an electronish puzzle

Hmm, looking for an alleged discussion on Google, I come across this little tidbit on electronish matters. The subject was "Diode with very low reverse leakage current?" and came up on January 10 this year. The two quotes below, demarked with '| ' and '- ' (to reduce quoting confusion), were lifted from the Google archive via cut-and-paste.

A question arose whether the b-c or b-e junction would have lower leakage when reverse biased. Somebody had cited 'Bob Pease in "Troubleshooting Analog Circuits" (page 66)'

I replied thusly:

- With all due respect for Mr. Pease, I'm going to stick my

- neck out and disagree. Most of the leakage of a diode comes

- from surface leakage and thermally generated carriers in the

- depletion region. For a normally built BJT, the B-E junction

- has less exposed surface than the B-C junction and also has

- a smaller volume depletion region, being both narrower and

- occupying somewhat less area. So, unless one requires a

- reverse breakdown higher than the several volts available

- from a B-E junction, it is the better choice. (I am willing to

- be proven wrong on this, but it will take some evidence.)

- I suspect Mr. Pease's advice was directed toward the case

- where more than 3 to 6 Volts of reverse standoff is needed.

Some would-be resident expert chimed in: | Well you know that alpha_F x Ieo = alpha_R x Ico, so that right there | tells you that Ico is an integer multiple or two of Ieo. Pease is | *never* wrong, so it must be that the Early effect narrowing of the EB | depletion region causes that junction reverse saturation current to | surpass the BC diode leakage in some exponential way with voltages | smaller than breakdown.

People who were paying attention in their semiconductor physics classes, or worked at becoming knowlegable, know that the Early effect occurs because of depletion width changes alright, but in fact, the effect only applies to a BJT biased in the traditional active region. It occurs because, as the b-c junction depletion region widens, (with increasing voltage), the base region narrows, reducing recombination losses which, effectively, increases beta. (With fixed base current, you can think of it simply as the reduced losses going directly toward increased collector current.)

The notion, presented with equations and all, that Early effect even has any meaning when applied to individual junctions, is patent nonsense and a good example of the kind of plunge into the unknown I have mentioned.

The same comments apply to the idea that alpha_anything would apply to individual junctions. As those who know know, alpha is also related to recombination losses in a two junction scenario. It does not apply to just one.

I did not even answer then, and neither did anyone else, so perhaps Fred would like to defend such "expertise" now. I'm taking bets on the direction this goes.

"Fred Bloggs" wrote in message news: snipped-for-privacy@nospam.com... [invective and commands cut]

--
--Larry Brasfield
email: donotspam_larry_brasfield@hotmail.com
Above views may belong only to me.

Fred is, incorrectly I think, referring to Pease's puzzle in which negative c-b *voltage* is measured as a result of zenering the b-e junction. Pease attributes it to light being generated in the zenered junction and photovoltaic action in the collector. Your bias situation is different, although you may be referring to a similar optical coupling.

Anyway, I'm interested, especially is there is a non-photocoupled mechanism of carrier transfer when the b-e is zenered. Please explain.

John

"John Larkin" wrote in message news: snipped-for-privacy@4ax.com...

[Diode leakage plunge cut.]

I could not find the alleged discussion using words that would surely arise if "it" had been "covered thoroughly". ("It" being the puzzle I posed as the OP.) But I would like to see that puzzle. With his cute titles in the online archive, I could not see anything about b-c zenering. Do you (anybody else) have some date, link, or title?

No, to first order. Yes, to higher order effects. I would like to let the discussion develop somewhat, (at least as much as it can along rational lines), before justifying that answer. At that time, I will provide an explanation which will be cogent and convincing to those who know the subject matter. If you do not want to participate [1], I would be happy to email it to you before then. (I do not have it written yet, so the outline would be first followed by better writing and supporting argument.)

[1. This is not a "Who's got the cajones?" challenge. It can be a fun discussion, as I have discovered in half a dozen interviews with people who made the claim. ]

--
--Larry Brasfield
email: donotspam_larry_brasfield@hotmail.com
Above views may belong only to me.

Oh, I'll just wait. My comprehesion of semiconductor physics is small and mostly qualitative, which suits me fine: I'm a circuit (not IC) designer.

But you are being a bit coy, which might annoy some people.

John

....

Good point. To forestall that for those susceptible to reason, I should point out that I have had 27+ years to think about this and the opportunity to discuss it with some people who really do know the subject well, including one who had devised some experimental ICs and new basic devices. I'm not sure it is quite fair to me to come in as a regular contender under those circumstances. And if I meant simply to dump a tutorial, I would not use the "puzzle" subterfuge.

As I said, it can be a fun discussion. I mean to let it run that way, not "settle" it with all the noise that would incite among the resident idiots.

--
--Larry Brasfield
email: donotspam_larry_brasfield@hotmail.com
Above views may belong only to me.

While I believe that to be irrelevent to the B-E reverse breakdown puzzle, it goes to the diode leakage issue, in which I have directly relevant experience.

Part of the reason I was willing to "stick my neck out" is that I developed the core RC oscillator of a function generator that was once sold commercially. (Not many, as it turns out, but that instrument's performance was not the reason!) If I had ready access to this resource then, I would have been asking the same question. As it happened, though, I had to do some research, thinking, and experimentation to find a good way to get accurate steering of pretty low currents. (The low end frequency limit was 20 mHz or so.) My solution, justified by the reasoning earlier posted, some interesting relations between leakage bulk resistivity, and breakdown voltage, as well as some experiments with numerous parts, and finally use in a stable RC slow oscillator, was to use the B-E junction as suggested by my post in January.

You conjecture about experience, Jim. I can most certainly concur with what you say about "leads", (or, more to point, packaging), but the question does not ultimately depend on that problem, one which must be controlled at any rate when low leakage is paramount. And that view comes from real, hard experience, contrary to your suggestion.

--
--Larry Brasfield
email: donotspam_larry_brasfield@hotmail.com
Above views may belong only to me.

For Pete's sake, tell us or don't. Since no sane person deliberately zeners b-e junctions on transistors that matter, it's not like it's important.

John

And inexperienced. Anything with leads... leakage becomes totally unpredictable.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
|  Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
|  Phoenix, Arizona            Voice:(480)460-2350  |             |
|  E-mail Address at Website     Fax:(480)460-2142  |  Brass Rat  |
|       http://www.analog-innovations.com           |    1962     |
             
I love to cook with wine.      Sometimes I even put it in the food.

Yeah, that's OK. We used to use transistors as reference zeners: take an NPN, leave the base open, and use c:e as a 6.2 volt zener. For some transistors, the forward-biased b-c junction temperature compensates the b-e zener very nicely. Nowadays, it's safer to just buy a reference IC.

But zenering a transistor does permanently damage to beta, so once it's a zener, it's always a zener.

Does anybody know how much zenering (in coulombs or whatever) it takes to hurt a common small-sig transistor?

And some people claim that indirect-bandgap semiconductors won't emitt light!

John

Alright. Since nobody got interested in this, I'll keep this quite short and simple.

Zenering injects majority carriers into the base. They do not contribute directly to C-B current flow for the same reason that majority carriers there naturally do not contribute. One 2nd order effect is that the extra majority carriers reduce the equilibrium density of minority carriers in the base region, reducing what is usually thought of as the C-B leakage.

You've overlooked fault analysis. While you are right that no deliberate zenering is prudently done to a transistor later used with normal active biasing, (and expected to stay quiet or maintain its beta), fault conditions can lead to such "abuse". Then, the question becomes: What comes out of (or into) the collector? If that matters, then the answer does.

My original interest in the problem arose from an assignment by a mentor to go work out all the single point failure modes in a circuit where it was thought to matter. I'm sure he planned to check that work thoroughly, but the puzzle question was one where he doubted the result I claimed and so I had to persuade him, which I did. Since then, I used it to sort out the posers from the knowers in interviews where appropriate.

--
--Larry Brasfield
email: donotspam_larry_brasfield@hotmail.com
Above views may belong only to me.

Somebody putting those suckers in the wrong drawer again?

Do you have a klaxon rigged to the switch of your bench supply? 8-)

Does it really work that way? The Pease thing suggests that the zener lights up and illuminates the c-b junction, which would cause photodiode-mode leakage. So there would be opposing effects, I suppose.

There are, somewhere on the web, some cool microphotographs of light pouring out of planar transistor b-e junctions.

So, you won't hire people who don't know about this?

John

Oh there is no would-be to it- I will be here long after you're dead.

