Talk:Bipolar junction transistor/Archive 1

Archive 1

Voltage or current?

"very sensitive to the current passing through the base."

I've heard it's actually the voltage that does it, in BJTs *and* FETs. current is just a byproduct in BJTs. - Omegatron 20:37, Jul 28, 2004 (UTC)

You can't have one without the other. Whether you see it as current-sensitive or voltage-sensitive just depends on which model you use. As a first approximation, a BJT can be modelled as a current-controlled current source. The proportionality constant between the base current and the collector current is β or h_FE. This is taught in many lower-level electronics courses, since it is often sufficient for circuits operating in the forward-active mode. The Ebers–Moll and Gummel–Poon models use diode-like equations relating voltage to current. When using the current-controlled model, it's useful pedagogically to assign causation to the current, and when using the Ebers-Moll model it may be more useful to assign causation to the voltage. -- Tim Starling 01:02, Jul 29, 2004 (UTC)

Math models obscure the question as they simplify the concepts. Look to the full-blown details in order to understand how transistors actually work. As with any diode junction, the B-E junction contains an insulating "depletion layer" whose thickness is controlled by applied voltage. Make this insulating layer thicker, and the diode turns off (reverse biased), or make the layer thin enough, and charges are able to get across; the diode turns on. E.g. the diodes's current is controlled by the voltage applied to the diode's terminals. Now in a BJT there essentially are two currents in the same B-E diode junction: the base current and the Collector current, and... both are controlled by the thickness of the insulating B-E depletion layer. And the thickness of the B-E depletion layer is controlled by the applied voltage Vbe. THEREFORE, both the collector current Ic and the base current Ib are controlled by the Base-Emitter voltage Vbe. Things only SEEM confusing because Ib happens to be proportional to Ic. In reality Ib cannot affect Ic, instead both are affected by a third variable: the B-E voltage. Weird, eh? At the most fundamental level, BJTs and FETs are both controlled by voltage. However, in a BJT the main current must cross an insulating layer of variable thickness, while in an FET the main current remains in the conductive regions while the insulating regions encroach from the side. If a BJT acts like the dark lens in a pair of variable-density sunglasses, then an FET is more like a variable-aperature optical shutter or iris: a variable-sized empty hole. --Wjbeaty 21:29, Mar 20, 2005 (UTC)

Can't agree with this. Whether a BJT is a current or voltage controlled device depends upon whether the base-emitter junction is current or voltage fed. If we use a current source, then the b-e voltage will float to whatever is appropriate to allow that current to flow. If we use a voltage source, then a current appropriate to the voltage will flow. So, in truth, the collector current is dependent on *both* the base-emitter voltage and current. However, given that beta (the current gain) is relatively constant, then it's easier to consider a BJT as a current controlled device: increase the base current by x, and the emitter current increases by beta times X. Phil Holmes 15:45, 19 August 2005 (UTC)

I agree in principle with what you are trying to say here but I am uncomfortable with how you have said it. Specifically, the manner in which a two-port is driven does not in any way affect it's fundamental nature. Further, claiming that the collector current is dependent on both the base-emitter voltage and current is misleading. I am not aware of an equation for collector current that, to first order, has independent terms for both base-emitter voltage and current. But, you have also said in effect (and correctly) that the base-emitter voltage and base current are not independent. Thus, the collector current equation can be written as a function of base-emitter voltage OR base-emitter current.

Now, consider an ideal transconductance amplifier. By definition, the input current is zero so driving this amplifier with a current source is, let us say, a problem. In the non-ideal case however, the input current may be non-zero and is related to the input voltage (linearly in the case of a resistive input or more generally, non-linearly as in the case of BJT). Thus, by solving for the input voltage in terms of the input current, one can look at the non-ideal transconductance amplifier as a non-ideal current amplifier. This has nothing to do with the way the amplifier is driven - it is simply an alternate perspective on the transfer equations.

If the base-emitter of a BJT is driven directly by a current source, it is reasonable to adopt the current amp perspective. Since the voltage developed at the input is relatively small and nearly constant over a wide range of input current, the BJT is an (approximately) ideal linear current amplifier from this perspective. If a voltage source is used instead, it is reasonable to adopt the transconductance amp perspective. However, since the source current varies greatly and non-linearly over a small range of input voltage, the BJT is a highly non-linear non-ideal transconductance amplifier from this perspective. Alfred Centauri 20:57, 19 August 2005 (UTC)

I think the only difference between what you've said and what I said is that I said *both* and you, in effect, replaced that with *either*. On reflection, I'd go along with that. Phil Holmes 08:30, 22 August 2005 (UTC)

It all boils down to whether we want to teach how transistors work, or whether we only want to solve some design equations. Never lose sight of the fact that the collector current is never directly controlled by the base current. If we drive the base with a constant current source, this creates a certain value of BE voltage. This BE voltage then determines the thickness of the depletion layer, which then controls the collector current. Yet if we could somehow hold Vbe constant while changing Ib, we'd find that Ib cannot control Ic. Be careful not to confuse the simplified "black box" mathematical model with the physics-based explanation of transistor behavior. Yes, simplified engineering equations say that Ic is determined by Ib, and these equations are incredibly useful... but these equations are "wrong" because they mislead us into believing that Ib directly affects Ic. It does not. Instead, Ib determies Vbe, and Vbe determines Ic, which shows that a transistor is a voltage-controlled device which pretends to be a current-controlled device. Many people seem to hate this fact, and try to talk themselves into believing that Ib directly affects Ic. But unfortunately there's no escaping the well-known transistor physics. If we want to know how transistors work, we have to confront the fact that the flow of Ib charges has no affect on the flow of Ic charges. Yet for simplified introductory classes where nobody cares how transistors actually work, it's fine to say that Ic is controlled by Ib... and then to never mention any complexities such as the fact that the control relies indirectly upon Vbe. On the other hand, if a more advanced textbook never mentions the Eber-Molls model, or worse, tries to use semiconductor physics concepts to "prove" that Ib affects Ic, then that author has a serious misconception, and is doing their readers a disservice. --Wjbeaty 20:02, 4 April 2006 (UTC)

mnemonic

i can't ever remember from the schematic symbol which side is emitter/collector and whether it is npn/pnp. any silly mnemonics for remembering or should i make one up? - Omegatron 16:21, Oct 18, 2004 (UTC)

It's not much of a mnemonic, but the terminal with the arrow is the emitter, and the arrow points in the direction of current flow. So if the arrow points from emitter to base, it implies the emitter is injecting positive charge into the base. Emitter positive, base negative, therefore pnp. -- Tim Starling 03:45, Oct 19, 2004 (UTC)

yeah, not much of one. something like e is for arrow and the N in the middle of PNP stands for "pointing iN". it has to be something you can just remember without thinking about it. but that is a really stupid attempt. ummmm... i dunno. i'll just stare at them until i remember i guess. :-) - Omegatron 13:41, Oct 19, 2004 (UTC)

Never Points iN and Points iN Proudly. that is a good one. and a bow/archer "emits" arrows? i guess that works. - Omegatron 14:50, Oct 19, 2004 (UTC)

The one I used was Not Pointing iN for NPN, and you could use Pointing iN Permanently for PNP if you need to remember both. This assumes you can remember that the emitter is the one with the arrow on. This last could'nt be more obvious really- its a bit like the symbol for a male! (who is normally the 'emitter' in sexual relations! --Light current 00:55, 5 February 2006 (UTC)

doping

In the illustration, there seems to be 4 types of semiconductor - p+, p-, n+, n-. What does this signify? The diagram looks to be more informative than the usual NPN or PNP layer discription. Where can I find more info? Thanks.

p⁺ means strongly doped with acceptor impurities, p^- means weakly doped with acceptor impurities, n⁺ and n^- mean strongly and weakly doped with donor impurities, respectively. The region of very strong doping leading up to the metal is called an ohmic contact. Strong doping reduces the diodic effect of touching metal and semiconductor, see Schottky barrier. As for more information, a google search didn't turn up much for me. You might want to try a textbook, e.g. ISBN 0471333727. -- Tim Starling 00:03, Nov 23, 2004 (UTC)

I've beefed up the sections on doping in the semiconductor article - this explains the n+, p+ nomenclature. --Phil Holmes 21:26, 29 August 2005 (UTC)

models

this should have some small- and large- signal models and such. - Omegatron 02:24, Mar 4, 2005 (UTC)

I'll draw up some diagrams for ones that we decide to use. Actually, this could be a good test for Wikipedia:Modular electronics diagrams. I'm going to start working on it. (Since I need to re-learn this for a project anyway). - Omegatron 16:44, Mar 6, 2005 (UTC)

Here are two active-mode models:

			C
			αiE
B
	iE ↓		Is/α
			E

B						C
	iB ↓				βiB
				E

What do you think? - Omegatron 18:09, Mar 6, 2005 (UTC)

Uses and possible technical details

Well I was thinking that a uses section for BJTs could be added to the article.

Uses - Current sources - TTL, transistor transistor logic

Technical Detail

-Low input impedance when compared to FETs and MOSFETs. Input impedance depends on amplifier configuration. - Output impendance is low. Also is dependent on amplifier configuration.

Amplifier configurations

Links to the following configurations maybe.

- Common collector - Common emitter - Common base

How gain works

Please see that I've changed the sentence that used to say "The proportion of electrons able to run the base "gauntlet" and make it to the collector is very sensitive to the current passing through the base." That's not correct - I've now clarified that this is fairly constant, but that gain occurs as a result of the emitter being more highly doped. Phil Holmes 08:47, 22 August 2005 (UTC)

Good catch. I was just reading that paragraph a day or to ago and the word 'proportion' didn't jump out at me at all. But don't forget that the light doping and the thinness of the base is also important to the gain. What is required is that the the diffusion length is long compared to the thickness of the base. This ensures that the odds of a charge carrier from the emitter recombining in the base region before reaching the base-collector junction are small. The smaller the odds, the greater the 'gain'.

One thing that I don't see mentioned in the article is that once the charge carrier reaches the base-collector junction, it is the electric field within the depletion zone that sweeps the charge carrier into the collector region. In a way (this is sure to raise some eyebrows!), it can be said that the role of the forward biased base-emitter junction is to increase the reverse bias 'leakage' current through the base-collector junction.

On another note, there is a statement in the opening paragraph that I'm concerned about but that I'm hesitant to change. This statement is: "BJTs can be thought of as current-controlled resistors". This is really wrong technically yet if it helps newcomers grasp the concept I'm not going to bother with it. Alfred Centauri 20:41, 22 August 2005 (UTC)

I've always been in favor of the "controlled resistors" analogy for newcomers, but it's perfectly acceptable to add "this is only a crude analogy" or whatever you want to make it accurate. — Omegatron 21:33, August 22, 2005 (UTC)

I'm not sure there is a better analogy at the newcomer level. I've thought about something like "the BJT is to electric current what power brakes are to pedal pressure" but I not sure if that even makes sense. Is there any way that an analogy can be made between a BJT CE amplifier circuit and a power brake system? Alfred Centauri 21:53, 22 August 2005 (UTC)

You can make a good analogy to a Globe valve where the plunger is controlled by a diaphragm. EGR valves work this way. There is a big flow that is stopped by the plunger, and then you can control the plunger by applying a constant pressure or vacuum to a diaphragm (FET) or a small, constant current to a diaphragm with small holes in it (BJT). I've been meaning to draw a picture of this for hydraulic analogy. — Omegatron 23:57, August 22, 2005 (UTC)

I just visited the hydraulic analogy page. I've often thought about this analogy so it is good to see you working on this. I have a suggestion for resistor. A small diameter pipe sounds like higher gauge wire. But a resistor is like an obstacle course for moving electrons so maybe a water resistor is like a restrictor plate. I'm picturing a disk with holes in it that fits inside the pipe. I doubt if such a thing is actually found in a hydraulic system but maybe it would help create a picture of electrical resistance in a resistor Alfred Centauri 03:06, 23 August 2005 (UTC)

The analogy of a BJT to a current controlled resistor is very poor. Alfred's comment towards the top of this page reminded me of how much that grated when I read the article for the first time. To explain. The resistor we would be refering to would be between the collector and emitter terminals. With no base current, we put a voltage varying between 1 and 5 volts (say) between those terminals, and see what current flows. In this state, it would be negligible. We now apply some base current and do the same. The result we will see is that beta times the base current will flow at 5 volts, and there will be very little difference at 1 volt. (The Early effect means there is some difference). This is very different behaviour from a current controlled resistor. With one of those, I'd expect to see "infinite" resistance with no base current, and a current at 1 volt that is 1/5 the current at 5 volts with base current applied. My suggestion would be to delete the analogy. Phil Holmes 11:44, 23 August 2005 (UTC)

We already agreed it's a poor analogy. I still think it's helpful for someone who has no concept of what it is. It's better than saying "a controlled current source" or "an amplifier". Do you have a better suggestion? — Omegatron 14:06, August 23, 2005 (UTC)

Frankly stuck for a good analogy. Perhaps it would be best not to bother with analogy and simply say that a BJT is a device that controls current flow between the collector and emitter by using a control signal between base and emitter? Incidentally, and you're probably aware of this, I found a useful reference when searching for source information to validate my memory on some of this. [1] Phil Holmes 15:52, 23 August 2005 (UTC)

The physical reason for gain

I was wondering if much more could be said on how current and voltage gain are obtained in a transistor. I have no idea myself, and i couldn't find any good explanation of it. I too think of a transistor as a variable resistance, but that analogy obviously breaks down when applied to an amplifying transistor circuit. I would greatly appreciate it if anyone could explain how gain is physically obtained. Fresheneesz 21:12, 10 February 2006 (UTC)

Does this not explain it ?

In normal operation, the emitter-base junction is forward biased and the base-collector junction is reverse biased. In an npn-type transistor for example, electrons from the emitter wander (or "diffuse") into the base. These electrons in the base are in the minority and although there are plenty of holes with which to recombine, the base is always made very thin so that most of the electrons diffuse over to the collector before they recombine with holes. The collector-base junction is reverse biased to prevent the flow of holes, but electrons are swept into the collector by the electric field around the junction. The proportion of electrons able to cross the base and reach the collector is approximately constant in most conditions. The heavy doping (low resistivity) of the emitter region and light doping (high resistivity) of the base region mean that many more electrons are injected into the base, and therefore reach the collector, than there are holes injected into the emitter. The base current is the sum of the holes injected into the emitter and the electrons that recombine in the base - both small proportions of the total current. Hence, a small change of the base current can translate to a large change in electron flow between emitter and collector. The ratio of these currents Ic/Ib, called the current gain, and represented by β or Hfe, is typically 100 or more.

