring from sages who have been there done that is relevant, and comes across as "techno babble", then maybe you should examine your own background. Maybe you are the one lacking in knowledge. What credentials have you got? Where do you get off rudely rebuking me? How many semi phy courses have you taken and passed? Do you have work experience in the semi industry.

    The problem is that electronics is a field where everybody sees themself as an expert. Anybody that's heard of Ohm's law thinks they are the equal to a professor. What folly.

  18. Sep 14, 2009 #17
    I am not questioning anybody's expertise, nor am I making any brave assumptions on people I encounter on internet by the limited knowledge they share with other people. There's just no way to see whether you are debating with an expert or a high school student.

    I don't need to show off by listing the courses I have taken or bragging about my own "credentials" and so on... This is not the right venue for that. Why don't we focus on the content instead of belaboring about the number of courses we took? I took a few SC courses, but apparently even a few of them was enough for me to set the basics right.

    Back to our discussion> You still owe the community the Ebers-Moll paper where THE CURRENT CONTROLLED DEVICE concept is being used FOR THE FIRST TIME in history to distinguish BJT operation... Why don't you post that, instead of your work xp?

    If you LOOK BACK and see what the OP INITIALLY asked (before being bombarded on Ebers-Moll, and history of BJTs etc...) it REALLy had an easy answer, and I am sorry, even if you are Schokley, your answer to the OP was unnecessarily complicated and off the track... Anybody who knows a decent amount of BJT and FET theory would agree. And he was drifted off to other parts of the problem without really appreciating the basic difference between a FET and a Junction Transistor.

    Now before we start another major debate on the operation principles of BJTs, let us agree on one thing: The collectro current on a BJT is mainly modulated BY THE BASE current, the sole purpose of the existence of a base terminal... You might want to check an elementary SC book (say, Pierret) before insisting on this one. Because it's really a VERY, VERY rudimentary piece. Of course, the base junction is intricately connected to the emitter junction by the common bias they share, but still what really changes the output current DIRECTLY depends on the number of carriers injected from the base... Without the base current there wouldn't even be any TRANSISTOR operation, because "base" terminal is analogous to the GATE terminal of a FET in BJT... And you are saying that the collector (drain) is modulating the ON-current of a BJT (FET) ? That's obviously wrong...The emitter is needed to EMIT (supply) a large amount of carriers to the collector and BASE current determines HOW MUCH of the emitted current will make it to the collector to be collected... That's basically it. And you reach to the conclusion that I don't know semiconductor physics from this wrong premise?... But I am not going to assume the same thing, don't worry. Because I don't know you well enough.

    The question was related to voltage control (Field-Effect) and current control ( Junction transistors, current injection), people don't care about what we know, they are coming here because they want intuitive and common sense answers.

    Years of "sagely" advising people must have taught you how to be concise and relevant. There are hundreds of ways to address a basic question from a beginner, you chose the worst one and you blame me for that.
    Last edited: Sep 14, 2009
  19. Sep 14, 2009 #18
    I'll search for the E-M paper. I'll buy it if I have to, as it always comes in handy. I'll immediately post it.

    Regarding bjt operation you say the following:
    [q] The collectro [sic] current on a BJT is mainly modulated BY THE BASE current, the sole purpose of the existence of a base terminal... You might want to check an elementary SC book (say, Pierret) before insisting on this one. Because it's really a VERY, VERY rudimentary piece. Of course, the base junction is intricately connected to the emitter junction by the common bias they share, but still what really changes the output current DIRECTLY depends on the number of carriers injected from the base... Without the base current there wouldn't even be any TRANSISTOR operation, because "base" terminal is analogous to the GATE terminal of a FET in BJT... And you are saying that the collector (drain) is modulating the ON-current of a BJT (FET) ? That's obviously wrong...The emitter is needed to EMIT (supply) a large amount of carriers to the collector and BASE current determines HOW MUCH of the emitted current will make it to the collector to be collected... That's basically it. And you reach to the conclusion that I don't know semiconductor physics from this wrong premise?... But I am not going to assume the same thing, don't worry. Because I don't know you well enough.

    I'm not saying what you think I'm saying. You say that w/o base current, there would be no transistor operation. I said exactly the same thing by calling base current necessary for bjt action. "Necessary" means that without base current, the bjt is non-functional. You say that. I say that. We are in perfect agreement so far. A bjt must have base current to work.

    No, I am not saying that collector current modulates the bjt ON current. That is plain wrong. The collector current is the quantity being controlled. We agree on that as well. So far there is no reason to be at odds. We agree.