Hey everybody- read what the pseudo-intellectual and pretentious hot-air bag Larry Brasfield wrote- and this is the individual who clearly came on SED to kick ass- explaining basic op amp theory, Miller effect, logarithmic distributions and lots of other stuff to us morons. This boy is getting the sh_t kicked out of him repeatedly- apparently the moron knows nothing about Ebers-Moll and SPICE because forward and reverse current transfer ratios and forward and reverse Early voltages are used to account for transistor operation for all combinations of forward and reverse bias operation- the Early voltages in particular are used to account for the non-zero conductivity of reverse biased output junctions that are current controlled. This Brasfield crap about "the effect only applies to a BJT biased in the traditional active region" is total horseshit and just plain wrong! Once again the pretentious windbag gets his ass kicked- but there will be no reaction- he is a psychopathic narcissist and pompous ass.

In article , John Larkin wrote: [...]

Who are you calling insane!!!!! :>

The pop-tronics surf synthesizer used a back biased E-B junction as the basic white noise source. It also used 3 free running multivibrators to make a "random" voltage to feed the variable gain.

Don't try this trick with a Darlington unless you want an only slightly random saw tooth.

BTW: If you do back bias the collector junction with lets say 20V and force a moderately large current into the E-B junction, the meter on the collector circuit twitches up scale just as there is a bright flash and a loud bang. I think this is due to the pyro-electric effect in the semiconductor more than from the plasma conduction, but I'm not sure. I think further experiments are in order.

--
--
kensmith@rahul.net   forging knowledge

Fred is right. He may not put it very diplomatically, but he is right none the less.

Purely descriptive nonsense and absent of any real meat like the energy band theory of solids.

There you go again with another of these incessant "let me tell you how I put someone down again" stories...pathetic.

Really? Too bad you can't self-sort.

"John Larkin" wrote in message news: snipped-for-privacy@4ax.com...

I'm only stating the effect of the 2nd order effect I mentioned. There are a few others. You have to add them all up if you're nuts enough to want to predict how they will show up in real devices.

That is an interesting effect, if it occurs. I have not researched or studied light emission by silicon diodes, having learned in school that Si is very inefficient at that due to its indirect bandgap structure. What puzzles me about the "emission while zenering" claim is that the extra carriers that have tunneled across the junction are just like the ones normally present. It would only be via recombination that there would be emission, and the majority carrier injection at any normal current is low enough that it should not change the density that much. The clincher is that minority carrier generation is essentially unchanged, and that would limit the emission rate whether some extra majority carriers were floating around or not. So I am quite skeptical of that claim. (But willing to be educated by anybody who really knows better.)

Those wouldn't be GaAs transistors? If it was Si transistors, I would love to see a link or a few keywords to aid a search.

Please, John. That "appropriate" would be poorly used if it meant what you suggest. I tried to avoid duplicating my criteria for asking about that, already mentioned here.

Why would I consider someone a poser who did not know that? If you knew me, I would not have to tell you the answer: Only if the interviewee's resume claims an understanding of semiconductor physics do I even bring out that question. (That's why I've only done it maybe 5-8 times.) Even then, if they are able to give a wrong answer, but with some supporting reasoning that makes sense (given their assumptions), I go on to discover if they can back out when we revisit the wrong assumption(s). (The few who get this wrong instead of just giving up usually have forgotten some basics of PN junction tunneling.)

I've never been the hiring authority, but I have often given a good hire recommendation to applicants who get a few things wrong in a tough interview yet are able to get to the issues when discussing those errors. Likewise, I have given a definite thumbs-down to a lot of people who are smart but much less smart than they imagine or pretend.

And just to quash the notion that I conduct a mean interview, several of the people who suffered thru it are among my longtime friends. One of them even shows up here from time to time.

--
--Larry Brasfield
email: donotspam_larry_brasfield@hotmail.com
Above views may belong only to me.

theory of solids.

I would not expect you to understand it, Fred. And while I imagine that you could copy many bits of theory into a post, I doubt that it would be to make the point ostensibly at issue.

someone down again" stories...pathetic.

What is truly pathetic is your projection of your own miserable outlook into a fantasy about what was a mutually enjoyed, healthy relationship between that mentor, (to whom I owe a lifetime of thanks), and me.

To suggest that I was involved in trying to put him down, as he continued my education by insisting that I defend my conclusion, is just plain sick. It reveals the depth of your depravity and decline.

--
--Larry Brasfield
email: donotspam_larry_brasfield@hotmail.com
Above views may belong only to me.