How Transistor works - verifying

[wbeaty]

Honestly Bill, the rebuttals to your site have been thoroughly presented.

Yes, I had my chance and now it's too late!

It's a very regretful situation, but that's just how things must be.

And besides, who do I think I am?! Someone coming in at this late date, wanting to give input long after the timer has expired.

I should be ashamed.


(Also, prev msg link is bad.)

 

[cabraham]

https://www.physicsforums.com/showthread.php?t=390291&highlight=wbeaty

That should do it. It's never too late to ask a question. I'll clarify anything that needs to be understood better.

Claude

 

[wbeaty]

That should do it. It's never too late to ask a question. I'll clarify anything that needs to be understood better.

You said that you and I had a discussion, first linking to that other forum with "Rachit," then linking here where I'd posted one single message (and in that instance I never stayed around to read any responses.)

That's very odd. No, I've not been involved in any "discussion."

But now I'm wondering ...back during that enormous "current controlled" debate, did you suspect that the anonymous "Rachit" guy was actually myself? That you and I did have a discussion, but I was hiding my identity?

 

[cabraham]

You said that you and I had a discussion, first linking to that other forum with "Rachit," then linking here where I'd posted one single message (and in that instance I never stayed around to read any responses.)

That's very odd. No, I've not been involved in any "discussion."

But now I'm wondering ...back during that enormous "current controlled" debate, did you suspect that the anonymous "Rachit" guy was actually myself? That you and I did have a discussion, but I was hiding my identity?

I only asked the question. I don't dislike Rachit, nor anybody who opposes me. I just want to re-examine everything until consensus can be reached. I have a pet peeve just like we all do. Mine happens to be debates that never end. At some point in time there needs to be a "Supreme Court" of some kind to resolve it. The SC is not always correct, & they have reversed themselves numerous times in US history.

The current controlled model of a bjt has always been presented as an external viewpoint, focused solely on driving the bjt w/ external circuitry, not examining the internal device physics. When this point was finally uinderstood after a lengthy debate consensus was reached. It has been universally accepted, even from critics of semicon OEMs, that the current control viewpoint does not address internal physics. Rather the CC approach is beneficial for driving the device, developing networks around the bjt. In short, a constant voltage source should never be used to drive the b-e jcn.

A constant current source biases said b-e jcn very well, but a CVS would likely destroy it. Nobody, including critics, disputes this assertion. So we had universal concensus, no dissent at all, that for an external view, CC is what works best.

We also concurred w/o dissent, that when operating a bjt at a speed high enough so that internal charge distribution & transit time, is no longer negligible, then the CC model is inadequate. I still repeat this - any time internal device physics is under examination, the CC model is inadequate.

So the debate then shifted to which internal model is best. The peer reviewed app notes from OEMs, & unis say QC, charge controlled. A handful of critics say VC, voltage controlled. Scan the debate & it was eventually settled w/ full agreement that QC is the best internal model.

Do not hesitate to ask for clarification. I want to be seen here as an approachable dude. I hope that everyone feels they can talk to me freely w/o contention. If I've erred somewhere I'll accept & welcome correction. If my explanations are not effective, I'll rephrase. Please feel that you can talk to me freely w/o any repercussions. So here it is.

External model neglecting internal device physics - current control.

Internal model considering internal device physics - charge control.

Atomic level considering the limits of physics knowledge - quantum mechanics.

I'll address any questions. Best regards.

Claude

 

[wbeaty]

I'll address any questions. Best regards.

Hmmm, a problem.

Again: did you suspect that "Rachit" and myself were actually the same person?

 

[cabraham]

Hmmm, a problem.

Again: did you suspect that "Rachit" and myself were actually the same person?

I did not assume that. I just wasn't sure. Some of his phrasings sounded like yours but that could be because he relied on your site as a reference. Just curious, Rachit is not a bad person. I just wanted to know his reasoning & sources for his contrarian position.

Don't take any of this as personal. I ask everybody else, you included, to provide sources & proof when taking a contrarian stance against a well established position in science. I provide sources & proof to support my conformist positions.

One more point, & no offense is meant. Contrarian positions are, for some folks, a lot more fun to ponder than the official viewpoint. There is something about being in a vocal minority that gives them a sense of being more enlightened than the rest. They consider the official viewpoint as "misconception, myth, based on limited anecdotal evidence, etc." What the critics need to do is carefully examine their own positions, & they will find that they make assumptions at the start that are not valid, or anecdotal, holding under limited conditions. To these folks, there is little excitement in conformist positions because one cannot take credit for it. But once a fact is established based on solid proof, there is no need to knock it down to prove oneself capable. There are many new ideas & inventions waiting to be discovered. I suggest that their effort be redirected.

I always refute the critics with solid immutable laws like conservation of energy & charge, Kirchoff's 2 laws, Maxwell's equations, etc. These are Gibraltars of science. Any theory which cannot conform to these laws are nothing but heresy. To prove the contrarian view requires disproving these well proven laws.