    [q] The emitter is needed to EMIT (supply) a large amount of carriers to the collector and BASE current determines HOW MUCH of the emitted current will make it to the collector to be collected... That's basically it. And you reach to the conclusion that I don't know semiconductor physics from this wrong premise?... [/q]

    Yes. The emitter is indeed needed to EMIT a large no. of carriers into the base & collector. But here is the 1 fine detail where we disagree. You said that the "BASE current determines HOW MUCH of the emitted current will make it to the collector to be collected..." I feel this is more correct:

    The no. of emitted carriers from the emitter that do not survive the trip to the collector, plus the no. of carriers injected from base to emitter, DETERMINES the BASE current.

    But after rereading your statement, I guess it could go either way. If we drive the bjt in the common base configuration, then the base current is determined by the emitter current. We force a constant current source into the emitter, so that the base current is determined by Ie/(1 + beta). The emitter current, and beta determine the base current.

    But if we connect a constant current source to the base terminal and configure the bjt as common emitter, then your scenario would be correct. The emitter current is determined by the base current times (beta + 1). Thus the collector current is determined by emitter current times alpha, but the emitter current is base current times (beta + 1). So in this mode, Ic is controlled by beta*Ib.

    We pretty much agree, even down to minute details. But the Ebers-Moll thing is relevant. The bjt was discovered by placing a germanium junction on a conductive surface forming a point contact device. From there, current gain was observed. Drs. Ebers and Moll examined this new device and wrote equations characterizing said device. The eqn for Ic began with reverse leakage Ics, as the c-b jcn is a reverse biased diode. The eqn for Ie was simply Ies*(exp(Vbe/Vt) - 1), as this is a well known p-n jcn relation.

    So 2 back to back diodes SHOULD have Ie given by the forward bias eqn relations, and Ic given by a reverse biased eqn. So Ic should be much smaller than Ie.

    But, that is not what happens. Ic nearly equals Ie. How so? When the emitter, let's say the device is an npn, emits electrons towards the base region, the electrons do not recombine in the base entirely like a diode does. But, the E field in the reverse biased c-b jcn attracts these electrons. Thus instead of Ic being limited to Ics, the reverse leakage of a back biased p-n jcn, we get a current transfer from emitter to collector.

    The collector exhibits constant current source behavior as:

    Ic = alpha*Ie, plus we add the reverse leakage from collector to base "Ics".

    Drs. E & M illustrated their diagram with a CCCS, where Ie controls Ic, with alpha being the constant (coefficient) of proportionality. It was in the paper, and I saw it with my eyes. Is it possible that the CCCS concept was published elsewhere prior to December 1954? I doubt it.

    This was, IMHO, the 1st time that a bjt was referenced as a CCCS. I am well aware that the E-M eqns are a 1st order external approximation, and do not delve into semi phy. I was just providing historical background, which I thought was appropriate.

    Oh well, I didn't mean to come across sounding like Mr. know-it-all. I don't believe in that. I was just trying to understand what background you had, and how to discuss the details of bjt operation based on the model you employ. No personal offense was intended, I assure you.

    I don't think we disagree at all.

  20. Sep 14, 2009 #19
    I was talking about this...If I were to explain it to someone, I'd prefer calling the "BASE" current the controlling handle because that's what we do in practice, especially in amplifier operation.

    I agreed that what you call the emitter current and the base current are connected (they basically feel the same bias across the p-n junction) but still, base control is making more sense to me... You describe the base current as if it were an emergent terminal current, just because we have some collector and emitter terminals, we observe a base current which is "necessary" for transistor operation...

    But I believe it's not the best description because the IDEA of transistor operation starts with a BASE control current... You START by assuming that you can control a large EMITTER current by a much smaller current through the coupling of the base junction to both emitter and the collector.

    So treating it like it emerges out of nowhere is slightly misleading. You describe the individual components that "constitute" the base current but this is not relevant. What we care is WHO CONTROLS WHAT, and BASE CONTROLS COLLECTOR CURRENT is the right answer here.

    This is exactly why people present BETA ( COLLECTOR CURRENT DIVIDED BY BASE CURRENT) as a figure of merit for a good BJT... We almost never care about the emitter current, per se. The sensitivity of the on-current to the minute base current is much more significant, because at the end of the day, sharp circuit characteristics, and all the other good stuff depend mainly on Ic/Ib

    So I still can't get how you conclude :
    because emitter current is NOT controlling anything... It is just the supply... At least conceptually it makes much more sense to think about it this way.
    If you post the Ebers-Moll paper and show us where they really make the seminal "current-controlled" device definition that would be great. I'll completely retract my criticisms because your remarks would then at least have a historical basis.

    But at this point, I still believe that going to an "approximate" circuit model to describe one of the defining device characteristics of two families of transistors is like trying to explain quantum mechanics by a semi-classical theory...