--Light current 21:44, 10 February 2006 (UTC)

mm.. not really. I mean, it does indeed explain the general operation of a transistor, but the way its written is a bit hard to understand. On top of that, gain takes a huge backseat in this description. After reading that paragraph perhaps for the 4th time or so, I still can't imagine how gain would occur. Obviously small changes in the base voltage produce large changes in the amount of current allowed throgh the transistor - thats the whole point of the device. However, gain requires that *more* current or *more* voltage come out than comes in, which seems 120% unintuitive to me and I think is *so* important that it should have its own section on this page. Also I'm pretty sure that the base *current* is more of a byproduct than a cause, and that the base *voltage* is really what produces the transistor changes. Fresheneesz 22:25, 10 February 2006 (UTC)

OK then, does this explain it any better:

Math models obscure the question as they simplify the concepts. Look to the full-blown details in order to understand how transistors actually work. As with any diode junction, the B-E junction contains an insulating "depletion layer" whose thickness is controlled by applied voltage. Make this insulating layer thicker, and the diode turns off (reverse biased), or make the layer thin enough, and charges are able to get across; the diode turns on. E.g. the diodes's current is controlled by the voltage applied to the diode's terminals. Now in a BJT there essentially are two currents in the same B-E diode junction: the base current and the Collector current, and... both are controlled by the thickness of the insulating B-E depletion layer. And the thickness of the B-E depletion layer is controlled by the applied voltage Vbe. THEREFORE, both the collector current Ic and the base current Ib are controlled by the Base-Emitter voltage Vbe. Things only SEEM confusing because Ib happens to be proportional to Ic. In reality Ib cannot affect Ic, instead both are affected by a third variable: the B-E voltage. Weird, eh? At the most fundamental level, BJTs and FETs are both controlled by voltage. However, in a BJT the main current must cross an insulating layer of variable thickness, while in an FET the main current remains in the conductive regions while the insulating regions encroach from the side. If a BJT acts like the dark lens in a pair of variable-density sunglasses, then an FET is more like a variable-aperature optical shutter or iris: a variable-sized empty hole. --Wjbeaty 21:29, Mar 20, 2005 (UTC)

--Light current 22:36, 10 February 2006 (UTC)

It does explain a transistor better. But I must be missing something fundemental, because I see no mention of gain in that description, nor can I see how gain arrises from that description. All of these descriptions seem to follow the variable-resistor analogy, while the concept of gain doesn't. Sorry if i'm being a pain in the ass, but I really think this is important. Fresheneesz 23:22, 10 February 2006 (UTC)

Well you do understand that current gain is the ratio of Ic/Ib ? Also, if you can control a variable resitor with a voltage or current, you can get gain , yes? --Light current 23:28, 10 February 2006 (UTC)

That must be what I don't understand. I understand the mathematics of Ic/Ib, but I don't understand why that it is the case. What I think I don't understand is that you can get gain with a voltage-based variable resistor. I would have thought that that is impossible. If the variable resistor analogy is held, then by, say, increasing the voltage in the base, the resistance through the CE junction goes down. But no matter how much one varies a resistor, it seems to me that the limits are a gain of 1 and a gain of 0 (when the resistance is infinity and 0 respectively). It seems to me that the resistance would have to be negative for one to get gain out of it (ie a battery or something). Fresheneesz 00:28, 11 February 2006 (UTC)

OK. If you have a variable resistor connected between a voltage supply and a load, then suppose somehow, you could alter the value of that variable resistor in sympathy with the input signal. Do you see how that will change the voltage across the output resistor?--Light current 01:00, 11 February 2006 (UTC)

Perhaps it might help if a sentence were added at the end of the paragraph originally quoted by Light Current. Put simply, this feature of the high doping of the emitter and low doping of the base means that a small change in current flowing between the base and emitter causes a larger change in current flowing between the collector or emitter: gain. --Phil Holmes 15:37, 11 February 2006 (UTC)

I think Fresheneesz must be thinking that the gain is happening from C to E or something, when in fact it is happening between B and C. But he is correct in thinking that it is the base emitter voltage that controls the collector current and the base currrent is just a by product!.--Light current 16:52, 11 February 2006 (UTC)

OH! Gain from B to C. That makes perfect sense. I've never thought of it like that. But looking back at the problem that confused me, It seems like the emitter and base get the same magnitudes of voltage. Heres a picture:

In my book this is labeled as "common-base" configuration

I can see how one would get gain using two different voltage sources (one at the emitter end, and one at the base end), but I can't see how gain could happen when the emitter and base voltages have the same magnitude. Fresheneesz 21:47, 11 February 2006 (UTC)

Well E and B have nearly (but not quite) the same voltages on them. Remember that there is an exponential dependence of the collector (or emitter) current on Vbe.--Light current 22:53, 11 February 2006 (UTC)

I didn't read the whole thing (and didn't understand transistors neither) but I'll try to help anyway, corrections are welcome:

gain is a mathematical concept that arises from equations. In the physical world we have lots of electrons (in the npn) pushing to enter the emitter and cross base's ultra thin layer (since the depletion (insulating) layer has been removed by direct polarization.) Then electrons are projected to the collector's layer since the ultra thin base layer has not as much holes to capture every electron. In the collector an electromagnetic field attracts the 99% of electrons and "collects" them to the collector's exit. A negligible amount of charges goes to the base "forming" a little current (that can be seen as an unwanted side-effect, hence the MOSFETs.) The whole process can be modelled mathematically as an amplifier with a gain of quasi 100 (the 1 electron that goes out the base against the 99 that go out the collector.)

BJT transistors are made in order to work this way efficiently. That's why they have a thin base layer (otherwise there would be no transistor effect) and an emitter more doped than the collector and very different surfaces (otherwise they would have been symmetrical and with major power losses.) --Zimbricchio 13:09, 7 June 2006 (UTC)

Here is another perspective of the bjt gain: For a well constructed transistor when a single carrier (hole for an NPN) is injected from the base into the base-emitter space charge layer, you will get approximately beta times as many carriers (electrons for an NPN) injected from the emitter into the space charge layer. The reason for the larger number of carriers from the emitter is the higher doping of the emitter. Once the carriers injected from the emitter reach the base most will diffuse through and end up at the collector. You get more than beta injected from the emitter because you lose a few due to recombination in the base, hence the requirement for a thin base. The imbalance of carriers from the base compared to the emitter is the primary cause for current gain. This explanation covers the case for an upside down transistor (where you try to use the collector as the emitter) where the beta is about one because the base and collector doping is comparable.

This explanation also is relevant to the current control vs voltage control discussion because it shows that the factor that controls the emitter current is the base current, the number of carriers injected from the base to emitter.

Volts must exist? Two majority carriers?

Light current - Your change of 'a voltage must be applied' to 'a voltage must exist' is interesting but rather opaque. What is your reasoning for making this change?

Also, isn't it true that a PN junction necessarily 'employs' both types of majority carriers? It might be misleading to make this statement especially in the opening paragraph. It is usually said that the BJT is a minority carrier device for the reason that the charge carrier current making the greatest contribution to the total electric current is the minority carrier in the base of the BJT. Alfred Centauri 13:52, 25 August 2005 (UTC)

a) I changed it to 'volts must exist' to remove the impression that you can just put any old battery across the BE junct. as the diagram shows. Because of the exponential relationship of current to voltage, current can vary tremendously with a few mV change. I admit its not the best wording! But I cant alter the diag to include a base resistor.

b) I mentioned 2 types of charge carrier to try to explain the name Bipolar Junction Transistor (as opposed to FET which has only one type of charge carrier) Again this may need rewording if you think its not clear.

See here for instance [2] Light current 15:52, 25 August 2005 (UTC)

Perhaps stating that current is carried by both holes and electrons would be a good wording? Phil Holmes 16:11, 25 August 2005 (UTC)

I've changed some wording related to my first question but I guess I got auto-logged out before I finished my edit. Alfred Centauri 18:35, 25 August 2005 (UTC)

Anybody know what is meant by 'high' doping and 'low' doping? Is the writer trying to differentiate between the majority carrier types in the base(P region) and the emitter(N region). If so, it is not explained very well. Light current 23:04, 25 August 2005 (UTC)

If I were to use those terms, I would be refering to high or low doping densities. The base region is lightly doped which means that the mobile hole and electron densities are relatively similar. That is, the majority carriers aren't that great of a majority. On the other hand, the emitter is highly doped meaning that there are very few minority carriers compared to the majority carriers. BTW, I like your addition of the word 'appreciable' but I would suggest that you parenthetically add what appreciable is in this context. I would think that something in the neighborhood of 1mA would be about right. Regarding your change of 'emitter' to 'base' - if it makes you feel better then so be it. Both are correct as long as $\beta$ is finite. Here's a question for you to ponder though. Let $\beta$ go to infinity and ask what current does $V_{BE}$ control? Alfred Centauri 01:44, 26 August 2005 (UTC)

I was thinking of the case when the collector current is small - but in that case, Ib = ~ Ie, so I'm not sure now! I guess I was thinking of the B-E junction alone.Light current 20:21, 26 August 2005 (UTC)

I'm not sure what your notation is here. Are these small-signal currents? Actually, it looks like the notation used on the original schematic that I have now modified. That is, for DC (constant) voltages and currents, the letter and subscripts should both be capitalized. So given that, you are correct that $I_{B}$ is proportional to $I_{E}$ as long as $\beta$ is finite and the BJT is in the active region of operation. So, it is equally correct to say that $V_{BE}$ controls $I_{B}$ or $I_{E}$ or $I_{C}$. However, if you let $\beta$ tend to infinity (which is not so farfetched - the small signal model for a JFET 'looks' like the small signal model of a BJT with $\beta$ set to infinity), $I_{B}$ is zero. In this limit, it is clear that fundamentally, $V_{BE}$ controls $I_{E}$ I'm teaching a junior level class on small signal modeling of transistor circuits this semester. It's good to clear out the cobwebs and review this stuff. It's amazing how much more sense it makes when you have to explain to someone else. I remember reading somewhere that a prominent theoretical physicist would, whenever he was stumped on some difficult problem, explain the problem to his dog. His dog would just sit there and look at him without interrupting and, as he would try to explain the problem to his dog, he would invariably see the resolution to his difficulty. Alfred Centauri 20:52, 26 August 2005 (UTC)

Yes, I was intending DC currents, but I've always been sloppy about the capitalisation I use (I just cant remember them all (any?). The problem does not practically occur in JFETs in that most reasonable voltages applied between gate and source won't cause a large current to flow and blow up your transistor junction( unless you apply the wrong polarity).-- that is the point I was tring to make. PS I must get a dog!!(or is WP just as good?):-) Light current 21:22, 26 August 2005 (UTC)

Text from Transistor

I did not know why I'm putting so much info on transistor, slaps forehead. The transistor page can get smaller if I link to bjt and fet. The following paste will be highly edited on transistor now. It is here for future use.

The bipolar junction transistor (BJT) was the first type of transistor to be commercially mass-produced. Bipolar transistors are so named because the main conduction channel uses both electrons and holes to carry the main electric current. Two p-n junctions exist inside the BJT, colector-base junction and base-emitter junction. When the BJT is not powered, the junctions are in unbiased thermal equilibrium with a depletion region formed at each junction. The arrangement of greatest interest is when the B-E junction is forward biased and the C-B junction is reverse biased because this makes amplification possible. Applying a forward bias voltage to the B-E junction unbalances the thermal equilibrium of the junction. In an NPN type BJT, the p-type base begins to inject surplus holes (holes not required to maintain thermal equilibrium) into the emitter where they quickly recombine in the n-type material at the emitter contact. Similarly, the emitter injects surplus electrons into the base. If not for the close proximity of the reverse-biased C-B junction, the B-E junction would behave like a diode–injected emitter electrons would recombine in the base and transistor action would not occur. However, in the NPN BJT, the electrons which are injected from the emitter into the base diffuse across the very narrow base region before most of them have time to recombine within the base region. The depletion region of the nearby reverse biased C-B junction contains an electric field, which sweeps any electrons close to it into the n-type collector, where they recombine with holes at the collector contact. Through this action, only a small amount of carriers in the base (holes) are needed to cause a large amount of carriers in the emitter (electrons) to flow through to the collector.

diagram

that diagram looks like its meant to represent a particular highly complex process but it doesn't specify which one and really serves more to confuse then inform. imo the first diagram in the article should show a minimal bjt and possiblly have diagrams of more complex processes further down. Plugwash 03:15, 7 January 2006 (UTC)

Dopant concentrations

At the moment, part of the article reads "By varying the voltage across the base-emitter terminals very slightly, the current allowed to flow between the emitter and the collector, which are both heavily doped and hence low resistivity regions, can be varied." AFAIK, this is incorrect. The emitter region of the BJT is the highest doped region, whilst the collector region has a relatively low dopant concentration --Rspanton 00:22, 28 January 2006 (UTC)

Correct. I've deleted the reference to doping at this point to correct this. (The author probably meant that the bulk collector is highly doped to minimise collector resistance, which is true, but it's confusing to state that here). --Phil Holmes 16:41, 28 January 2006 (UTC)

The transistor models

I think the article should include the transistor's signal models, including the reference to the model's components such as gm, r0, rπ, Cμ, Cπ, etc... It should represent the transistor model to a full range of frequencies, from DC to the infinite. Afonso Silva 21:49, 3 February 2006 (UTC)

You are talking about a particular transistor model (the hybrid-pi model). Would it be best to start a new page about that model, or add it to this one? I am unsure about the relevant wikipedia policies. Rspanton 00:01, 5 February 2006 (UTC)

I think a new page on hybrid pi woud be appropriate.--Light current 00:50, 5 February 2006 (UTC)

Yes, that was an example, I think the article should include a section with a brief description of the models, with a picture and the description of the main parameters on every type of signals. Those models include not only the hybrid pi model but also the Ebers-Moll model or the T Model (I don't know more), that section would later include links to the main articles and also explain the advantages and applications of each model. The article lacks that kind of information and much more. By the way, Rspanton, what policies are you talking about? Afonso Silva 23:41, 7 February 2006 (UTC)

Yes . One page on Transistor 'hybrid pi' model, one page on Transistor 'T' model. (I think Ebers Moll is the same as the T model). I dont think there are any more common models!--Light current 00:42, 8 February 2006 (UTC)

Maybe I'm wrong, but I think the Ebers Moll and the T Model are separate models, the Ebers Moll model is used to predict the operation of the BJT in all modes (active and saturation). The T model is mainly used to analyze the BJT behaviour in AC operation. Afonso Silva 21:22, 15 February 2006 (UTC)

High-frequency behaviour

The article should also include a reference to the transistor behaviour at high frequencies, and the existance of parasite capacitances that influence the device behaviour. Afonso Silva 21:50, 3 February 2006 (UTC)

Alpha

In my ECE book, it says that alpha is different during DC operation and AC operation. It defines α_DC to be I_C / I_E and α_AC to be ΔI_C / ΔI_E. It goes on to say that the ac alpha is formally called the "common-base, short-circuit, amplification factor". This seems to be inconsistant with the article's "α_F. Fresheneesz 20:46, 9 February 2006 (UTC)

Forgive me if you already know this, but if you're ECE, you should get comfortable with the small signal model concept. In analog design, you have steady-state (DC) and small-signal (AC) models. You need DC and AC versions of each parameter, because each one is used for different purposes. DC is needed to find the operating point, output range, etc. AC is needed to find the gain. In reality AC and DC operation are happening at the same time; it's just more convenient to analyze them separately. - mako 21:33, 9 February 2006 (UTC)

picture of the structure section

I've replaced the original picture by a much more simple one, that illustrates the contant of the section much better. If someone disagrees with that, just revert the change.