Any theory which is contradicted by said laws cannot be taken seriously. The critics are the ones spreading myths & misconceptions. Everything we need to know is taught at the accredited unis in engineering (full) & physics curricula. Web sites preaching info which counters the uni teachings must be scrutinized, & I always find that assumptions are made early which are invalid, & hence produce invalid conclusions.

But admission of an error is rare. Once a critic has boldly declared the engr/phy community to be wrong, mountains of evidence refuting their erroneous view is futile. They are then in a position where they must defend their contrary view,& attack the official view just to save face. That is why I tend to be careful about debating an issue. If I am not experienced in a topic, I tend to trust those who are. Nothing is worse than taking a hard stand on an issue I am not an expert on, then after others present evidence to the contrary, I realize "Oh s***, I'm wrong!" I avoid that by debating only when I know the facts.

Oh well, I do run on. Best regards.

Claude

 

[Studiot]

Everything we need to know is taught at the accredited unis in engineering (full) & physics curricula.

Probably the two most famous counterexamples to this declaration are

The gross miscalculation of the age of the Earth by Lord Kelvin.

The address to the Royal Society by Oliver Heaviside "Gentlemen, shall I refuse my dinner because I do not fully understand the process of digestion?"

 

[cabraham]

Probably the two most famous counterexamples to this declaration are

The gross miscalculation of the age of the Earth by Lord Kelvin.

The address to the Royal Society by Oliver Heaviside "Gentlemen, shall I refuse my dinner because I do not fully understand the process of digestion?"

I stand by what I posted. One person, such as Lord Kelvin, does not disprove my point. Lord Kelvin put his foot in his mouth on more than one occasion, esp in his later years. The unis teach what is known at this moment to the best of their ability. New discoveries will be made & the science/uni canon will be updated to reflect new info.

As far as OH & dinner goes, that is hardly a rebuttal to my point. I didn't say that all is known. Rather, if we wish to understand how to utilize a bjt to its full effectiveness, unis are a good source, as are semicon OEMs.

The internal physics involves QM, of which the science community has only a limited understanding. We have much to learn about QM, & the unis provide only a limited amount of enlightenment here, because man's knowledge of QM is limited.

I hope that clears up my statement. BR.

Claude

 

[wbeaty]

I did not assume that. I just wasn't sure.

Nope, I've never been on that forum. I think I registered long ago when I was adding forums to the big list on my Electronics Hobbyist page. I suspect that Rachit is this one professional chip designer who uses SED and other newsgroups, and appears to be a big fan of my articles. He probably emailed me a few times years ago.

 

[wbeaty]

Don't take any of this as personal. I ask everybody else, you included, to provide sources & proof when taking a contrarian stance against a well established position in science.

What contrarian position? What do you mean exactly?

My transistor article comes right out of standard textbooks. It's what I was taught in EE classes at the U. of Rochester. It's all based on the main bjt equation, the DC transconductance equation or "Shockley's Equation" which describes the effect of the potential barrier in the BE junction of the bjt: Ic=Is*(e^(Vbe/vt)-1). That equation appears in vast numbers of EE textbooks, if not all of them. I think Horowitz and Hill devote an entire chapter to it. In Art of Electronics, first they go over Ic=hfe*Ib and the "little man in the transistor." Then they drop the simple stuff and delve into professional-level design techniques. IIR, Sedra and Smith do the same. (Might you recall the undergrad semiconductor physics text you yourself used back in school?)

But my transistor article was written for the Electricity exhibit at the Boston Museum of Science (well, the initial version was.) Therefore my goal is to explain diode switching to little kids and grandmothers. I cannot introduce any math at all. I can't even use the central concept of "diode potential barrier." In order to translate the diode equation Id=Is*(e^(Vd/Vt)-1) into everyday language, I focus on visualizing the changing width of the depletion zone; the variable insulating region found in all PN junctions.

The general public can't deal with Id=Is*(e^(Vd/Vt)-1), but many of them can understand a thin flat insulator which dynamically grows and shrinks in thickness, and which presents a barrier between two conductors. My original transistor article was intended to have several animations where the Depletion Zone is depicted as glass, and the rest of the carrier-filled semiconductor material is depicted as metal. Also, I planned articles covering Diode first, then BJT later, so everyone can pick up a critical concept: the BJT is really just a fancy kind of diode switch.

You're totally familiar with Ic=Is*(e^(Vbe/Vt)-1), right? It's the "Ohm's Law" of the transistor.

 

[Chunkysalsa]

Interesting article to read. I don't know anything about transistor yet to contribute to your discussion.

I thought it was interesting to see you responding here when the other day my intro circuit analysis professor gave us the link to that article as a source for writing a small mini-paper on how transistors work.

 

[Studiot]

Out of interest, on what website was this 'discussion' with this 'rachit' fellow please?

I ask because I remember a member called ratch on who had a particularly frustrating discussion style on another forum. He kept referring to the amasci website.

 

[wbeaty]

I thought it was interesting to see you responding here when the other day my intro circuit analysis professor gave us the link to that article as a source for writing a small mini-paper on how transistors work.

Yeah, that article has attracted fairly huge web traffic, plus staggering amounts of ego-swelling congratulatory email, way too much for me to have any hope of answering. The physics educators love it. As S. Hawking discovered, any new explanations of physics, if they're purely verbal+pictures, tend to attract big crowds.

So I was a temporary minor online celeb. But you cannna makes a living on th intertubes, unlessuns you publishes a paper-based book.

 

[wbeaty]

Out of interest, on what website was this 'discussion' with this 'rachit' fellow please?

It's the same one Claude repeatedly posted here, the one with forty-three pages. I thought you participated? See the first page of this current thread, down near the bottom.

I ask because I remember a member called ratch on who had a particularly frustrating discussion style on another forum. He kept referring to the amasci website.

Look back there again, it doesn't say 'ratch,' it says 'ratchit.'

 

[cabraham]

What contrarian position? What do you mean exactly?

My transistor article comes right out of standard textbooks. It's what I was taught in EE classes at the U. of Rochester. It's all based on the main bjt equation, the DC transconductance equation or "Shockley's Equation" which describes the effect of the potential barrier in the BE junction of the bjt: Ic=Is*(e^(Vbe/vt)-1). That equation appears in vast numbers of EE textbooks, if not all of them. I think Horowitz and Hill devote an entire chapter to it. In Art of Electronics, first they go over Ic=hfe*Ib and the "little man in the transistor." Then they drop the simple stuff and delve into professional-level design techniques. IIR, Sedra and Smith do the same. (Might you recall the undergrad semiconductor physics text you yourself used back in school?)

But my transistor article was written for the Electricity exhibit at the Boston Museum of Science (well, the initial version was.) Therefore my goal is to explain diode switching to little kids and grandmothers. I cannot introduce any math at all. I can't even use the central concept of "diode potential barrier." In order to translate the diode equation Id=Is*(e^(Vd/Vt)-1) into everyday language, I focus on visualizing the changing width of the depletion zone; the variable insulating region found in all PN junctions.

The general public can't deal with Id=Is*(e^(Vd/Vt)-1), but many of them can understand a thin flat insulator which dynamically grows and shrinks in thickness, and which presents a barrier between two conductors. My original transistor article was intended to have several animations where the Depletion Zone is depicted as glass, and the rest of the carrier-filled semiconductor material is depicted as metal. Also, I planned articles covering Diode first, then BJT later, so everyone can pick up a critical concept: the BJT is really just a fancy kind of diode switch.

You're totally familiar with Ic=Is*(e^(Vbe/Vt)-1), right? It's the "Ohm's Law" of the transistor.

As I've stated repeatedly, there are 3 basic equations which can be labeled as "terminal relations":

1) Ic = beta*Ib

2) Ic = alpha*Ies*exp((Vbe/Vt) - 1)

3) Ic = alpha*Ie

Equation 3 is the law of transistor action. Equation 2 is often miswritten, as you just did. I've highlighted the factor "alpha" & the saturation current is "Ies". In a diode there is but one value of Is. In a bjt, there are 2 junctions each w/ their own value of "Is", due to differing doping densities in collector vs. emitter. Hence "Ics" & "Ies" are used in the Ebers-Moll equations.

All 3 equations are needed to fully describe a bjt. But only 1 of the 3 input quantities can be the directly controlled input, the other 2 being incidental, but inportant nonetheless. We can control the base current & then eq 1) gives us Ic. But this method results in beta dependency, which we usually avoid. Eq 3) is very reliable. If we set Ie to a known value, Ic is very predictable. Alpha is around 0.98 to 0.998.

We never control Ic w/ Vbe. The Ies term is very temp dependent. It also has a positive temp coefficient. A voltage source directly across the b-e jcn could kill the device. It's never done. We control a bjt by controlling its current.

In the process of doing so, a voltage develops across the b-e jcn. We use eqn 2) to compute transconductance of the device. This is the upper limit for the transconductance of the stage. The stage gm cannot exceed the devices gm.

Likewise, beta is the upper limit for the stage current gain. The stage current gain cannot exceed beta, that of the raw device.

Your presentation of bjt operation reads like a scathing indictment of OEM & uni app notes. It reads as if your version is the only valid one. My beef is that the explanation given in uni texts is spot on, & does not need any correction. You present E-M to the exclusion of the other 2. E-M is one law for bjt operation. The other 2 are just as important. A bjt is a 2 port device. It cannot be explained in 1 equation. Likewise for a FET.

As I said, to develop bjt networks, the CC model is good until storage/transit time becomes an issue at high speeds. Then QC takes over. I've used CC & QC forever, & haven't gone wrong. What have you used?

Claude

 

[wbeaty]

As I've stated repeatedly, there are 3 basic equations which can be labeled as "terminal relations":

1) Ic = beta*Ib

2) Ic = alpha*Ies*exp((Vbe/Vt) - 1)

3) Ic = alpha*Ie

Yep, dead normal textbook stuff.

But I begin to suspect that you didn't read my previous message (number 45.) Hint hint.

Equation 3 is the law of transistor action. Equation 2 is often miswritten, as you just did.

Miswritten? No, it's simplified, just as uni texts commonly do, for devices where alpha ~= 1. Must we now go and make a list of the undergrad textbooks which do this? Does your school have an engineering library with all the current double-E texts behind the ref desk? Well, doing all that is irrelevant, for the reasons explained in my previous message.

And again as before: might you yet recall which text you yourself had in your undergrad semiconductor physics course, the one that first covered transistors?

I've highlighted the factor "alpha" & the saturation current is "Ies". In a diode there is but one value of Is. In a bjt, there are 2 junctions each w/ their own value of "Is", due to differing doping densities in collector vs. emitter. Hence "Ics" & "Ies" are used in the Ebers-Moll equations.

Sure, but I think you're missing something important. Hint again: read my previous message 45, the one you quoted.

What is the goal of my transistor article?

That transistor article ...what was it's goal? What org was it written for, and who are their clients?

 

[cabraham]

You say in article 45 that lay people cannot handle E-M eqn. So you went into physics & potential barrier. I just don't see how that explains bjt action. That explains diode action. The relation between I & V is a diode relation. E-M is derived from Shockley's diode eqn, then combined w/ eqn 3), the alpha eqn.

The barrier potential explains the I-V properties of diodes & all p-n junctions including bjt. But to explain bjt action, we need eqn 3). If the base region of a bjt was super wide, say 1.0 mm thick, & every electron emitted from emitter recombines in base w/ holes, then Ic is near zero. Yet the I-V relation per Shockley is still valid.

A bjt w/ a thick base is merely 2 back to back diodes w/ no transistor action. Ebers-Moll equations include alpha to account for bjt action. With alpha near or at zero, Ic is near zero. Vbe can be 0.85 volts, but Ic is about 0. Without high alpha, Vbe matters not.

Again, the potential barrier description accounts for I-V log/exp properties. Of course Vbe must be non-zero in order to sustain emitter current which gives rise to collector current. But Ic = alpha*Ie is the eqn that separates a true bjt from back to back diodes.

FWIW, it's very hard to explain bjt action to lay people. But I've had the greatest success w/ the base region being so thin, carriers are yanked into the collector before they get a chance to recombine. Potential barrier details involve thermally generated electron hole pairs, phonon interaction due to lattice vibrationm thermal energy, band gaps, Fermi levels, recombination, etc.

The thin base region resulting in carriers yanked into collector before recombination can occur is simple, less involved, & requires no advanced math. Your website article chastises the whole science community for not explaining bjt action well enough. As I said, I've had hundreds of colleagues enjoy great success as an EE in hardware development, whos understanding of bjt is based on OEM app notes & uni teachings.

Also, I've aleady stated that the 3 terminal eqns are simple external models not accounting for device physics. But you then say that to go deeper we use E-M. I've already told you that charge control is the model used when transit time & charge distribution are relevant. Increase a bjt speed to hundreds of MHz. None of the 3 terminal eqns give a good answer. Or to take a bjt out of saturation requires knowledge of the stored exces minority carrier charge value. Neith eqns 1, 2, or 3, provide this. The QC model is needed.

Then to go even deeper, we need QM. Transconductance is not more basic than current gain. THey are both terminal quantities. You seem to think eqn 2) in more basic than 1) or 3). That is not so. I can elaborate if needed.

Anyway, let's not make this an endless campaign. You & others can have the last word. I'll answer a question if asked. BR.

Claude

 

[wbeaty]

I'll answer a question if asked

Just two questions.

First, might you yet recall which text you yourself had in your undergrad semiconductor physics course, the one that first covered transistors?

Second question:

You say in article 45 that lay people cannot handle E-M eqn. So you went into physics & potential barrier. I just don't see how that explains bjt action.

I haven't clearly stated my reasoning yet, so do you want to hear it? That's my second question.

 

[Studiot]

Originally Posted by Studiot
Out of interest, on what website was this 'discussion' with this 'rachit' fellow please?

It's the same one Claude repeatedly posted here, the one with forty-three pages. I thought you participated? See the first page of this current thread, down near the bottom.


Originally Posted by Studiot
I ask because I remember a member called ratch on who had a particularly frustrating discussion style on another forum. He kept referring to the amasci website.

Look back there again, it doesn't say 'ratch,' it says 'ratchit.'

Many thanks for this answer. I have now had time to review this website with the following results.

Comparing this website with the one I linked to in post#21 of this thread I think that 'Ratch' and 'Ratchit' are one and the same person. I note he signs himself Ratch in cabraham's E-Tech thread, although his handle is 'ratchit'.

In June 2008 he started the thread I linked to (post#21) by referring to your site (amasci) as proof that 'transistors are voltage controlled not current controlled'.
There was significant discussion, including the nature of the term 'control', though not of the gargantuan proportions of the one in 2010 in cabraham's link.

I apologise to claude if we had primed the pump for that argument in 2008.
I felt that his summary in post#39 here was particularly good.

However I would take issue with equations posted in post#50.

My version of equation (3) has another term which is significant in certain types of transistor and reminds us that there are other agents that affect, and therefore can 'control', the collector current.


3) Ic = alpha*Ie + Ico


go well

 

[cabraham]

Just two questions.

First, might you yet recall which text you yourself had in your undergrad semiconductor physics course, the one that first covered transistors?

Second question:

I haven't clearly stated my reasoning yet, so do you want to hear it? That's my second question.

Kittel - solid state physics as undergrad.
Muller & Kamens - Device Fabrication For Integrated Electronics at grad school, MSEE.
Sze - Physics of Semiconductor Devices, grad school Ph.D.-EE.

If you have reasoning you'd like to explain, sure, by all means do so.

Claude

 

[cabraham]

Many thanks for this answer. I have now had time to review this website with the following results.

Comparing this website with the one I linked to in post#21 of this thread I think that 'Ratch' and 'Ratchit' are one and the same person. I note he signs himself Ratch in cabraham's E-Tech thread, although his handle is 'ratchit'.

In June 2008 he started the thread I linked to (post#21) by referring to your site (amasci) as proof that 'transistors are voltage controlled not current controlled'.
There was significant discussion, including the nature of the term 'control', though not of the gargantuan proportions of the one in 2010 in cabraham's link.

I apologise to claude if we had primed the pump for that argument in 2008.
I felt that his summary in post#39 here was particularly good.

However I would take issue with equations posted in post#50.

My version of equation (3) has another term which is significant in certain types of transistor and reminds us that there are other agents that affect, and therefore can 'control', the collector current.


3) Ic = alpha*Ie + Ico


go well

Thanks for your feedback. I am well aware of the additional "Ico" term in eqn 3), & what you've presented is correct. But when we describe quantities that "control" a device, we are not usually referring to leakage & other parasitic flaws.

In a bjt, the objective is to control collector current w/ some sort of input signal, as found in uctlrs, transducers, photodiodes, etc. The inherent leakage current Ico, which exists due to the non-ideal nature of the reverse biased c-b jcn, is present & varies greatly w/ temp.

Of course Ico plays a role in determining Ic, but it is not something we use to control Ic. It is an inherent property of the bjt, one which fortunately has a very small influence on bjt Ic behavior, for silicon material.

In the 1950's when germanium was the dominant bjt material, Ico was a real problem at medium to high temps. Designers had to account for the large Ico c-b leakage when employing Ge devices at temps above 50 or 75 C. The limit was around 100 C.

Then silicon replaced Ge around 1959, & Ico for Si is generally small enough to neglect. Again, it's there, but Si devices can operate to the mil temp range of 125 C & beyond w/o Ico being too large an eror. It is an error term for sure, & your eqn is more precise than the simplified version I presented.

But Ico, & I believe I'll get universal backing/concensus on this, is NOT a "control" quantity. It influences Ic for sure, but we don't control Ic by setting a value for Ico. I think this whole question revolves around the meaning of "control".

Ib, Vbe, Ie, Vbc, Ies, Vt, Ico, etc., all have influence over Ic depending how the device is driven. If the b-e jcn is driven by a true current source or voltage source w/ a large series resistor, then Vt, & Ies determine the Vbe value at a given temp. Also, Ico adds to whatever Ic value is obtained from eqn 1) or 3).

Which quantity are we adjusting to get a specific value of Ic? That is what we mean by control. Again, your version of eqn 3) is more precise than my simplified version. Ico exists indeed & influences Ic. But if Ico is 2.7 uA, & we bias the bjt at Ic value of 1.0 mA, the error is just 0.27%. For larger Ic value, the error is less.

Again, w/ Ge devices, the presence of large Ico values forced the designer to take it into consideration. Circuit topology was built around the need to mitigate large Ico values. EEs from the 1950's can give you insight into this practice. BR.

Claude

 

[Studiot]

I think this whole question revolves around the meaning of "control".

I have no doubt you have great knowledge of semiconductors and have already commented how worthwhile your technical analyses are.

But I also feel there is a lack of acceptance that others may also have something valid to say and a corresponding willingness to listen to them as well as to expound to them.

Here is a simple experiment to prove that it is possible to 'control' Ic with the base not even connected.

You can show this by allowing photons to enter the transistor, biased suitably between collector and emitter only. It is possible to switch on the transistor by this means.

As a matter of interest how do multiple emitter transistors fit into your control scheme? and how do you describe control by emitter injection?

 

[cabraham]

I have no doubt you have great knowledge of semiconductors and have already commented how worthwhile your technical analyses are.

But I also feel there is a lack of acceptance that others may also have something valid to say and a corresponding willingness to listen to them as well as to expound to them.

Here is a simple experiment to prove that it is possible to 'control' Ic with the base not even connected.

You can show this by allowing photons to enter the transistor, biased suitably between collector and emitter only. It is possible to switch on the transistor by this means.

As a matter of interest how do multiple emitter transistors fit into your control scheme? and how do you describe control by emitter injection?

Thanks again Studiot for your feedback. With photon stimulation, instead of forward biasing the base-emitter junction, photons provide the energy to transition valence electrons into conduction. Though the base lead is not brought out to the outside world, the base-emitter junction carries a conduction current due to photon stimulation.

There is no base lead terminal, but there is a base region internally. Emitter region emits electrons which are yanked into the collector via the E field of the reverse biased c-b jcn. Have I answered your question.

Your last question is a good one. "Control by emitter injection" is how I describe the bjt in general. Emitter electrons injected towards the base quickly become collector current. Base holes injected into the emitter recombine there w/ electrons in the emitter. But base injection does not directly produce collector current.

The reason the base is doped w/ acceptor ions (for a p type base, i.e. npn bjt device), is to improve c-b reverse breakdown voltage & leakage current. A heavy doping of acceptor atoms into the p base results in more base current for a given collector curret, an undesirable thing. But doing so reduces Ico, a bad thing, & improves c-b junction blocking voltage ability.

What function does the injection component of base current serve is as follows. THe lower the doping density in the base, the lower the base injection current & the higher the beta value. Superbeta bjt's at op amp inputs use these devices. But the c-b blocking voltage ability is a few volts, & the leakage current from c-b is horrendous.

For bjt devices that are to operate w/ a Vce of 50, 100, or more volts, w/ low leakage current c-b, i.e. Ico, the base must be doped heavier than is optimum for high beta. Thus current gain is sacrificed for low Ico & high c-b blocking voltage.

It's all about tradeoffs. Ic is produced by Ie. Ib is needed, but we wish to minimize it. If we minimize Ib too much, Ico goes through the roof, & Vce,blocking plummets. So base doping is optimized according to how high Vce must block, & how low Ico needs to be. Higher voltage devices have lower beta. Can't get around that. Did I help?

Claude

 

[Studiot]

Thanks again Studiot for your feedback.

When I look over this thread I see many of the same points and formulae that were raised in the 2008 thread I linked to.

In particular, since others may have missed them

1)What happens if you try to force current control via the base.

2)What happens if you force voltage control via the base.

3)What happens at different frequencies - a very important point since the bjt is not only a DC device.

4)Whether you are interested in the internal workings of the bjt or worings of a circuit using a bjt.

5)The role of the emitter.

The 2008 thread I linked to contains some further information, including references to orifinal articles and other practical demonstrations on how to vary Ic including by varying the input frequency only (= control of Ic by frequency) and a great deal more on the photon aspect.

I would agree the demonstrations were contrived to show that anyone method only achieves partial control, since it can be subverted by another.
I also agree that the weight of practical experience of countless engineers and scientists ove the years have found current control configurations to be the most useful.

Thank you for some excellent insights.

 

[wbeaty]

Kittel - solid state physics as undergrad.
Muller & Kamens - Device Fabrication For Integrated Electronics at grad school, MSEE.
Sze - Physics of Semiconductor Devices, grad school Ph.D.-EE.

Thanks! I wanted to make sure we were "on the same page," approximately. My own text titles I don't quite recall, but I think they're in a cellar box from 1979. I'm almost certain that the main one was Sze above.

If you have reasoning you'd like to explain, sure, by all means do so.

OK. But first...

Do you now understand that my article is not aimed at engineers? Right? Did you take note of who the original intended audience was?

 

[Studiot]

Did you take note of who the original intended audience was?

That too is a very good point.

 

[wbeaty]

Comparing this website with the one I linked to in post#21 of this thread I think that 'Ratch' and 'Ratchit' are one and the same person. I note he signs himself Ratch in cabraham's E-Tech thread, although his handle is 'ratchit'.

Yep probably the same, but probably avoiding ban by mods. Following your links, I note a constant parade of odd questionable events there. Mods freely using personal attacks? And misusing moderator power while participating in battles with users? Multiplying sockpuppets and constant deceptive practice. The usual anti-trolls rule against namecalling is missing.

In June 2008 he started the thread I linked to (post#21) by referring to your site (amasci) as proof that 'transistors are voltage controlled not current controlled'.

<red sweaty face>BUT THEY ARE!</red sweaty face>

But seriously, my article originally was written in response to this question: what's a good way to explain the BJT to my grandmother? Or equivalently, how can we simplify the BJT down to the point where its operation becomes obvious to anyone?

Here's an appropriate comment: http://www.scottberkun.com/blog/2006/there-are-two-kinds-of-people-complexifiers-and-simplifers/"

Of course that question is entirely different from this one: what's the single best transistor model, the One True Path which all Proper chip engineers should use? Never ask that one, since it's a recipe for endless "Swiftian Battles."

A Swiftian battle is where two populations are driven to nearly-murderous rage over disagreement regarding the One True Way to crack an egg. In Gulliver's Travels, the two Lilliputan countries were slaughtering each other because every Proper citizen knows that morning breakfast eggs must be cracked on the pointed end ...and those despicable Outsider Others who disagree, they're disgusting heretics who need prompt incineration. :) Johnathan Swift clearly was well acquainted with the academics of his time. Tiny little men? Best way to "crack an egg?" Obvious rage and a desire to silence the blasphemous opponents?

My transistor article inadvertently triggers one of these battles when I wrap the original goal inside another one: "How do they REALLY work." The flames break out. Even Win Hill of Art of Electronics arrived and put in his two cents.

 

[wbeaty]

I also agree that the weight of practical experience of countless engineers and scientists ove the years have found current control configurations to be the most useful.

Ooo, here's an idea and a trick question inspired by the transistor war...

What if Ib was actually zero in BJTs?

What if hfe didn't exist? We'd be screwed, right? We could no longer use the linear relation between Ib and Ic. The diode exponential function would become part of everything. To prevent extreme distortion, only very small Vbe signals would be allowed.

Right? :)

 

[Studiot]

I think I have already described, here or in my references, the situation when Ib=0.

It is not my purpose to discuss the inner working of other sites, except to observe that progress is generally facilitated by goodwill. With goodwill moderation is often self moderation.

I have described, as has Claude, methods for calculating the variation of Ic with frequency by considering current or charge.

I have further described, in later posts in my link, an experiment which shows how to set up a transistor so that this change in Ic with frequency is independent of base voltage.

I asked Ratch for a derivation of equations showing how to calculate with the voltage only model in this case, with no response.

You have the same opportunity here.

I am not wedded to any particular transistor model. surely the 'best model' is the one which yields the required information with the least calculative effort. As such there is no best model, since models and calculations, correct at 2Hz will not suffice for 2GHz. A qualitative model such as the man cranking the variable resistor is very useful for achieving uderstanding but little use in real circuit calculations. I usually offer a man turning a tap (faucet) to control the flow.

go well

 

[cabraham]

Ooo, here's an idea and a trick question inspired by the transistor war...

What if Ib was actually zero in BJTs?
What if hfe didn't exist? We'd be screwed, right? We could no longer use the linear relation between Ib and Ic. The diode exponential function would become part of everything. To prevent extreme distortion, only very small Vbe signals would be allowed.

Right? :)

If Ib was zero, it would NOT be a bjt! How can an n-p-n structure have Ib = 0 when b-e jcn is forward biased? An E field exists in b-e region oriented so that holes move from base to emitter & electrons move from e to b. The n-type emitter has an abundance of free electrons, they being the majority carrier w/ the jcn under fwd bias. Likewise the base has an abundance of free holes, same reasoning.

It is impossible for an E field to selectively impart motion upon the emitter electrons & NOT on the base holes. How can Ib ever be zero? We can minimize Ib by doping base region very lightly w/ acceptors, much lower than emitter doping of donors. This reduces no. of holes injected into the emitter under forward bias, so that Ie >> Ib.

If we reduced base acceptor doping density to zero, we still have base current. Intrinsic silicon still conducts current in the presence of an E field. The device is now no longer n-p-n, but rather n-i-n ( "i" = "intrinsic").

To reduce Ib to zero literally means we must make the base an insulator. Then Ib = 0. There are 2 problems. The device is no longer "bipolar". The word "bipolar" literally means "2 polarities". An insulator for the base region makes the device unable to conduct electrons emitted from the emitter. When b-e is fwd biased, how can emitted electrons get through the base & onward to collector if said base is an insulator?

A device w/ Ib = 0 is simply NOT BIPOLAR. A fwd biased p-n b-e jcn in close proximity to a rev biased n-p c-b jcn is why the device is called bipolar. An insulated input electrode that forces Ib to zero is not bipolar.

We have a transistor constructed with an insulated input electrode such that the input current dc value is zero. It is a MOSFET. The JFET is similar only that the input jcn is reverse biased resulting in near zero gate current at low freq. Neither of these devices is classified as bipolar. Both are voltage controlled. They need gate current to operate. But Vgs is the quantity directly controlled, with Ig being indirect & incidental, yet important.

If a device has Ib = 0, it is not a bjt. It's that simple. How can a b-e p-n jcn be fwd biased w/o base current? Bill you have to examine what you're saying & do a sanity check. An insulated input device is not a bjt! Period.

How can I make things any clearer. If Ib were zero how do I explain bjt action? Well, if we make Ib very low, as in a superbeta device, Ie still controls Ic. My current control model, straight from OEMs, uses Ie to control Ic, not Ib. Ib is a factor, but Ie is what Drs. Ebers & Molls used in their 1954 paper as the control quantity for Ic. I have the paper if you need it. I'll email it to anyone.

Another question is "what if Vbe was zero?" An ideal p-n jcn under forward bias should have zero voltage at any finite current value. One could say that Vbe can vanish as well. If every electron emitted from emitter entered the p type base region w/o a potential barrier being formed, the case for a perfect rectifying p-n jcn, Vbe equal zero, but Ic is still alpha*Ie. This requires a material w/ a band gap energy of zero.

One can theorize zero Vbe and/or zero Ib. Real world semiconductors have non-zero Ib & Vbe both. If both vanished, we still have Ic = alpha*Ie. That is still the transistor action law. An emitter resistor is connected to the bjt, & input signal is at the base. Ideally zero current enters the base, & Vbe is zero. The entire input voltage appears across emitter resistor Re. Ic = -Ib + Ie = 0 + Ie = Ie. The output voltage of this emitter follower is exactly Vin.

Hence we have an ideal buffer. A voltage gain of exactly one (zero Vbe drop), w/ infinite input impedance (Ib = 0). Ideally Ib & Vbe can both vanish. But we always have Ic = alpha*Ie = 1*Ie.

In reality, like it or not, Vbe & Ib are facts we must deal with. I'd love it if my emitter followers could swing to the rail (Vbe = 0). I'd love to drive a 25 amp motor w/ zero base current in the bjt buffer.

I can wish for zero Vbe & Ib all I want. Wishing isn't getting.

Any other questions? Best regards.

Claude

 

[wbeaty]

If Ib was zero, it would NOT be a bjt!

OK. But first...

Do you now understand that my transistor article is not aimed at engineers? Right?

Did you take note of who the original intended audience was?

 

[wbeaty]

Any other questions?

Yes. Still looking for answers:

Do you now understand that my transistor article is not aimed at engineers?

When I stated who the original intended audience was, did you see that message?

 

[cabraham]

Yes. Still looking for answers:

Do you now understand that my transistor article is not aimed at engineers?

When I stated who the original intended audience was, did you see that message?

Yes, I am well aware of the target audience. Offering explanations that differ from established canon is not the problem. Bringing in info that counters the canon is where I have trouble.

I realize that QM, charge control, doping density, band gap, etc., is more than a lay person can digest. But you made statements that were downright contrary to canon. That is why I made my remarks.

Everybody has their unique way of explaining things, & I make no claims that my explanations are superior. Personally, I agree with your method of examining the device as 2 back to back diodes opposing. Also, it is all important that we distinguish between 2 diodes & an actual functioning bjt.

The best method I used to teach this concept to non-electrical majors (civil, machanical, chemical) when I taught at uni, was as follows. Make the base region so wide, say 1.0 mm, that every e- (electron) emitted recombines in the base region with an h+ (hole). Here we have 2 back to back diodes & nothing more. There is no transistor action whatsoever.

But there is a potential barrier, a depletion zone, & an exponential I-V curve.

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

Ic = alpha*Ie = 0, since alpha is zero for such an arrangement.

We have no transistor action at all, nothing but a fwd biased b-e jcn diode, & a rev biased c-b jcn diode. The logarithmic/exponential relation holds steadfast because it is a diode relation. We still have no bjt functionality.

Reducing the base region width results in an increase in collector current Ic since alpha is now increasing. When the base region is very thin, on the order of 1.0 micron, Ic is almost the value of Ie.

In the 1st case we had near zero Ic, the 2nd case shows Ic almost equal to Ie. What changed? NOT the potential barrier relation, not the I-V "Ohm's Law" relation, not the Shockley diode eqn. What changed is alpha. When alpha is near zero, we have a pair of back to back diodes per the exponential eqn above.

Ie is exponentially related to Vbe, Ic = 0. In the 2nd case, alpha approaches 1 & Ic = alpha*Ie = alpha*Ies*exp( ).

Bill I would recommend approaching the explanation of bjt action when addressing lay people in terms of the base being too thin to allow recombination. I personally like your 2 diode approach. If you start w/ a p-n jcn diode, explaning recombination, you can then advance to bjt. The difference is that the e- from the emitter pass through the super thin base so quickly, there is not enough time for recombination. The current from base to emitter literally overshoots the base & gets yanked into the collector.

Anyway, I believe that something along those lines might work for lay people. Again, I don't claim to be the best teacher at all. I get your point about the target audience. Thanks for your feedback & interest. BR.

Claude

 

[wbeaty]

Yes, I am well aware of the target audience.

OK, thanks! And yes, comparing narrow base to extremely wide base is an excellent teaching strategy of which I previously hadn't heard.

Regarding canon, earlier you stated that the Ebers-Moll model is a CC model, and their original paper employed current-controlled current sources. Correct?

Are you certain that you're not misinterpreting something? N.b. that your assertion regarding Ebers-Moll being a CC model goes entirely contrary to numerous undergrad Uni books which give the following CV (large signal transconductance) equation as the central feature of Ebers-Moll model:

Ic=Is*(exp(Vbe/Vt)-1) (for large hfe of course, alpha ~=1)
​

Is this not canon? If the above CV equation isn't "Ebers-Moll model," then you've discovered a vast flaw in an enormous number of textbooks.

 

[Studiot]

Cabraham's approach in stressing the width of the interdepletion region is excellent and fundamental - or at least it was when I learned about transistors back in the 1960s. (post#68)

Pierce devotes several pages to calculations of the effect of base width on carrier density, potential diagrams etc and has an excellent diagram on page 101, which I have appended.

For expansion of these formulae (posts #68 & 69) I also recommend the monograph by E H Cooke-Yarborough of the Atomic energy Research Establishment at Harwell

An introduction to transistor circuits (1957)

go well