    Why not start from the bottom, which is simpler and more accurate?
  21. Sep 14, 2009 #20
    I'll get the E-M paper at the university where I study. I'm sure they have it. Meanwhile here is a Univ of Berkely lecture note on E-M. They employ a CCCS, with Ic = alpha*Ie.

    As far as Ib being the "controlling current" vs. Ie, Ie is usually employed.

    With a classic bias network, a resistive divider whose midpoint sets the base terminal voltage wrt ground, an emitter resistor, and a collector resistor, the emitter current is controlled via the resistors. The Vbe value is very close to 0.65V. As temp varies and hfe varies, the Vbe remains near 0.65V. The Ie value stays constant over the span of temp and device variations. When Ie is established at some quiescent value, then

    Ic = alpha*Ie.

    Alpha is 0.99 +/- 0.01 for all devices at all temps.

    If however, we set Ib at some value, then

    Ic = beta*Ib. This is not a good thing as beta is device dependent, temp dependent, and current level dependent. A circuit which employs the base current to control Ic is called "beta dependent". Except for switching applications, controlling Ib is usually not done.

    Amplifiers, including monolithic op amps, do, however, employ base current control of Ic in the 2nd stage. The 1st stage is an emitter coupled differential stage. The emitters of the diff pair have their emitter current controlled via a current source/sink, or large resistor. The 2nd stage is a common emitter stage. The 1st stage outputs a current source which feeds the base terminal of the 2nd stage bjt. The 2nd stage collector current (output) is beta*Ib. But, the open loop gain of the op amp depends on beta. The higher the beta, the higher the open loop gain. Global feedback corrects this problem.

    Thus, for the 2nd stage, Ic = beta*Ib. Here, the collector current is indeed controlled by the base current, not emitter current.

    We can control Ic with either Ib or Ie. Ie is usually the preferred way of doing it, but Ib is often used. It depends on the application.

    From a device physics view, however, Ie determines Ic as Ie supplies the carriers, not Ib. But Ib is absolutely necessary or the device is non-functional. Likewise, Vbe is necessary but is not the controlling quantity. Without Ib and Vbe, there is NO Ie at all.

    FWIW, I agree that beta is indeed a very good figure of merit for a bjt. It describes how effective it is at amplifying current. Beta is very very important, especially the worst case minimum value. A good design will function consistenly well for any beta value above the wcm (worst case minimum). But the overall performance of the network is improved when using a device with a high wcm beta value. No argument there. Beta is all important especially when there is only 1 stage of amplification. Beta is all important here. With several stages, the beta values multiply, and a high beta for each stage may not be needed. The fewer the stages, the more important is beta.

    Make sense?


    Also, when used as a switch, the base current is usually controlled in a manner that forces a base current greater than Ic/beta_minimum. The collector saturates, and beta dependency is not an issue.

    Attached Files:

    Last edited: Sep 15, 2009
  22. Sep 14, 2009 #21
    I was thinking about the active-mode amplifier configuration (probably the most common use, or at least the principal) but I agree that if you bias the device differently, you'll see different effects as you explain...

    For digital, I think we operated in the Saturation mode, to get constant logic values, but "amplification", I still insist, is mainly due to the fine tune between Ib and Ie (therefore Ic)

    I was objecting specifically to "emitter control" of collector current in Forward-Active operation but it seems that we agree on these issues now.
  23. Sep 17, 2009 #22
    I obtained the 1954 Ebers-Moll paper, but it is too large to upload on this forum. Send me an email, and I'll send it to you. Cheers.

  24. Sep 17, 2009 #23
    Is it this paper?

    Large-Signal Behavior of Junction Transistors
    Ebers, J.J. Moll, J.L.
    Bell Telephone Labs. Inc., Murray Hill, N.J
    Proceedings of the IRE, 1954

    If it is, you don't have to send it to me,
    just point out the relevant portion in the paper.

    If it's not, post the name of the paper so others can see it as well.

    I have access to the journals, so it's no problem
  25. Sep 17, 2009 #24
    Yes, that's the one. The illustrations in the left hand column of page 1765 depict a current controlled current source where "Ic = alpha_n*Ie". Actually ther'e another CCCS for the upside down bjt operating in inverse mode. As I said earlier, every bjt is really 2 bjt devices in 1, a right-side up bjt in normal mode, and an upside down bjt in inverse mode. That's why the E-M equations have 2 terms for each current. I hope this helps. BR.

  26. Sep 18, 2009 #25
    Well depicting a current controlled current source, and defining the concept of "current-controlled devices" for the first time in history are different things.

    You were claiming that the reason a BJT is cited as a 'current-controlled' device had its roots in this 1954 paper.

    Having briefly checked the paper, I see that neither Ebers nor Moll really means to define anything other than their elegant model.

    I didn't ask the paper to study the E-M model, I just wanted to clarify an important misconception.

    Thanks for posting it,
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook