Art of Electronics, 3rd edition, corrections

I totally understand. ...Jim Thompson

Far too many elementary electronics texts state that in order for a BJT to be in normal bias, the CB junction has to be _reverse-biased_.

That's actually true if you account for the built-in junction potential, but not if you're measuring voltages in the external circuit. Good BJTs will operate nearly normally with V_CE ~ 0.3V or even a bit less. They do slow down on account of the increased capacitances.

Current mirrors are full of devices where V_CE = V_CB ~ 0.6V.

Cheers

Phil Hobbs

No- it's just a bunch of junk written by an intern about 40 year old techno logy. If you read the HDK in detail, you will find the dI/dt is spec'd for "recom mended" gate drives, which are anything above the maximum minimum gate trig ger level. Of course, it's understood that once the gate is triggered, the drive is applied long enough for complete turn-on.

It's going to be hard to destroy something more modern like this (Fig.10):

Those old zener circuits predated the precision threshold gate drives but t hey're still usable for low energy discharges. You can use them to discharg e the massively large energy storage circuits by going with something like this:

Please view in a fixed-width font such as Courier.

. . . . . ----+-------------------------------+---- . | | . | | . | --------||--------+---/| . | \ | | | . | | | | | . | +-- | | . |0.1u | | | . === [Rz] [Rg] | . | | | | . -----------+------------+------+----+---- . . . . SCR1 small . . SCR2 large . . .

Whereas SCRs are so new, and all. ;)

We were discussing the hoary age of the problem, after all. And if you have to make up lies about its author to discredit it, I think my point is adequately made.

If you read the HDK in detail, you will find the dI/dt is

But the zener circuit doesn't always do that, especially with slowly varying inputs. Something with some snap helps. The zener circuit isn't guaranteed to destroy the SCR, but as the databook says, using something just a bit better makes sure you get the rated capacity of the SCR. Your circuit looks like it would work fine, but it has a lot of parts compared with, for instance, using a 723 or 431 type voltage regulator chip. (The MC3423 has been discontinued.)

Like I said, the problem is simple to understand, easy to fix, and has been known for decades.

Cheers

Phil Hobbs

Actually they are...do you think stuff like the azipod 17MW 6600V synchrono us motor cycloconverters are built from 1960s thyristor technology?

The particular section you quoted was redacted by a marketeer.

The thyristor has all the "snap" you need. Once the main terminal currents hit a threshold where the sum of junction current gains ( current dependent alphas) exceed one, it takes off on its own, that threshold should occur at spec'd IGT, and that abusive gate drive of 5x abs max they quoted isn't going to make things switch much faster.

What a horrible data sheet. No definitions, few contexts, no test circuits shown. Fig 10 doesn't mention the gate drive.

I'd expect that you could destroy it with a simple zener crowbar circuit. I certainly wouldn't assume that a soft trigger is safe.

"Modern" chips tend to be physically small, which can make them cheaper and less rugged.

ng

hnology.

commended" gate drives, which are anything above the maximum minimum gate t rigger level. Of course, it's understood that once the gate is triggered, t he drive is applied long enough for complete turn-on.

):

Right, TECCOR doesn't know what they're doing...

Whoever writes their data sheets doesn't.

SiGe's and and in particular cryogenic SiGe's are agile to this: the B-E drop increases to 0.9 - 1.1 V ballpark and you can pull the collector significantly below the base without ... well, I would hesitate to say 'any' ... er ... noticeable adverse effect.

There's the neat circuit doi:10.1109/TASC.2012.2227638 by Drung utilizing this:

Vsupply

I .--o--. I I .----o o---. I I I I I I I I I I/ \I I Vin+ ------I I------ Vin- I I>

The errors are beginning to stream in now, see our errata page,

I keep seeing references to chapter 3x - but cannot find any chapter 3x. For instance p170 col 2 line 15. "Going beyond this first look at jfets, chapter 3x includes some ..." or for instance last line of p167 where reader is referred to para 3x.4

Is the "X" a placeholder that was going to be filled-in?

piglet

As we explain in the Preface (fittingly on page xxx), a companion book, "The Art of Electronics: The x-Chapters" will be coming out in a few years. This book will have piles of advanced material we couldn't fit into the 3rd edition. For the portions we've already written, we can give you some cross-references now. These are indicated with an x after the section number, etc., for brevity. You'll find these x-Chapter cross references all over, but IIRC, especially in Chapter 3.

Cute. That's a lot like the two-JFET cascode trick, with a higher-Vp part on top and a lower-Vp one on the bottom.

Cheers

Phil Hobbs

Ahh, that explains it! Thanks Win, I was so excited when the book arrived that I just dived straight in without reading the preface. Waiting several years is going to be tough but worth it.

piglet

I think there is a chicken and egg problem here - few people buy OTAs so the marketing departments at chip companies don't know that they are nice things, so none of them have been designed for decades, so there are none of them with rail to rail outputs or modern processes etc. so few people buy them.

They are used a lot within chips - perhaps as much as opamps.

My favourite use is for driving the gates of large MOSFETs in things like constant current sinks. The gate capacitance helps the compensation and it is stable, whereas if one uses an op-amp then the op-amp output impedance and the gate capacitance forms a pole, and along with the other pole in the op-amp compensation, and any parasitic poles it is very likely to oscillate.

Chris

The LM13700 is pretty nearly rail-to-rail, if you don't use the buffer. My main issue is that it's too slow for almost anything--only about 1 MHz at maximum I_ABC, and slower still at lower current. I have a couple of dozen VTC VA713 and VA2713 70-MHz OTAs in my drawer from 20 years ago, but they're long obsolete.

Bits of them, for sure--diff pairs driving current mirrors for level shifting. Dunno about full OTAs, but maybe so.

That's an interesting approach--put the dominant pole on the output. Parts like the LM8261 do that as well.

Cheers

Phil Hobbs

What characteristics are desired in an OTA? Defined gm? Dominant pole at output? ...Jim Thompson

Lots of RRO amps are c-load stable. AD8565 for example.

pole at output?

Differential inputs, low output distortion, low offset, optional linearizin g diodes as in the 13700, and much higher speed, say >10 MHz at 100 uA tail current.

TI makes a weird part (OPA860) that they call an OTA but it's more of an im proved BJT.

Cheers

Phil Hobbs