Thank you! It looks fine. I wanted to do that but I don't have drawing skillz Snafflekid 23:40, 15 February 2006 (UTC)

Ebers-Moll Model

Great article! I'm not overly familiar with the Ebers-Moll model, but in the diagrams, the emitter dependant current source refers to I_CD - which doesn't appear to be present in the model. Should the I_ED on the collector side be changed to I_CD?

I believe so too, I_ED is the current from Emitter to Base, it has no place between the Collector and Base. Also, Why are these currents not named I_EB and I_CB? Sounds more logic since we're talking about the Base, not the Drain. -Pelotas 09:58, 9 July 2006 (UTC)

I think there is a slight error in the unapproximated Ebers-Moll equation for iB, looks like VBC and VBE are swapped.

image

most top image has in npn transistor arrow denoting thats npn but pnp lacks it, and its often used on schematics, maybe some info about or image update could solve that? 193.238.17.211 (talk · contribs · WHOIS)

I've just reverted back to the png images. TBH i'm getting mighty pissed off with people replacing well done pngs with poor SVGs. Plugwash 22:52, 2 May 2006 (UTC)

Can you tell us ignoramuses what the difference is please? 8-)--Light current 23:25, 2 May 2006 (UTC)

The arrow was missing on the NPN transistor. Plugwash 23:34, 2 May 2006 (UTC)

There is also the blurry grayness but thats probablly a result of mediawikis SVG renderer and not something you can do much about Plugwash 23:35, 2 May 2006 (UTC)

So png's are prefereable?--Light current 23:39, 2 May 2006 (UTC)

Well for web use a png designed for the size it will be displayed at beats everthing else. However apparently the svg pushers have convinced the powers that be that svg is better as it scales better for other output formats.

In an ideal world, SVG would be best, because it is a vector format. But for whatever reason, mediawiki's rastering process leaves out the arrows. I discovered this when editing Image:Photomultipliertube.svg; the arrows do show up in Inkscape. - mako 21:29, 3 May 2006 (UTC)

http://bugzilla.wikimedia.org/show_bug.cgi?id=5163 says it's a bug in librsvg. - mako 21:38, 3 May 2006 (UTC)

using alpha and beta

Hello. The parameters alpha and beta appear before any explanation about them is given. Could anyone please write some info about it..? Thanks, 132.68.145.85 19:27, 23 June 2006 (UTC)

Ive changed page now to explain alpha, beta and their relationship. Is it any better now? 8-|--Light current 19:39, 23 June 2006 (UTC)

I've added that there is a big difference between the Beta in forward or reverse operation mode, however I am not entirely sure of the numbers. I believe the ideal forward operation current gain is mostly about 300, but the 100...1000 region is probably correct. In reverse the gain is surely smaller than 1? Also, should not be mentioned that another reason for that big gain in forward operation is the extremely small dimension of the base? - Pelotas 10:02, 9 July 2006 (UTC)

Logarithms

I'd like to know more about using bjts to compute logarithms.

The base-emitter voltage is proportional to the log of the base-emitter or collector-emitter current. A diode can do the same things, as shown in this book: [3].

Current or voltage control - that is the question

Moved here from my talk page;--Light current 19:54, 1 August 2006 (UTC)

Bipolar transistors are fundamentally current-controlled devices, notwithstanding the hobbyist page you cite. I'll sit back and let someone else correct it back for now. Dicklyon 19:14, 1 August 2006 (UTC)

There has been a detailed and exhaustive discussion on this topic well before you joined. The consensus of opinion was that the BJT is voltage controlled. I dont think anyone will revert it. but we'll see ! 8-)--Light current 19:54, 1 August 2006 (UTC)

OK, I read that discussion, but I don't agree that a consensus was reached. I think we need to add a section to the article about voltage- and current-control views. The current-control view is most commonly used in circuit engineering, since beta is a simple linear parameter that describes the transistor well. Dicklyon 19:38, 1 August 2006 (UTC)

I agree that the current control explanation appears to work and describe things, but is it the truth of the matter? I think not! 8-|--Light current 19:54, 1 August 2006 (UTC)

I think the issue of truth is a bit misplaced here. It's all about description of physical processes, in which no part of the situation can be called the cause as opposed to the effect. A better model, if you want truth, is the charge-controlled model, since it explains collector current as a function of minority charge concentration in the base region, and therefore contains photo-transistor effects, turn-off recovery-time effects, etc. Voltage and current are just the terminal properties we use to explain charge in the base region. Maybe we should write that up, too. Dicklyon 19:59, 1 August 2006 (UTC)

Yes the charge control expalnation would be good addition. 8-)--Light current 20:03, 1 August 2006 (UTC)

OK, I added it, to the section on current control and voltage control, as you noticed and already mutilated. Please don't think that the current-control and voltage-control models should be associated with driving the base from a current source or voltage source (the latter being a ridiculous thing to attempt with a BJT, notwithstanding the funny diagram that shows it that way). In general, the transistor is in a circuit that needs to be analyzed as a whole, and the current-control model almost always makes that job easier if the circuit is intended to be approximately linear. Dicklyon 21:24, 1 August 2006 (UTC)

Dick, I would truly enjoy watching you present this view at a design review of 20 professional analogue i.c design engineers of a mainstream i.c. manufacture. There are those that actually routinely design 10,000 transistor analogue circuits, and actually understand what they are doing, and there are those that don’t. There are also those that don’t know that they don’t.

Kevin aylward (talk) 16:16, 22 February 2009 (UTC)

Kevin, I would truly enjoy seeing you act like a mature contributor to wikipedia; what are the chances? Dicklyon (talk) 23:39, 22 February 2009 (UTC)

Sorry Dick. I do not mutilate. I either improve or delete. Misleading info must be exterminated.--Light current 21:26, 1 August 2006 (UTC)

Apology accepted, as Stephen Colbert likes to say. When you changed "(current control)" to "(if unusually driven by a current source)", you destroyed the parallel explanatory structure of the sentence to insert your irrelevant opinion of what kind of base drive might be usual. It makes no sense anyway, as it seems to imply that you think maybe the base would be driven by a voltage source. As I said, how the base is driven is not relevant to the alternative ways of modeling the transistor action. Dicklyon 21:33, 1 August 2006 (UTC)

Yes I can see you are going to be a worty adversary! I look forward with relish!--Light current 23:11, 1 August 2006 (UTC)

I believe it's spelled 'warty'; and there's no need to be advesarial here, just open up your mind to the alternate points of view. That's what I've done by allowing the voltage-control view to stay. Dicklyon 23:54, 1 August 2006 (UTC)

Sorry I missed out the 'h'. my spelling is always terrible . Ask anyone here!--Light current 00:14, 2 August 2006 (UTC)

The Gummel–Poon charge control model subsection is just a stub, but let's leave it and maybe someone who understands it better or has time will elaborate it. Dicklyon 00:36, 3 August 2006 (UTC)

Anyone disagree with this?

See what's happening here? THE TRANSISTOR IS NOT CONTROLLED BY CURRENT. Instead it is controlled by the base/emitter voltage.

   7. THE P-TYPE AND N-TYPE ARE CONDUCTORS BECAUSE THEY CONTAIN MOVABLE CHARGES.

   8. A LAYER OF INSULATING MATERIAL APPEARS WHEREVER P-TYPE AND N-TYPE TOUCH.

   9. THE INSULATING LAYER CAN BE MADE THIN BY APPLYING A VOLTAGE. 

(extracted from Bill Beatys pages at amasci.com)

If so , why? --Light current 23:06, 1 August 2006 (UTC)

If by conductors you mean semiconductors with free majority carriers in them, then yes, 7 is OK. If by a layer of insulating material you mean a depletion region, in which the majority carriers are absent, then yes, 8 is OK. And 9 is OK; you can make the depletion region thin or nil by forward biasing the junction. So yes, that's all fine. However, it does not make a case for "THE TRANSISTOR IS NOT CONTROLLED BY CURRENT". Anyone who uses BJTs knows that you control the collector current by the base current, and that you get a current gain of beta, and that the base–emitter junction has to be forward biased to get such a current.

Oh dear.. .Dicky… err... anyone that is a professional analogue i.c design engineer knows, from trivial inspection of this text, that you are not.

Kevin aylward (talk) 16:28, 22 February 2009 (UTC)

That's my comment you're responding to, not someone named Dicky. But you're right I'm not a professional IC designer at this time, and when I was doing IC design I didn't do a lot of bipolar. And I realize the current-control and voltage-control views are both used. I'm not sure why I stated it above as strongly as I did, three years ago; but the point is that the collector current is based on base charge, which is has its own dynamics, influenced by the B–E junction I and V among other things. Don't push one view to the exclusion of the other, OK? Dicklyon (talk) 17:48, 22 February 2009 (UTC)

As I said in the article, the voltage-control and current-control views are related to each other by the b–e junction V–I curve, and these are both simplifications of the internal processes of the transistor. It is not necessary to deny one to use the other. Dicklyon 23:52, 1 August 2006 (UTC)

If you agreee with 8 & 9, why is a transistor not voltage controlled? PS Pls use my proper name which is 'Light current'. or if you wish to be less formal you can call me 'LC'. Thanks! --Light current 00:19, 2 August 2006 (UTC)

Light voltage, I mean LC, I did not disagree that you can consider a transistor to be voltage controlled. I thought I had stated this clearly; please re-read. I believe it is you who is saying a transistor is NOT CURRENT CONTROLLED. I say it is. Still not clear? Consider re-writing 9 as "the depletion region becomes thin when a forward current is flowing across the junction." Dicklyon 01:01, 2 August 2006 (UTC)

How do you get a current to flow if not by applying a voltage. BTW Dont make fun of my name. You cant afford to with a name like Dick! Think of the fun i could have with that!--Light current 01:05, 2 August 2006 (UTC)

Actually, Dick IS my name. If Light is yours, I'll tred more gently as you must be sensitive about it. How else do you get a voltage on the junction except by putting a current through it? See the symmetry? Dicklyon 01:19, 2 August 2006 (UTC)

Which is more fundamental: current or voltage(emf). ie which comes first?--Light current 09:00, 2 August 2006 (UTC)

I see a bit of confusion here, and it seems to be caused by semantics. People can argue forever over whether the transistor is "voltage controlled" or "current controlled." And each side means something different by that. And each side is trying to win an argument, not trying to clarify matters. (They already know the truth. In any religious battle, the first rule is to never question the viewpoint you're defending.)

Instead, what if we ask a different question: how do transistors work?

In explaining the transistor effect, if we drill down through various ideas, we find a central concept: when the base terminal injects charged particles into the base region, there is no way for the motion of those particles to influence the other charged particles being sent out by the emitter. There is no mechanism by which the base charge-flow can directly affect the emitter's charge-flow. Yet if we inject a tiny base current, we obtain a huge emitter current! The hfe parameter is no illusion. Base current obviously does control emitter current. How? It's because any change in Ib absolutely requires a change in Vbe... and Ie is determined by Vbe because Vbe determines the thickness of the depletion layer. Base current can control emitter current through a multi-step process involving Vbe and the depletion layer. Charged particles injected by the base do in fact influence the charged particles sent out by the emitter. They just can't do it directly.

So is a transistor controlled by current? Yes, if by "controlled," we mean that the emitter current can be changed by changing the base current. Is it "controlled" by voltage? Yes, if by "controlled" we mean that the emitter current can be changed by changing the value of Vbe. And those people who choose one side or the other can fight forever about the "right answer." They'll almost come to blows over the One True Path to transistor enlightenment, since their own viewpoint is obviously good and true and right, while the opposing side are all a bunch of worshippers of Darkness and Confusion whose blasphemy must be ridiculed. After all, young students could be misled from the One True Path! Think of the children!  :)

On the other hand, it's hard to become a victim of either meme if you laugh at religious wars. But that's a very difficult habit to teach. Instead, it's easier to take some devout "Current Believers" and shatter mental grip of that meme by demonstrating an alternate religion. So some become outspoken "Voltage Worshippers." With luck they'll eventually see the folly in either viewpoint, and be immunized against both, yet accept both. The "Transistor Zen" practitioner views BJTs from both viewpoints, switching back and forth as needed, while also seeing the duality as an illusion and rejecting the notion that a single best answer can exist. --Wjbeaty 03:03, 2 August 2006 (UTC)

Very nicely articulated, Bill. But then weren't you the guy with the all-caps denial of the current-control view, who LC quoted? Confess it! Dicklyon 05:26, 2 August 2006 (UTC)

Of course! The voltage-control view is "right" because it gives us a deep grasp of the transistor effect. If we watch the impact that the depletion layer has upon transistor currents, then we can figure out how transistors work. But while designing circuitry, unfortunately the voltage viewpoint is far less useful than the current-control viewpoint, because current-control gives us a powerful rule of thumb called hfe.--Wjbeaty 06:10, 16 August 2006 (UTC)

It is an extract from Bills pages and I should have credited it to him. I omitted to do so. Apologies 8-( --Light current 08:52, 2 August 2006 (UTC)

It seems to me the debate can be settled if we can agree whether the voltage controls the current in a resistor, or whether current controls voltage. If we can sort that out, we can apply the same logic to a diode, and then to the transistor base-emitter junction.  :-) --Phil Holmes 16:34, 5 August 2006 (UTC)

In resistor physics at the micro level, the e-field accelerates the charge carriers, and they in turn start moving. Isn't this a one-way causation? After all, the acceleration of the charge carriers does not create the e-field. (And... while gravity causes dropped rocks to fall, a falling rock does not magically create a strong gravity field.)

On the other hand, transistors are less akin to resistors and more akin to diodes. Perhaps the real question is: are diodes turned on by forcing charges forward through the depletion layer, or are they turned on by applying a forward voltage across the depletion layer? --Wjbeaty 06:10, 16 August 2006 (UTC)

There's no real answer to the question, though. You're very comfortable with voltage as cause and current as effect, but they're both just a relaxation to the conditions imposed by the external circuit. At the electron level, you can think about a field accelerating the electrons, or about the electron distribution creating a field. Don't think it can only be one way. Dicklyon 06:24, 16 August 2006 (UTC)

Its probably energy flow between the wires that causes the effects (current and voltage) at the terminating resistor.--Light current 17:12, 5 August 2006 (UTC)

Does a wheelbarrow accelerate because you push it, or does the accelerating wheelbarrow cause you to be pushing it? Surely the whole point is whether "control" means "physically control" (i.e. a belief of cause and effect) or "mathematically control" (i.e. apply this function to that value). It is very hard to comprehend a transistor being "physically" controlled by current, because our bodily experience tells us that force causes motion/acceleration; therefore voltage causes current, and so transistors are (physically) voltage-controlled. However the mathematics is not tied to a physical interpretation, and by rearranging equations you can model transistors as current-controlled or voltage-controlled; however this is completely detached from a physical interpretation.

I don't really have much bodily experience with pushing via alteration of electric potentials, so they're both abstract to me. It seems foolish to limit our conception of quantum phenomena (electrons) to things that match our bodily experience. Dicklyon 15:28, 30 August 2006 (UTC)

2007 Revival of argument

The BJT is VOLTAGE CONTROLED, period. It works according to generalised Newton’s laws, to wit, F=qE. That is, to make a charge move in the collector circuit, requires a *force*. It is the *electric field* that makes charges move. Current (ignoring magnetic effects), can not make charges move, so the idea that base current causes collector current is simply daft. It has no basis in physics. We can argue that all voltage is due to charge, so charge is the real cause, but this misses the point. We apply a voltage to the base emitter, and the collector current is caused. Whatever base current exists, just comes in the wash.

For those of you that claim that base current causes collector current, please supply a mathematicall derivation of such effect. Additionally, please explain, why base current does not appear in the 1st order expression, Ic=Io.exp(Vbe/Vt) —Preceding unsigned comment added by Kevin aylward (talk • contribs) 12:06, 7 September 2007 (UTC)

Kevin, such strong statements must come with a strong source. What you're saying makes very little sense, since the field the moves charges to the collector has little to do with the physics of charge injection at the B–E junction. The physics of current control is more based on emitter current (which is where the collector current comes from), not base current, but they happen to be nearly lienarly related so they get used interchangeably in models. And since first-order expression omitting any current or voltage are easy to write, that's no kind of argument. In any case, we should not argue, but cite sources for viewpoints. Dicklyon 15:16, 7 September 2007 (UTC)

That is, to make a charge move in the collector circuit, requires a *force*. Nonsense. You've made a serious error in reasoning, my friend, and an ironic one at that since you invoked Newton. The electric field accelerates a charge. The mere movement of charge does not require any electric field (force) at all. Alfred Centauri 23:36, 7 September 2007 (UTC)

Now why did I just *know* that someone would say that. Yeah, so I am ignorant of basic F=ma. Try getting a current in a resister without a force. —Preceding unsigned comment added by Kevin aylward (talk • contribs) 11:52, 10 September 2007 (UTC)

There's some truth on both sides; in a semiconductor, electrons tend to give up their momentum in collisions, so you need a field to make a net drift current, according to the Einstein relation (kinetic theory). But none of this is particularly relevant to the point being argued, which is whether the transistor can be equally well described as voltage-controlled, current-controlled, or charge-controlled. They all work, and to try to make just one variable be the cause and the others the effect is very contrary to everything we know about physical systems. Dicklyon 00:30, 8 September 2007 (UTC)

so you need a field to make a net drift current. Rather, associated with a net drift current in a semiconductor is an electric field. It really can't be argued that this field makes the drift current because it can be equally well argued that the collisions due to the current cause a charge density gradient that in turn causes the field. But, that's precisely the point you made when you said that "to try to make just one variable be the cause and the others the effect is very contrary to everything we know about physical systems", right? Alfred Centauri 01:49, 8 September 2007 (UTC)

Not precisely the point I was trying to make, but OK, close enough; I won't argue. Dicklyon 03:48, 8 September 2007 (UTC)

OK, precisely was a poor choice. But, I do think we're on the same page. The bottom line, IMHO, is that this question of whether the BJT is a voltage-controlled device or a current-controlled device is a false dichotomy. Regards, Alfred Centauri 14:50, 8 September 2007 (UTC)

Grammatical structure

It seems to be conventional to write the terms such as base–emitter incorrectly with a hyphen instead of an en dash. But shouldn't we strive to do it right here, instead of following the typical error, so that the meaning of the terms become more clear at least to those who understand English punctuation? I've made a start in various sections, but it's a pain to fix throughout since my browser's editor won't find and replace. Dicklyon 21:09, 1 August 2006 (UTC)

Why bipolar?

LC reverted my claim that the term BJT comes from requiring junctions of both semiconductor polarities. I found the current claim not credible, because "the main conduction channel employs both electrons and holes" is not true; only minority carriers flow through the base to the collector; or there some other notion of "main conduction channel" that I'm missing. So I'm trying to find a reference, and having trouble. I found this one [4] that says the BJT is a current-controlled resistor, but that sure won't fly with LC.

I'm willing to believe I got it wrong, but it's also wrong as written. Perhaps it can be corrrected to say that both types of carriers are involved, since there is a small component of the opposite type in the base–emitter current. Dicklyon 02:15, 2 August 2006 (UTC)

Ah, I see where it came from. LC references [5] which says "The device is called “bipolar” since its operation involves both types of mobile carriers, electrons and holes." But somewhere along the way the meaning got distorted to "the main conduction channel employs both". Let's fix it. Dicklyon 02:20, 2 August 2006 (UTC)

Just to keep you going, Dick, Ive found a ref that may interest you:

The main difference between field effect transistors and normal junction transistors is that in FETs the current is only carried by the majority charge carriers. Field effect transistors are thus "unipolar" transistors, ie only one type of charge carrier is responsible for their actio, unlike normal transistors, which are "bipolar"

Page 1 para 1, Field effect transistors,(multiple authors) Mullard Ltd, 1972, ISBN 0 901232 42 4.--Light current 10:36, 2 August 2006 (UTC)

However, it's not totally on point, as in BJTs the collector current is only carried by the minority carriers. Still, this may be the right reason, if we just phrase it right. It would be nice to see it in a book on BJTs. Dicklyon 18:48, 2 August 2006 (UTC)

Dont forget that the term Bipolar Junction Transistor is a relatively recent term. Originally they were just called 'normal' transistors or perhaps 'junction' transistors to diff them from point contact types (perhaps). THe term 'bipolar' in this usage aint been around too long!--Light current 21:26, 2 August 2006 (UTC)

Good point. I was able to look through my old transistor books today, including Shockley's Electrons and Holes in Semiconductors with applications to transistors, and all seven editions of the GE transistor manual, and didn't find bipolar in any of them. I wonder who coined it. I seem to remember a book on BJTs I used in college in the 1970s, but it's missing. Dicklyon 23:55, 2 August 2006 (UTC)

Aha! I have that classic book also! Must get round to reading & digesting it sometime. The term must have been coined when FETs became popular I would think 8-)--Light current 23:58, 2 August 2006 (UTC)

Looks like we're wrong about how recent the term is. I found this from 1958 in Google Book Search:

"The operation of the field-effect transistor depends essentially on the presence of just one type of carrier, the majority carrier. For this reason it may be called a unipolar transistor. The junction transistor, however, depends for its operation on both types of carrier and is hence a bipolar transistor."

p. 520, vols 2-3 (unclear if this means two volumes bound as one?)

Biondi, F.J., H.E. Bridgers, J.H. Scaff and J.N. Shive (eds.) Transistor Technology, 1958.

Dicklyon 00:27, 3 August 2006 (UTC)

Interesting. What I was saying tho was that common usage of the term is relatively recent (say last 20 yrs). Im sure we didnt use this term when I was studying.--Light current 13:24, 3 August 2006 (UTC)

More like 35 years and I would agree. I studied from a book entitled something like "BJTs, FETs, and Microcircuits" around 1971. I just looked it up, and it's by E. James Angelo, 1969. Perhaps that started the common use of the term, by teaching it to college students. Dicklyon 18:33, 3 August 2006 (UTC)

I studied from a book called 'Integrated electronics' by Millman and Halkias pub. 1972. THey do mention the term BJT in the introduction to the transistor chapter, but everywhere else, its just 'transistor'. --Light current 22:40, 3 August 2006 (UTC)

I have referred a Book called "Solid State Electronics Fundamentals - Chih Tang Sah " here he has explicitly mentioned that Bipolar is because of conduction in BJT is due to "Majority"(In collector and Emitter) & "Minority"(In base) charge carriers and not anything else.

Suggestions for this article

In scanning over this article, it seems that it is very heavy on BJT usage in circuit topology and general application, but extremely light on the actual device operation and the various processes that govern its behavior. For example, the sadly short "Explanation" section discusses an NPN BJT in forward active mode but never makes it lucid to the reader that this is the case. I think this section should be rewritten and generalized into something better. I'd propose bringing in a simple energy band diagram as well as one or two figures showing carrier distributions under the common operation modes (active, saturation, and cutoff; inverted mode is rarely used). Personally, I find it highly useful to visually show the six or so significant current components within an active-mode BJT when explaining its operation.

Speaking of which, this article is nearly totally about BJTs in active mode (as it probably should be), but never makes it clear that this is the only mode that really provides useful current gain or why. That seems to be a fundamental failure since the primary use of BJTs in modern devices is as a high gain amplifier. The article seems often guilty of making statements without explaining them. For example, the section about the Ebers Moll model mentions "normal" operation but doesn't define what "normal" is. From the rest of the article, the reader might be led to assume that "normal" means "forward active". We also have very brief mention of junction breakdown without any explanation of the basic breakdown mechanics. These should at least be briefly summarized, perhaps with a nod to the PN junction article (assuming there's a decent explanation of breakdown mechanics there... I haven't looked).

Then too, there is the discussion of the hybrid parameter BJT model. Therein it is explained that the model is useful for "small signal" analysis, but never what small signal conditions are or why they must be maintained to keep the transistor in forward active mode (it also neglects to mention that the H-parameter model is only useful in active mode). Maybe a simple common-emitter single-BJT amplifier with a resistor biasing network would be good for illustrative purposes here? I dunno; I'm admittedly more keen on device operation and design than circuit topology. A full discussion of a single transistor amplifier might be too far out of this article's scope anyway.

Anyway, I just wanted to vent some of my complaints with this article before I messed with it too much. I made a few changes here and there and decided I should check in on the talk page before going much further. I don't want to step on anyone's toes or anything, just make some needed improvements to the article. If nobody has any objections, I'll probably try to write a decent explanation of BJT operation in terms of carrier concentrations and movement. Of course, I'm far from an expert on semiconductor physics (just a humble research student), so your feedback is greatly appreciated! -- uberpenguin @ 2006-08-22 05:49Z

Go for it. But if it gets real technical, try to keep that in subsections such that a more general audience can still find the article useful. Dicklyon 06:23, 22 August 2006 (UTC)

NPN BJT under equilibrium.

Okay, I threw together a simple energy band diagram for an NPN BJT under equilibrium. If you like the style, I'll also make one for active mode and use these and perhaps a minority carrier concentration diagram to help illustrate things. I don't plan to include all this in the top section, but in a subsequent section that goes into a little more detail about the device's operation. Anyway, let me know what you think. -- uberpenguin @ 2006-08-22 23:54Z

Any correct diagram is a worthwhile addition to an article. So, yes please!--Light current 00:19, 23 August 2006 (UTC)

NPN BJT in active mode.

Cool... Here's the diagram for forward-active mode. These should help a lot in explaining the device operation. The only thing I'm concerned about is that a lot of the terminology that must be used to describe BJT operation is heavily rooted in semiconductor physics, and I'm not sure whether there are any decent articles on Wikipedia that explain the transport mechanics and current components in semiconductors or the specifics of PN junction operation. Is it okay for the purposes of this article to just link where possible and hope that those articles get improved eventually? -- uberpenguin @ 2006-08-23 01:58Z

Yes I would say so.--Light current 02:09, 23 August 2006 (UTC)

I know nothing about transistors and I didn't find this page as useful as I had hoped. The non-technical section really needs some help in order to be useful for the general audience (maybe put it in terms of when this wire is 'on' the other wire is 'off' or something). The explanation makes this idea more complicated than it has to be. Thanks everyone!

Proposed merge of PNP and NPN

I see no reason for there to be two articles that should at most be a couple of sentences in this one. -- mattb @ 2006-09-11T03:31Z

• Support. I agree, merge them in here. Dicklyon 03:37, 11 September 2006 (UTC)
• Support — Omegatron 04:44, 11 September 2006 (UTC)
• Support. I felt rather silly making the NPN article, even though it was requested. — Matt B. 05:20, 11 September 2006 (UTC)
• Support CyrilB 12:13, 12 September 2006 (UTC)
• Support Lincher 03:13, 17 September 2006 (UTC)
• Support --Light current 23:03, 29 September 2006 (UTC)

Done. -- mattb @ 2006-09-29T23:10Z

Regions of operation

This whole section is sort of a mess, and I thought we might as well talk about it since there seems to be a disagreement. The section is very confusing and, on the whole, misleading because it sort of combines an explanation of operation regions with small signal modeling. In reality there are four well defined regions of operation (or biasing modes) for a BJT:

• Forward active — EB junction forward biased, BC junction reverse biased
• Reverse active (inverted) — EB junction reverse biased, BC junction forward biased
• Saturation — Both junctions forward biased
• Cutoff — Both junctions reverse biased

"Linear" is NOT a BJT operation region (it is for the MOSFET). I can only assume that whomever wrote that was combining or confusing small and large signal models. Operation mode of a BJT depends on nothing but the biasing of the two junctions; things like external current limiting circuitry do not directly affect the operation region (unless they change the biasing of one of the junctions). Most of the time BJTs will be biased squarely into the forward active region, wherein the common small signal models later described in the page actually apply. -- mattb @ 2006-09-29T22:49Z

Cut-off region

While it is true that a transistor is in cut-off region when both junctions are reverse biased, that is not as strictly necessary condition. Furthermore, it is not even typical. Consider, for example, the grounded-emitter switch configuration. When the base is at or near ground, it is in cut-off; it is not possible to reverse bias the base–emitter junction in a circuit that can't pull the base below ground. So let's just say what's true and verifiable, OK? Dicklyon 20:53, 30 September 2006 (UTC)

As I stated in my edit comment, your view seems to reflect a simplified CVD model of the junctions in a bipolar transistor, not the actual physics of the device. If you'd like, I would be happy to provide references from three modern semiconductor physics textbooks and one modern microelectronic circuits textbook that verify exactly what I asserted. I think this is merely a conflict of terminology understanding, since some googling confirmed that a few sources claim that cutoff exists for small forward bias on either junction. However, this isn't correct usage of terminology and is not claimed by any semiconductor device design book I'm aware of.

Taking the example you gave (if I understand you correctly, and assuming an NPN device), the transistor is actually in forward-active mode for small input voltage. Since $V_{BE}$ is very small (as you said, close to ground potential), very little electron current diffuses across the EB junction, and there is therefore a relatively small minority carrier perturbation in the base to provide carriers to be swept over the BC junction to the output. Now, once the potential at the base becomes larger than a few multiples of the thermal voltage (${kT}/q$), the current through the EB junction quickly becomes significant. This current is amplified and a larger current is pulled through the collector and load. Obviously this causes a voltage drop at the collector, which quickly ends up at a lower potential than the base as the collector current rises. When this occurs, the device goes from forward active to saturation and the collector current doesn't change much with increasing $V_{BE}$. So you see, the simple emitter-grounded BJT switch is in forward-active when "off" and saturation when "on". The "turn on" voltage of such a switch is a few times the thermal voltage, which is approximately where diffusion current starts to dominate over the small R-G current in the EB junction.

I encourage you to look at how TTL works; it uses transistors primarily in saturation and reverse-active mode, but not cut-off. Anyway, to show good faith, I will not make any further changes to the section until we've talked things out to your satisfaction. Please let me know if you'd like me to cite those textbooks I mentioned. I'd also be happy to sketch some potential diagrams to illustrate what I have explained if that will help. I realize that you have much more electronics experience than myself, and I mean no disrespect to you. However, semiconductor device research and physics is my area, I think I've provided a clear enough explanation, and I'd be happy to cite some sources that confirm proper usage of the four BJT biasing regions' terms. -- mattb @ 2006-09-30T23:30Z

Matt, since in your first couple of rounds you left the statement "the base–emitter voltage is too small for any significant emitter current to flow. In typical BJTs manufactured from silicon, this is the case below 0.5 V or so," I assumed you had no quarrel with that; your additions narrowed the scope of cut-off a lot from there. So, which is it? I do understand semiconductor physics well enough, but there may be a disagreement over the terminology, among various texts. I can see the logic of what you're saying, which is that it's in forward active region even when the current is too small to measure, by far. Please do show me the references. I showed a couple that disagreed, and didn't find any that defined cut-off the way you want to. I did find some that said the transistor is in cut-off when the junctions are reversed biased, which is obviously true, but they did not specifically say that the cut-off region is limited to that condition. That is, reverse bias is sufficient, but perhaps not necessary, for cut-off. Who is the authority on such terminology? Furthermore, if we decide to define cut-off that way, should we then change the part about transistors used as switches, since in many cases (e.g. H-bridge drivers) they will never be in cutoff by that definition? Dicklyon 23:42, 30 September 2006 (UTC)

Dick, I cited my four references in the below section. If you'd REALLY like, I'll try to make some photocopies of relevant pages from the texts, but that will take some time (two of the books are lent out to friends right now and I don't own a scanner) and I'd rather not have to do it if possible. I don't have any quarrel with the statement you quoted, but the magnitude of current flow does not determine the biasing modes of the BJT in any semiconductor physics text I have seen. A quick google search found numerous online documents that confirm my voltage-only definition of biasing modes, but I hate referencing online documents, which is why I mentioned the text books. It seems that the only sites which agree with the current-related definition of biasing/operation regions are those that use a CVD-like simplification for the transistor. There's nothing wrong with that, of course, because that's a useful simplification for small signal modeling, and obviously small signal models are what we typically use for real electronics design, not the full mathematical models. However, I have never seen operation modes defined with anything but the signs of $V_{EB}$ and $V_{BC}$. Personally, I think these definitions make the most sense because they directly define the positioning of quasi fermi levels within a device, which ultimately is what determines the rates of drift, diffusion, and R-G current. Of course, at a circuit level potential and current are so closely related that it's a difficult distinction to make, so I certainly acknowledge that there's a fine line to tread here. Anyway, I'm sticking with my definition as being backed up by a number of semiconductor physics texts, and I will provide copies of my references if that's what it will take to change your mind. -- mattb @ 2006-09-30T23:57Z

P.S. - I'm not contesting at all that the transistor is being used as a switch in the described configuration, just how we define the device biasing regions. There's obviously no perfect solid-state switch (especially with a bipolar device), and the small "off" base current is negligible regardless of what we choose to call the transistor's biasing mode at that point. -- mattb @ 2006-10-01T00:03Z

P.S. (again) - Perhaps we can strike a balance between the voltage/current definitions. The more I think about it, the more it seems to me that the voltage explanation makes slightly more sense and is easier to quantify, but the current definition has more to do with the operation of the device within a circuit and is therefore equally valid. Would it be acceptible to you if I added something like "the biasing regions overlap for very small applied voltages" and briefly explain that this has to do with the exponential nature of the I-V relationships? -- mattb @ 2006-10-01T00:26Z

Proposed "regions of operation" section

I was going to change the regions of operation section to the following, but I won't do it if I'll just be reverted on the spot, so I'll leave it here until we work this out. I reference "Semiconductor Device Fundamentals" by R.F. Pierret, "Physics of Semiconductor Devices" by S.M. Sze, "Semiconductor Physics and Devices" by D.A. Neaman, and "Microelectronc Circuit Design" by R.C. Jaeger and T.N. Blalock. All of these quite clearly agree with my proposed text. I also reference google, which will quickly verify the biasing conditions for the four BJT operation regions. I can provide photocopies from some of these books upon request, if necessary, but I encourage you to just google it since a quick search for "bjt biasing modes" will verify all of this (ignoring the handful of sites that use a CVD model simplification for teaching transistors in circuits without bothering with actual device physics... a pox on them for the "lie to students" method of simplification).

Anyway, the proposed text follows my signature. -- mattb @ 2006-09-30T23:40Z

---

Bipolar transistors have four distinct regions of operation:

• Forward-active (or simply, active): The emitter-base junction is forward biased and the base-collector junction is reverse biased. Most bipolar transistors are designed to afford the greatest common-emitter current gain, $\beta _{f}$ in forward-active mode. If this is the case, the collector-emitter current is approximately proportional to the base current, but many times larger, for small base current variations.
• Reverse-active (or inverse-active or inverted): By reversing the biasing conditions of the forward-active region, a bipolar transistor goes into reverse-active mode. In this mode, the emitter and collector regions switch roles. Since most BJTs are designed to maximise current gain in forward-active mode, the $\beta _{f}$ in inverted mode is very small. This transistor mode is seldom used, usually being considered only for failsafe conditions and some types of bipolar logic.
• Saturation: With both junctions forward-biased, a BJT is in saturation mode and facilitates high current conduction from the emitter to the collector. This mode corresponds to a logical "on", or a closed switch.
• Cutoff: In cutoff, biasing conditions opposite of saturation (both junctions reverse biased) are present. There is very little current flow, which corresponds to a logical "off", or an open switch.

While these regions are well defined for sufficiently large applied voltage, they overlap somewhat for small (less than a few hundred millivolts) biases. For example, forward active mode may be alternatively viewed as cutoff mode for very small biasing of the emitter-base junction or as saturation mode for very small biasing of the base-collector junction. However, transistor behavior in these regions is usually only considered in transient analysis and is avoided for principle device operation.

---

I cant see anything wrong with the above statements. 8-)--Light current 00:18, 1 October 2006 (UTC)

I added a little bit about region overlap, which will hopefully address Dick's concerns as raised above. -- mattb @ 2006-10-01T00:32Z

That all looks good and fine, provided you show me the refs where that's how cut-off is defined. Dicklyon 00:45, 1 October 2006 (UTC)

Cut off (or the off state) is defined in the Motorola switching transistor handbook (1973) as where there is minimum current flowing through the transistor (C to E) and there is maximum voltage (Vce) across the transistor. Is that a satisfactory defn?--Light current 00:55, 1 October 2006 (UTC)

That's a mixture of both definitions, which is equally valid. I'm still voting for the applied junction voltage sign definition (with the note about region fuzzyness), however, if only because it's valid and it's a very simple and methodical way of explaining things. -- mattb @ 2006-10-01T01:03Z

Very valid, simple, and methodical is nice, if you're writing your own text. But here we need to document the actual use of the term in the field, not clean it up to suit. So we need those refs. Dicklyon 01:10, 1 October 2006 (UTC)

On the one hand, you're right, but on the other there is a balance to be struck here. This is (supposedly) an encyclopedia article, and is therefore an overview and summary of the topic. Simple and methodical (as long as its correct) is often preferable to a full explanation for obvious reasons. The biggest thing I've gotten out of this dialogue is that the definitions of the operation regions are less universal than I had originally thought, so I don't really see a problem with us picking a simple explanation as long as it is a correct one. Industry usage? I learned analog electronics from a guy who was a senior analog engineer at TI for several years. He never indicated that there is any other way to reference biasing regions than the voltage-sign method. I learned semiconductor device operation and physics from the guy who I now work for, someone whose research is all about semiconductor devices (HBTs, APDs, MEMS). Same story. If you have alternative experiences to relate, I'm all for hearing them. I'm certainly not suggesting that alternative definitions of the biasing regions are less valid, far from it. I'm merely asserting that for two equally correct explanations, the simpler one is preferable. -- mattb @ 2006-10-01T01:49Z

Okay, after a hunt for a cable that can only be described as "epic" in scale (stupid proprietary connector), I took some pictures of the two of my four references that I have on hand. I apologize for the bad flash, but this is a really terrible camera. You should still be able to read the important parts of the text. Pierret: [6], Jaeger: [7] [8]. If you really want to make me work hard, I can try to hunt down my other two sources as well, but I hope you'll trust that I'm not lying to you and accept these as well as the google results. -- mattb @ 2006-10-01T01:30Z

Matt, good job. I just wanted to know what the references really did say exactly. The photos were one way to show me. Be sure to ref those, or at least one, in the article. One thing still unclear to me is how you get the pulldown transistor to have a reverse-biased B-E junction in a TTL logic gate. The paragraph that spans the Jaeger pages says the cut-off region is most commonly used in logic switching as TTL, but I don't see how it can be with that definition. Same goes for RTL and DTL circuits; all use the grounded-emitter configuration, where the B-E junction can never be reverse biased. So this story retains a nagging inconsistency. Pierett is not so explicity self-contradictory, but close. Dicklyon 05:53, 1 October 2006 (UTC)

Here's a twist: this book says that VBE < 0.6 V is "reverse biased". Is that the way you're thinking of it? I always assumed that reverse biased meant a negative relative potential, but this is one way to make the definition as used in your references not contradict the idea that the cutoff region is used in logic switching. Dicklyon 06:05, 1 October 2006 (UTC)

Hmm... Well that's a rather interesting way of looking at it. I guess they are going from the perspective that the junction's built-in potential works against applied bias and are calling "zero bias" what I've usually heard termed "flatband voltage". However, I've never heard anyone call bias anything but applied voltage; all of the equations I've ever used explicitly separate biasing voltage and built-in voltage. I think that book is providing a very non-conventional explanation.

Anyway, I'm not entirely sure how to answer your logic question. Digital logic frankly isn't my thing. I thought that under TTL the BJTs are in saturation and reverse active mode. I know ECL uses BJTs in cutoff, but I don't really have an off-the-cuff answer for you there... I guess I could thumb through the relevant pages to see if I can find an explanation. -- mattb @ 2006-10-01T06:18Z

So is it okay for me to copy my proposed text into the article, or is there anything else that we should talk about first? -- mattb @ 2006-10-02T19:53Z

I would still feel better if you added text to indicate that the definition of cutoff is sometimes extended to include the region of essentially zero current even if the junction biases are zero or slightly forward. This makes it compatible with statements of cutoff as the "off" state in logic circuits. Or, you can go ahead and add your text, and I might "improve" it in that direction. Dicklyon 20:40, 2 October 2006 (UTC)

Well, that's what the text about region overlap was intended to address, but feel free to modify the text once I add it to the article. I'd say that zero bias isn't a situation worth talking about much since it is only theoretically attainable (that is, there will always be some tiny potential difference in a circuit). We talk about it when describing device state in equilibrium, but the entire "operation/bias region" notion implies a non-equilibrium condition. -- mattb @ 2006-10-02T21:11Z

Some questions about convention

In my recent edits to this article I've noticed that convention is sometimes switched around. I think we should introduce and adopt convention for common terms in this article, if only to save us the trouble of typing very common phrases out over and over.

My suggestions:

• Use NPN transistors for most of the examples. We'd need only have a few lines explaining what the primary differences between the two types are. The operation region section should also note that "forward" and "reverse" biasing is always with respect to the p side of the junction.
• NPN convention means that we'd probably use $V_{BE}$ and $V_{BC}$ to denote junction voltage (unless explicitly talking about PNP), since these are positive for NPN forward biasing.
• Similarly, $I_{CE}$ for collector-emitter current since this is the direction of the majority conventional current flow in an NPN device in active mode.
• The emitter-base/base-emitter junction will be referred to as the "BE junction"
• Collector-base/base-collector: "BC junction"

Since these conventions are pretty arbitrary, I'd be happy with pretty much anything so long as its consistant, so let me know. I think maintaining consistancy with these terms will help the article's readability a good bit. -- mattb @ 2006-10-01T00:41Z

Collector base junction is normally refered to as the CB junct here. 8-)--Light current 00:46, 1 October 2006 (UTC)

Again, it's arbitrary, though I think it makes more sense to reference them with respect to the p sides. -- mattb @ 2006-10-01T00:48Z

Yeah but its nrmally refeered to as the Collector- Base junction in all the literature Ive read!--Light current 00:56, 1 October 2006 (UTC)

Well, in general: $V_{XY}=V_{X}-V_{Y}$, $I_{X}$ is the current going into terminal X. - mako 08:02, 3 October 2006 (UTC)

Yes, but I'm asking to agree upon a uniform convention for what to call the junctions. That's a little more arbitrary, though I think it makes the most sense to reference them P-N order. -- mattb @ 2006-10-03T16:48Z

P-N order? You mean like B-E for NPN and E-B for PNP? Seems nutty. Dicklyon 19:31, 3 October 2006 (UTC)

My previous comment was addressed to your bullet points 2 and 3. Regarding the junctions, my textbooks seem to be inconsistent in convention, but one simply calls them the emitter and collector junctions. - mako 19:46, 3 October 2006 (UTC)

Well if it seems nutty to you, can you suggest an alternative? As I said earlier I don't care very much about exactly WHAT convention we use as long as the article is consistant, which it currently is not. -- mattb @ 2006-10-03T20:44Z

Pick one. Cite what book you're picking it from. I haven't seen the one I called "nutty"; have you? Dicklyon 21:40, 3 October 2006 (UTC)

Hybrid PI Model

Would someone be willing to write about the Hybrid PI small signal model for the BJT (to go with the other models described here). The small signal approach is a very common way of modeling many nonlinear electronic devices, and the BJT should make a prime example of this. Any comments? --Dirkbike 04:09, 3 October 2006 (UTC)

The h-parameter model described IS basically the hybrid pi small signal model. The section could definitely stand for some cleanup, though. -- mattb @ 2006-10-03T04:42Z

Just wanted to check why my edit adding a note on biasing techniques was removed? Is there some other article where this should be linked? Any other reason? Xcentaur 06:34, 16 October 2006 (UTC)

Well, the collector current article pretty much has no business existing, and I've since changed it to a redirect to this page. Furthermore, it's bad writing style to announce what you're about to enumerate. Especially if it's already been explained in the preceding text. What's more, the wikilink to the biasing article was malformed and the article it links to is of rather poor quality as far as scope and tone go. -- mattb @ 2006-10-19T00:17Z

Regarding inverted mode biasing...

Some recent additions by an anon editor point to some apparant advantages of inverted mode biasing. Frankly I'm a little skeptical of the claims, but I didn't want to revert on sight partially because I don't really understand what the text is trying to convey... Here's the contentious part:

"However the transistor in this mode can be closed more "tightly", the remaining weak collector current being up till one order of magnitude smaller than during the normal wiring. Because of this the inverted wiring is sometimes used in transistor detectors"

The given reference is a rather oldish book (which may explain the germanium transistor example) which I do not have access to. Comments on this? I really can't wrap my mind around how purposely designing with a transistor in a configuration it isn't designed for could be of benefit. I've only ever seen inverted mode used for adding fail safes to circuits, and even then it's rare. -- mattb @ 2006-10-19T00:13Z

Anybody? I'll delete the text if nobody has any explanation of why it should stay. -- mattb @ 2006-11-02T01:45Z

The text you quote is gobbledegook. It needs deleting. Be my guest! 8-)--Light current 01:48, 2 November 2006 (UTC)

I agree. If there's a good source for this idea, let's see it and see what it's trying to say. Dicklyon 02:00, 2 November 2006 (UTC)

Well OK we could make something up to make this make sense. But its not worth it. Its talking about the low leakage phenomenon in inverted mode transistors to enable them to operate as more efficient detectors I think. But .do we care...??--Light current 02:51, 2 November 2006 (UTC)

While we're at it, I've never seen impact ionization effects described as a region of operation. Along with zener breakdown, avalanching is almost always classified as a non-ideality of the device, not a biasing region. Can you show some sources that classify breakdown as a region of operation? -- mattb @ 2006-11-02T02:56Z

Certainly! High speed switching transistor handbook, Motorola 1973 (8th printing). Chapter 9: Avalanche mode switching.--Light current 03:14, 2 November 2006 (UTC)

Forgive me for remaining skeptical, but none of my literature uses this sort of terminology, and google is a little light on it as well. I'll wait for a few other opinions, though. -- mattb @ 2006-11-02T03:40Z

Let me briefly express my concern a little better. If we were to call avalanching (or how about "breakdown" since the Zener process often is not insignificant) a mode of operation, why not also call high frequency gain cutoff a mode of operation? How about the catastrophic high-current-magic-blue-smoke-releasing mode of operation? You'll forgive my sarcasm, but the point is that modes of operation are arbitrarily defined by some standard, and the standard I've always seen used doesn't really take into account non-idealities of the transistor and mostly concentrates on biasing conditions. -- mattb @ 2006-11-02T03:44Z

Subscript T on alpha

Where does the T come from? I understood the F for forward, like on beta. Why the change to T? Dicklyon 17:36, 10 December 2006 (UTC)

I've seen common base gain written with several subscripts, but upon reading the section I've realized that there is some confusion of terms here. The article as it currently stands is incorrect in a couple of places. $\alpha _{T}$ is usually the symbol (I've seen) used for the base transport factor, and is almost correctly defined. It is actually the proportion of the majority carrier current component in the collector to the majority carrier current component in the emitter. So for a PNP transistor, $\alpha _{T}={I_{Cp} \over I_{Ep}}$. The bigger problem with the article is how it defines $\beta _{F}$ in terms of $\alpha _{T}$. This is incorrect since $\alpha _{T}$ is NOT the same thing as common-base DC current gain. If we use $\alpha _{dc}$ for the common base current gain (or whatever subscript you fancy), $\alpha _{dc}=\gamma \alpha _{T}$. $\gamma$ is the emitter injection efficiency; the ratio of the majority carrier current in the emitter to the total emitter current, or $\gamma ={I_{Ep} \over {I_{Ep}+I_{En}}}$ for the PNP. Obviously this is very close to 1 in a typical highly doped emitter. Now, defining $\alpha _{dc}$ in this manner, the relationship between "alpha" and beta as given in the article is correct: $\beta _{F}={\alpha _{dc} \over {1-\alpha _{dc}}}$. Going back to the earlier alpha definitions, you can see that $\alpha _{dc}={I_{Cp} \over I_{E}}$ (again for the PNP). So there's obviously a mix-up of terms in this section, especially as regards the base transport factor and the common base DC current gain (which are not the same thing). I'll try to fix things. -- mattb @ 2006-12-10T19:49Z

P.S. - You'll notice I threw the word "approximately" into the article text a few times. This is because these simplified relationships do not take into account unwanted current flow between the collector and base due to thermal generation at that reverse-biased junction. Normally you can ignore this current component since it is very small, but for very tiny base current, RG currents dominate and you can observe their effects. -- mattb @ 2006-12-10T20:10Z

Matt, do you have a reference for your definition of alpha? I always thought it was just the C/E current ratio. Dicklyon 21:01, 10 December 2006 (UTC)

Remember, there are two "alphas" here; the one you're defining is the DC common-base current gain (neglecting very low-current effects). The gist of all that I wrote is that $\alpha _{T}$, the base transport factor, is not the same thing as $\alpha _{dc}$ or $\alpha _{F}$, the common-base DC current gain. As for sources... I'll go with the three textbooks I have; Semiconductor Device Fundamentals (Pierret 1996), Microelectronic Circuit Design (Jaeger, 2004), Semiconductor Physics and Devices 2nd Ed (Neamen, some year I don't remember). You can likely google the terms "base transport factor" and "emitter injection efficiency" to verify what I've said, as well. -- mattb @ 2006-12-10T22:41Z

Sounds good, if we make it clear that there are several, and put back the usual one that's 1:1 with beta. Dicklyon 23:33, 10 December 2006 (UTC)

Reading difficulty

Holy crap guys, even the discussion page is largely unreadable. I'm not an electrical engineer. Can someone please translate some of the pertinant parts of this page into english.216.57.70.226 22:52, 15 March 2007 (UTC)

Can you be more specific? This isn't the Simple English Wikipedia, so the article cannot be expected to omit technical terminology where appropriate. I don't think that this article is excessively jargony, but it isn't written for the lay man. Some understanding of basic electromagnetics, semiconductor physics, and circuit theory is probably required to totally understand the contents, and frankly one can't hope to gain a good understanding of what a BJT is and how it works without some prerequisite knowledge. I would stick to the transistor article if you're looking for a less technical overview; I think that article does a slightly better job explaining the broad picture. -- mattb @ 2007-03-16T02:32Z

more specifically in the intro paragraph I'd like a sentence or two written for a layman describing the objects function. a layman shouldn't have to know to go to the other description of this object on the transistor page to get a simple answer.

from that paragraph I currently understand:

• it has three-terminals(legs)
• what its constructed of
• some applications where it is used
• why it is called what it is called
• why Charge flow changes in it
• that it is unlike a unipolar transistor
• how it is classified

why would a layman want to use it in a circuit? you apply current and then what do you get? how is it useful? —Preceding unsigned comment added by 173.59.6.62 (talk) 11:41, 7 June 2010 (UTC)

Theory and modelling section

I think this section is getting a bit too long. I suggest moving it to a new article on BJT models and just leaving a more condensed version. I'm planning on writing an article on the hybrid pi model (to avoid the redudant parameter explanations on the common emitter, common base, etc. pages) and logically it should be grouped with the h-parameter model currently in this article. Problem is its gonna make the article even more overwhelming, furthering the need for a new one. Any ideas? Roger 20:18, 11 May 2007 (UTC)

Don't be afraid to split stuff into new articles. — Omegatron 02:18, 12 May 2007 (UTC)

I tend to agree with splitting off transistor models into a separate article in order to keep this one a high-level overview. Perhaps just leave behind a short section summarizing the major models and their uses. You don't want to bog down the primary point of an encyclopedia (overview) article with a lot of details. -- mattb 02:51, 12 May 2007 (UTC)

Just split off the Ebers–Moll model to a main article, as is done with the Gummel–Poon model. Maybe also with the h-parameter model if it gets too big. Dicklyon 04:09, 12 May 2007 (UTC)

Small signal amplifiers

Hi. I just removed the following from the biasing article. Perhaps it should be added somewhere in this article (maybe under a section on applications of BJTs).

---

Configurational Bias

There are three configurations of a transistor. Configuration is the method of connecting any one terminal of transistor common to both input and output circuits. The three types are listed below.
1. Common base (Grounded base) (CB)

Here base is common to both input and output. Emitter-base jn is fwd-biased and collector-base jn is rev-biased.
See full article: Common base

2. Common emitter (CE)

Here emitter terminal is common to both input and output. Emitter-base jn is fwd-biased and collector-base jn is reverse biased.
See full article: Common emitter

3. Common collector (Emitter follower) (CC)

Here collector terminal is common to both input and output. Load is connected to emitter.
See full article: Common collector

Comparison

Parameter	CB	CE	CC
Phase shift between input and output	zero	180	zero
Current gain	less than 1	high	high
Voltage gain	high	high	less than 1
Power gain	moderate	high	low to moderate
Input resistance	low	moderate	high
Output resistance	high	moderate	low

---

Category: Transistor types

Category:Transistor types has listed NPN and PNP - both of which are disambiguation pages.

Should this be fixed by creating NPN Transistor and PNP Transistor as redirects, and adding them both to Category:Transistor types? --RobBrisbane 12:59, 16 June 2007 (UTC)

I think that is a great idea. NPN transistor and PNP transistor might be better names though. -- RTC 07:34, 17 June 2007 (UTC) Especially as one of them already exists as a redirect. -- RTC 07:37, 17 June 2007 (UTC)

Done. -- RTC 16:00, 19 June 2007 (UTC)

Voltage or current redux

I have again reedited back in a correct account of BJT operation. As per wikipedia citation guidelines, I have referenced a professor at a noted university stating that the BJT is voltage controlled. If a user desires to include text that states that the BJT is current controlled, please cite a respected source that supports that claim, and the details of the phyiscs that supports that claim. Kevin aylward 15:02, 8 September 2007 (UTC)

I have reinserted my removed edits on voltage control as, quite frankly, what I say is the way it is. Further explanation can be found at http://www.kevinaylward.co.uk/ee/voltagecontrolledbipolar/voltagecontrolledbipolar.html. I note Bill Beaty has previously tried to rectify this misunderstanding on how the BJT actually works, many would be wise to take note. Kevin aylward 09:51, 8 September 2007 (UTC)

I have now also added an email respose from Bart Van Zeghbroeck, professor of EE, which should be the final nail in the coffin of those current controlled adherents. Kevin aylward 12:29, 8 September 2007 (UTC)

This was already touched on in #Voltage_or_current.3F. We should definitely rectify any and all misunderstandings of true physics principles (as Beaty is usually very good at), but the way you're writing about it in the article is unacceptable. See WP:SOAP, WP:COI, and so on. If you're intent on "burying" your "opponents", Wikipedia is not the place for you.

If you're willing to cover this in a neutral, authoritative manner, let's discuss how best to proceed.

(Beaty's explanation) — Omegatron 16:46, 8 September 2007 (UTC)

Violations of WP:NPOV, WP:SOAP, and WP:COI by editor Kevin alward

I have reverted [9] the recent edits by Kevin alward for violation of WP:NPOV, WP:SOAP, and WP:COI (linking to his own webpage to support his edits). —Preceding unsigned comment added by Alfred Centauri (talk • contribs) 13:46, 8 September 2007 (UTC)

I have reverted back to my edits. As per wikipedia citation guidelines, a professor at a noted university has been cited stating that the BJT is voltage controlled. If a user desires to include text that states that the BJT is current controlled, please cite a respected source that supports that claim.

The link to my page is in the external links section, it was not in direct support of my claims, therefore does not violate the link to own page. The support for my view is

It is a NPOV as it is the view held in any and all academic physics texts book. To the contrary, there are no authority citations showing that the BJT is current controlled. Look, its

F=q(E + vxB)

Its that simple. The electric field makes charges move. A flow of base charge can not cause the collector current.

The non NPOV is the current controlled view, since there is no academic support for such a view. Kevin aylward 15:09, 8 September 2007 (UTC)

Kevin: your edits were reverted for numerous reasons. Please see your talk page for details. Now, to address your argument.

Look, its

F=q(E + vxB)

You seem to have forgotton that the flow of majority charge carriers from the emitter to the base is due to diffusion, not the Lorentz force from any applied voltage. This is fundamental, Kevin. Let's review: injection of carriers from the emitter to the base is due to diffusion, not acceleration from an applied field (in fact, the electric field within the BE junction 'pushes' the majority carriers away from the CB junction).

The acceleration of the injected charges towards the collector is in fact due to the field associated with the Collector-Base voltage, not the field Base-Emitter voltage. Alfred Centauri 16:07, 8 September 2007 (UTC)

(Kevin, I've moved your quoted comments below mine. Please do not insert your comments in-line with mine again.) Alfred Centauri 20:24, 8 September 2007 (UTC)

Yes, there is a diffusion current and an opposite electric current that balances it due to the in built potential (barrier). The applied Vbe modifies this potential barrier height, and hence the emitter/collector current.

Can you produce a physics derivation showing that the intrinsic cause of collector current is due to the actual magnitude of the base current?

Yes, Vce accelerates the charge from the base to the collector, as already pointed out on the main page. This is well known and illustrates the need for an electric field to move charges.

I tided up the reference to the proff to sound more kosher.

Kevin aylward 16:43, 8 September 2007 (UTC)

Voltage - Current Control Controversy

This section does not report a controversy; instead it promotes one PoV. 1Z 16:43, 8 September 2007 (UTC)

Why?

It is a matter of fact that the BJT is considered a voltage controlled device in standard semiconductor physics. It is truly astounding that there is such a common misconception that the BJT is current controlled, simply because base current exists.

For me the history of this goes back 30 years, to when I was an EE student. I still remember our physics professor going, look, ignor what you may have heard, the BJT is a voltage-controlled device, not current controlled, and this is why.... My latter reference here to Bart Van Zeghbroeck, University of Colorado, is just one of an infinite number of people that support this view. I just cant imagine any university academic saying you know, the BJT is base current controlled. Its just nonsensical. But I suppose...

It is also a matter of fact, not a POV, that the majority of professional analogue designers, i.e. that done by i.c manufactures (billions and billions of chips), use the i=gm.vi, voltage controlled model. It the one that works. —Preceding unsigned comment added by Kevin aylward (talk • contribs) 17:08, 8 September 2007 (UTC)

It is easy to imagine university academics propounding any position on this. But the ones that are serious about the physics teach that the collector current depends on the excess charge in the base diffusing over to the depletion region near the collector, where the field can grab them and accelerate them into the collector. The Early effect is easily explained by that edge of the depletion region moving, so that more charges can diffuse to it easily. As to what designers use, you may be perfectly correct, and we shoud include that fact if it is verifiable. Dicklyon 20:53, 8 September 2007 (UTC)

I dont have a problem with this, but see below on what charge controll actually means.

They also teach that one can use exactly the same underlying principle of calculation for the mosfet, to wit, calculate the charge in the channel due to gate voltage, and then use this charge to calculate drain current, to wit, KW/L.(Vg-Vt)^2.. No one seems to have a problem with then stating that the mosfet is a voltage controlled device.

From a big picture view, the mosfet and bipolar are, essentially, the same (draw them). The both have the same npn junctions, the p-well forming the gate length, is the “base”. The charge in the channel/base region is “induced” by gate/base voltage. Indeed, a mosfet will behave as a parasitic bipolar if one uses the well connection as the base. Kevin aylward 10:28, 9 September 2007 (UTC)

How to proceed: Kevin, to make progress, it would be useful if you would clearly point out the parts of the present text that you disagree with; you can mention them here, or add {{citation needed}} or {{dubious}} tags to passages that you think are not verifiable. You can add new parts, and they will stay if supported by verifiable reliable sources, but may be challenged or removed if not. But if you say in the text that something is incorrect, or that there's a controversy, that very definitely needs a reliable source. It's better to treat multiple viewpoints without making a controversy of it, unless there's a genuine controversy ongoing in the field that has already been reported and analyzed in secondary sources; it's not OK to make up a controversy, or to call one approach incorrect, from the fact that you find multiple viewpoints out there. Dicklyon 17:36, 8 September 2007 (UTC)

Furthermore, to say that the collector current is directly or physically a function of what's happening at the the base–emitter junction is absurd on the face of it.

I disagree. To say that it doesn’t is absurd. The applied voltages directly modulate the junction potentials Kevin aylward 10:28, 9 September 2007 (UTC)

The only influence is via the excess minority carriers in the base; that's why in all accurate models the collector current is charge-controlled and there's at least one state variable for stored based charge in addition to the terminal voltages and currents. Dicklyon 17:41, 8 September 2007 (UTC)

What causes the excess minority carriers? See below on what "charge-controlled" actually means Kevin aylward 10:28, 9 September 2007 (UTC)

I suspect your main point is really not in the physics but in the design field, which is your experise. The statement "In linear circuit design, the current-control view is often preferred, since it is approximately linear" may need to be modified, though "often" may be true, to indicate that the transconductance (voltage control) view is even more often what's used in traditional circuit design. Dicklyon 17:49, 8 September 2007 (UTC)

My point is both. The basic physics and how transister circuits are designed by professionals.

"In linear circuit design, the current-control view is often preferred, since it is approximately linear" may need to be modified. I would certainly agree with that to the extent that in small-signal analysis models, the BJT is a linear transconductor. On the other hand, designers, whether we realize it or not, are effectively linearizing the BJT by using it as a current-controlled device when they have significant series resistance in the base and/or emitter circuit. All one has to do to see this is to convert the circuits looking out of the base and emitter leads of the BJT to their Norton equivalents. Alfred Centauri 21:18, 8 September 2007 (UTC)

Here's another data point [10] and here's the payoff quote:

"This results in a collector current, controlled by a (smaller) base current. ... You can think of bipolar transistor as a current amplifier... or as a transconductance device"

Now one might argue that The Art of Electronics is not an 'academic' reference. On the other hand, it is highly regarded by professional designers while the authors, Paul_Horowitz and Winfield_Hill are indeed academics. It is my opinion that this reference combined with Dicklyon's recently added reference (in the main article) precludes any edit to this article that claims that the BJT is definitively voltage-controlled. Alfred Centauri 00:27, 9 September 2007 (UTC)

Unfortunately for this view is that, Winfield Hill and I have been in correspondence on several occasions in the past, he often posts to the electronics NGs. I am quite confident that his view is that the BJT is fundamentally a voltage controlled device. Even Steven Hawking writes things loosely that are not strictly true, just to convey a basic point. I will attempt to contact Winfield, and get a definitive statement from him to include in this discussion.

Secondly, Bill below explains quite clearly why the BJT can not possible be base current controlled. The only reason I can see why this explanation is not conclusive on eliminating any alleged base current control is a basic lack of understanding of physics. It pointless trying to prove a point to those that do not have the background to understand what is being proved.

The BJT *is* definitively voltage-controlled becuse that is how standard physics describes BJT, for those that have actually taken BJT semiconducter course, to wit:

The phrase "charge-control model" means that, YES, the collector current is taken to be a dependant variable from knowing charge in the base region. HOWEVER, this model calculates by fiat, those charges from the known terminal *voltages. The charges are taken to be dependant variables on the independent terminal voltages. The charges are calculated from voltages. This is the way it is. The reference I have to mind and have actually read, is Modeling the Bipolar Transistor by Ian E. Getreu but I have to will try and confirm this in case my memory on the details of this text is unclear. My guess is that 100% of those making charge-control model claims on the BJT page, have never actually looked at the formal derivation of that model. They use the words, but simply don't know what the words mean. Kevin aylward 10:28, 9 September 2007 (UTC)

You can think of bipolar transistor as a current amplifier Of course you can!

But think for a second. Why didn't Horowitz/Hill simply flat-out state that "The bipolar transistor *is* a current amplifier?" Why do we have to "think" of the transistor as being a current amp? Because it's not one, although it saves lots of trouble if we pretend that it is. Ic=hfe*Ib is a simplified model which works great as long as you don't look inside the transistor. But... is it real, or is it a Lie to children? Here's an analogy: we can also think of electrons as orbiting an atom's nucleus like little planets, and it's a great aid to simplified explanations ...but it's wrong. Ic=hfe*Ib works great for basic circuit design, but it forms a large stumbling block for anyone who tries to understand the internal operation of a transistor.

Here's another way to say it: please explain how the base current can have an effect upon the collector current. This is very important issue. Anyone who claims that Ib really does control Ic must tell us how this control actually works. I know of no reason why the motion of the carriers of Ib could have an effect upon the value of Ic, but I do know a straightforward explanation for the origin of Beta:

As with any simple silicon diode, the thickness of the BE depletion layer is proportional to the voltage applied across B and E. Change Vbe, and you change the depletion layer thickness. And because Ic and Ib are both determined by diffusion across this single depletion layer, Ib and Ic values are "locked together." Ib does not DIRECTLY control Ic. There is no DIRECT path by which the Ib current can be amplified to alter Ic. Instead Vbe determines both Ic and Ib, with Ib being smaller because Ib is caused by relatively rare majority/minority carrier recombination in a very small Base volume. If we connect a current source to the base terminal, Vbe will change, and the width of the BE depletion layer will change, and this alters the value for Ic. Ib did indirectly determine Ic, but it did so by altering Vbe and simultaneously changing the width of the BE depletion layer.

If there was some weird way way alter Ib yet hold Vbe constant (and prevent the depletion layer thickness from changing,) then we'd find that we could set Ib to any value we want, yet it's purported effect upon Ic would be gone. If Vbe can't change, then Ic can't change, and the equation Ic=hfe*Ib no longer works. This illustrates the idea that Vbe controls Ic, and Ib does not.

So, if we want to build a simple amplifier, it helps if we think of transistors as being current amps, even though they don't internally work that way. (And if we want to explain atoms to little kids, it helps if we think of electrons as orbiting the nucleus, even though atoms really aren't like that.) But if we want to become advanced students, and we want to develop a firm grasp on the overall picture of transistor internal operation, then we must first remove a large mental stumbling block from our minds: the false idea that Ib somehow can alter the value of Ic.

PS Imagine if school kids with no physics knowledge insisted that electrons really do orbit around atomic nuclei, and they started a huge fight on the pages of WP to fight with physics teachers who want to explain quantum numbers and the 3D shapes of atomic orbitals. It would be silly and embarrassing. That's how I see this fight between "current control" versus "voltage control." It's the kiddies versus the professionals, and only on WP can the kids win, since emotional opinions rule the day on WP, and knowledge and expertise is basically irrelevant here. The WP rules empower ignorance, guarantee flamewars, and cause thoughtful people to leave in disgust. Sad, and it didn't have to be that way. --Wjbeaty 03:41, 9 September 2007 (UTC)

Why are you knocking down this straw man? Has anyone said that the BJT is a current-controlled device? For the base–emitter junction, it makes just as much sense to treat its I–V curve with either variable being independent and other dependent, depending on how you drive it. In either case, the DC collector current is going to be a substantial fraction of the DC emitter current, but getting the details and AC right will require looking at the physics of excess minority carriers in the base region. Is there something in the article that you would change to explain this better? Dicklyon 03:53, 9 September 2007 (UTC)

By the way, the "kiddies" interpretation you've proposed is really quite offensive, bordering on incivility. Some of the editors may not have the same education you do, but they are not dummies and not just playing around, as far as I can tell. I trust you didn't mean to include me in that. In any case, please review WP:NPA and return to discussing how we can improve the article, instead of talking about other editors and their motives. Dicklyon 04:00, 9 September 2007 (UTC)

Yes, I can see how the kiddies reference can sound offensive, yet, for me it does catch the essence of the truth. By far the majority of professional analogue ic designers, and I mean the ones that design the billions and billions of analogue ics for a living, just don’t design BJTs from the point of view of base current controlled collector current. How do you suppose Barrie Gilbert could have come up with his Gilbert cell multiplier without the voltage controlled model? In my experience, it is truly only the novice designers that have learned bad habits that use this base current controlled idea. Its usually knocked out of them by a mentor engineer in the first few years, if their lucky. We use formulas like, max gain is Va/Vt, Av=gmRL. We correct for base current by subtraction of rbb’.ib from the vbe applied to get vbe’ at the junction etc. etc. I know this may sound pretentious, but this is just how it is. I don’t know how to make these type of facts sound less insulting. The truth often hurts. Kevin aylward 10:55, 9 September 2007 (UTC)

By the way, the "kiddies" interpretation you've proposed is really quite offensive. Dick, the only people that can offend you are those that you grant that power to. Wjbeaty diminishes only himself with such rants. That's too bad because up until now, I have found Wjbeaty to be quite civil and rational. What I don't understand is why Kevin and Wjbeaty, who otherwise seem quite intelligent, can't come to grips with these basic facts: Wikipedia is not about the truth and Wikipedia is not a forum to reveal the truth (from WP:V: "The threshold for inclusion in Wikipedia is verifiability, not truth"). However, it does not follow that grade school physics texts are of equal authority to graduate physics texts despite Wjbeaty's inference to the contrary.

There is no 'fight' here between voltage-control and current-control as Wjbeaty claims. That would be like arguing if the voltage across a resistor causes the current through the resistor or the other way around.

However, let's ask this question: What causes the measured voltage across a conducting PN junction? Is it due to a voltage that has been applied to the junction and that causes diode current? I suggest anyone thinking about answering that question think clearly and carefully before doing so. Hint: How does one actually apply a voltage to a diode etc. and does that process involve the movement of charge (otherwise known as a current)? Alfred Centauri 04:46, 9 September 2007 (UTC)

I have though about it. As noted above, charge generates the power supply voltage, but this is way far removed down the chain. At the operational level, we have a supply voltage that modulates the charge in the BJT, causing transistor action.

Indeed, it's all kind of silly. Why don't they read what the article says, and let us know if something about it needs improving, instead of all this arguing about what nobody has claimed? I think what it says is balanced, correct, verifiable, and not in conflict with all this stuff he's saying, so what's the beef? Dicklyon 04:52, 9 September 2007 (UTC)

It isn’t. It claims that the BJT is base current controlled. It isn’t. The most authoritative university reference on the nature of the BJT that I see, is the one I referenced, which states categorically that the transistor is voltage controlled. I will let you know when I get a response from Winfield. Kevin aylward 10:55, 9 September 2007 (UTC)

Kevins Edits

For those that desire to revert my inclusion of:

It should be noted here that the standard physics model expressed by the term “Charge Controlled Model” is open to misinterpretation as in that in this approach, the base charge is considered the dependent variable as a function of the independent terminal voltages. These currents are then calculated from the voltage dependant charges. This is in contrast, to calculating the currents without the intermediate charge calculation.

Please explain why you disagree with the above claim. The above is a matter of fact,and is clearly requied to avoid misunderstandings. Please explain what physically causes the charge in base. How is this charge casually related to whatever is claimed to be driving the transistor such that it can actually function as an amplifier to an external signal.

Kevin aylward 11:38, 9 September 2007 (UTC)

It is completely irrelevant whether I disagree with this claim or not, Kevin. Why is it that this is so hard for you to grasp? Can you answer that? All that matters is that what you've written above can be verified by an authoritative source. You said: "It should be noted here that the standard physics model expressed by the term So, “Charge Controlled Model” is open to misinterpretation". Please provide an authoritative citation for this statement.

But, there is another thing I want to discuss. You've continued to edit the BJT article while claiming a consensus for your views on this talk page when there isn't. This article is not a battle ground for the truth about BJT operation. And, as I said earlier, there are well known, authoritative references that clearly state that the BJT can be thought of as voltage controlled, current controlled, and charge controlled. Why? Because, these variables are not independent of each other. To claim that one variable is the one while the others are not is just plain silly.

Finally, the tone of your statements on this talk page (and the main article) is often inappropriate. The fact is, the other editors involved in this dicussion are well educated and have the credentials to back that up if that's what you're looking for. Do stop to consider, before hitting that 'Save page' button, that well educated people often disagree. Alfred Centauri 13:19, 9 September 2007 (UTC)

Kevin, I have reverted your edits to my comments as they are considered vandalism. Once again please do not insert your comments in-line with mine. Alfred Centauri 15:00, 9 September 2007 (UTC)

Well, this is a new one to me. Do you post much on the news groups? Its de-facto that the correct NG etiquette is to interleave the originator and replied text. Otherwise it’s a complete nightmare to follow the flow of the debate.

I'll just dump all the prose down here then, and let people try and figure out what goes with what.

Of course its relevant. If you can't explain what “charge controlled model” means, then you can't make any claims about it. Why is this so hard for you to grasp?

It is de-facto misunderstood by the actual comments on this page. Apparently, no one else here can offer an explanation as to what “Charge Controlled Model” actually means, despite quoting it at length, therefore it is trivially obvious that it is being misinterpreted. It does not need an authoritative citation on that basis. How can anyone use a principle if they don’t know what the principle even refers to. Its like stating that conservation of energy means that that I like blue colours.

Additionally, I have stated that I am trying to track down a reference. Or do you consider me a liar?

Oh? You specifically stated:

"Nobody has claimed that base current physically controls collector current."

It is that fact that there appears to be a consensus on, and which I have edited.

Apparently it is. Wikipedia is an encyclopaedia, so as far as reasonable, one does indeed to expect it to represent the truth as far as current knowledge holds.

enjoy. Kevin aylward 17:49, 9 September 2007 (UTC)

Do you post much on the news groups? I have in the past but remember that on usenet, each post is a separate entity and, with the proper reader, one can see the hiearchy including the original post in it's unaltered form. Here however, the talkpage is presented at once and, if your interleave your comments with mine, the question of who said what when becomes almost impossible to determine as it is hard enough to do this already. For example, in the comments I reverted, you wrote:

"Oh? You specifically stated: "Nobody has claimed that base current physically controls collector current.""

But, in reality, I didn't write anything like that. The closest thing I have found is this: "Has anyone said that the BJT is a current-controlled device?" but that was written by another editor. So, once again, please leave my comments intact and simply quote them in your comments if you wish to reply to individual parts. Alfred Centauri 18:08, 9 September 2007 (UTC)

Kevin has taken what I wrote completely away from its meaning in context. "Nobody has claimed that base current physically controls collector current" is in no way contradictory to the statements in the article that he is changing. I have also explained these things to him carefully in email. He seems to want to provoke an edit war. Better that he should learn a bit more about wikipedia editing, such as by reading WP:V and WP:RS and try to find a way to interact more productively. Dicklyon 18:26, 9 September 2007 (UTC)

Winfield Hills Response

I have now recieved an email response from Winfield -- Kevin aylward 17:43, 9 September 2007 (UTC)

The email is reproduced here, unedited:

Start of Winfield quote:

Here's my statement, in response to your request.

Although one can use a small current to bias and use a bipolar transistor, this is generally not a very successful way to use BJTs to make an amplifier, because the current gain, or beta, is very unpredictable from one part to another. Instead we use voltage biasing, and take advantage of a BJTs highly-predictable base voltage, as expressed in the famous Ebers-Moll equation. Usually we employ an emitter resistor, a current source, or another matched transistor, in the bias circuit. After setting up the transistor's operating condition, we finish our analysis and design by carefully considering the impact of the range of base currents we might encounter, e.g., by using the h_FE or beta spread as expressed or estimated from the BJT's datasheet specifications. Yes, beta plays a role, but a secondary role.

Sometimes the use of a high-value emitter resistor or a current source, etc., mentioned above, may dominate the circuit analysis, and we can simplify our description of the transistor to a fixed Vbe drop, like 0.65 volts, etc., and ignore Ebers-Moll, but this does not take away from its importance and significance. After all, in this situation, its Ebers-Moll that lets us determine if the Vbe change is small enough to ignore.

To understand the gain of a BJT amplifier circuit we calculate the BJTs transconductance, which does not vary from part to part, and which is accurately predicted from the Ebers-Moll equation. Once we know the transistor's operating bias current, from the steps above, we know its transconductance: g_m = 40 Ic (at room temp). To be clear, transconductance expresses a (small) change in collector current resulting from a (small) change in a BJT's base-emitter voltage. So it's a voltage-driven amplifier we're analyzing (not a current-driven amplifier). it is Ebers-Moll that lets us to not only determine the amplifier gain, but also its distortion. We can add the small influence of the Early effect to improve the accuracy of the analysis.

Sometimes we place our BJT in a multi-stage feedback circuit, which may dominate in the amplifier gain analysis, and we can simplify the transistor description and ignore transconductance, but this does not take away from its importance and significance. After all, it's what we use to calculate the open-loop gain.

In summary, the amplifier design process is most accurately described by considering the BJT to be a voltage-controlled device, rather than a current-controlled device.

End of Winfield Quote:

I agree with virtually everything Winfield says here particularly this: "the amplifier design process is most accurately described by considering the BJT to be a voltage-controlled device" (emphasis mine). But please note that is a far cry from the claim that "the BJT is a voltage-controlled device". The amplifier [circuit] design process is fundamentally based on models that may or may not have a physical basis. Thus, this info from Winfield doesn't rise to the level of confirming one physical description of transistor operation or the other. Alfred Centauri 18:22, 9 September 2007 (UTC)

I too support and agree with Winfield's statement, and I'm sure there are sources we could cite if someone wants to put something about that into the article. But it says nothing about the physics of the device, and does not address the issues around Kevin's edits, as far as I can tell. He certainly isn't calling the charge-control model a misinterpretation, or denying that the device can be modeled in terms of beta. Dicklyon 18:30, 9 September 2007 (UTC)

I am not saying that the charge controlled model is a misinterpretation. I am saying that people are misinterpreting what the charge control model actually means. I have no bone to pick on that model itself. It is correct. People just don't seem to understand what it is saying. As for the not denying that one can’t use beta in general, this is left to Winfield’s followup email.

Kevin aylward 19:31, 9 September 2007 (UTC)

OK, but let's get away from worrying about "people misinterpreting" and get back to improving the article. What in the article is incorrect? Please provide sources to back up your assertions. Dicklyon 22:29, 9 September 2007 (UTC)

Winfield Second Response

After writing my response, I wandered over to Wikipedia to see what all the fuss was about. I found the little segment wherein H&H AoE is quited to support a current-control model. I think one badly misreads our book in so doing, because we quickly introduce Ebers-Moll to the reader and use it thereafter in our analysis.

You may copy my paragraphs into the Wikipedia discussion if you wish. Ditto for this letter.

As for the Wikipedia article, my preference would be to see both approaches mentioned, but with the serious disadvantage of relying on a given beta stressed, and the advantages of using Ebers-Moll (and its daughter equations) for accurate calculations also stressed.

The so-called "exponential function" aspect is BJT reality, but it's absolutely not a problem, being easily dealt with by using the small-signal transconductance g_m at a given operating current.

In fact, what Ebers-Moll lets us do is observe exactly how g_m changes with collector current, as the output signal traverses from the minimum to maximum voltages, and thereby determine the amplifier's distortion. To my knowledge this cannot be analyzed using the current-control approach.

Kevin aylward 19:25, 9 September 2007 (UTC)

Actually, if you assume constant beta and a diode at the B-E junction, the current-control model trivially leads to the same transconductance versus collector current relationship. But I agree with Winfield that we should represent both methods and that their relative usefulness can be pointed out with reference to a reliable source such as his book. I see no reference to H&H in the article, so I'm not sure what he's referring to about using it as a source for the current-control idea; perhaps an old talk comment. Anyway, let's focus on what the article needs, old talk aside. Dicklyon 22:28, 9 September 2007 (UTC)

I think one badly misreads our book in so doing. Mr. Hill is mistaken. The quote from AoE was not introduced to support a current-control model over any other model. Even a casual read of my comments on this issue reveals that my position is that the voltage-control / current-control model dichotomy is a false one. That is, it is my position is that the views are duals. The fact that AoE mentions both (actually, charge control is also mentioned on the same page, so in fact three views are mentioned) is exactly the reason I put the link and quote up. It has been my understanding that Kevin is convinced that there is only one correct view - the voltage-controlled view and that is why I have been pushing back.

But now, what I'm concerned about here is a subtle confusion regarding context. Mr. Hill appears to be speaking from a design perspective which is a different context than the physics perspective. So, just to make sure that I'm on the same page as everyone else, here is my question: In what context is this discussion of how a BJT is controlled - a design context or a physics context (or both)? Alfred Centauri 23:05, 9 September 2007 (UTC)

One can say that there are two basic approaches to modelling, one based the “real” physics of the situation, the other based on a functional curve fit, which ignores the reason for the relation

The base current controlled view is an approximation, functional curve fit. It is a black box, non physics based model, that whilst producing a basic overview of one main characteristic of BJT operation, is not a sufficiently accurate approximation, or general enough to perform useful design of the majority of production viable BJT circuits. An accurate model that does allow such production of satisfactory BJT circuit designs, is the voltage controlled, physics based model.

The main page needs to address that fact that the base control view is a poor approximation.

It needs to address the fact that a valid authority reference, Bart Van Zeghbroeck Professor University of Colorado Department of Electrical and Computer Engineering, specifically denied that the transistor is a base current controlled device. There is no basis to keep removing this reference, as it is a rebuttal, valid under wikipeda guidelines, to the allegation that the BJT may be considered as a base current controlled device. Yes, all alleged views should be represented, including authoritative references that claim only one view is the correct one.

Kevin aylward 11:04, 10 September 2007 (UTC)

It is, and should be, both. But you are right that Kevin seems to be mixing them up, using statements about design to push ideas about the physics; or so it seems to me. Dicklyon 23:07, 9 September 2007 (UTC)

I am indeed quite clear on both the physics aspect and the design aspect. Winfield, is only addressing the design aspect with his comments here. I am addressing both. From a physics point of view, the transistor is a voltage controlled device. That is, the terminal voltages directly determines the charge in the base, and such charge, directly determines the collector current. From a design point of view, the base current control approach is severely limited in its ability to predict transistor behaviour, for example, multipliers, current mirrors, and distortions. For example see, http://www.kevinaylward.co.uk/ee/bipolardesign2/bipolardesign2.html

And whilst I really do not want to step on any toes, here, I have great respect for Winfield’s views. It is most likely that that my experience of the design of 50,000 transistor analogue integrated circuits, including production designs at major semiconductor companies, e.g. Texas Instruments, gives me an alternative insight. Of course, I could be a twat, but there you go.

Kevin aylward 11:04, 10 September 2007 (UTC)

The Art of Electronics

I've beefed up the section in question with more specific info from, and citations of, Horowitz and Hill. Please comment or further improve it if it's not quite right. I've tried to take Kevin's concerns into account, especially in the final paragraph about what models are used by designers and why. Dicklyon 02:43, 10 September 2007 (UTC)

Bart Van Zeghbroeck on Voltage Control

Result of newsgroup sci.electronics.design discussion 04 December 2004 19:57 - Re: FETs Vesus Bipolars, Why More Efficient?

Ratch, I hate to be the one telling you this but the BJT is indeed a voltage controlled device. The voltage applied to the base emitter junction controls the collector current and the base current is a result of the additional hole injection (for an npn BJT) into the emitter as well as the recombination in the base-emitter depletion region and the quasi-neutral base region. It is tempting to claim that the BJT is controlled by the base current, since that is how a BJT is typically biased; the exponential variation of the current with the base-emitter voltage makes a voltage bias impractical. Any circuit designer will also tell you that any voltage bias can be replaced by its Thevenin equivalent current source. Hopefully this provides you some ammunition to claim that either one can be claimed when treating the device as a black box. Finally, you'll find that a MOSFET biased in the subthreshold region has characteristics that are very similar to that of a BJT.

Bart Van Zeghbroeck Professor University of Colorado Department of Electrical and Computer Engineering Campus Box 425 Boulder, CO 80309-0425 Office ECEE1B41

Kevin aylward 11:55, 10 September 2007 (UTC)

βF

The article, in the Bipolar junction transisitor#Voltage, current, and charge control section, contains the sentence "In analogue circuit design, the current-control view is sometimes used since it is approximately linear. That is, the collector current is approximately βF times the base current" (emphasis added). As far as I can tell, F is not defined in the article, and I don't know what it could be. (Or maybe this was supposed to be β_F, for forward β? --Gerry Ashton 16:05, 10 September 2007 (UTC)

I fixed it back to subscript F for forward. Dicklyon 17:36, 10 September 2007 (UTC)

Ok, however, you reedited back in "with critical gain and speed parameters". This is inaccurate. Critical is not a requirement. As I have noted, and indeed as Winfield noted, it is standard practise for professionals to design with the voltage control model. It is not reserved just for "criticle" parameters. Kevin aylward 08:27, 11 September 2007 (UTC)