Art of Electronics, 3rd edition, corrections

Fig 3.36B Add an output

P 157 typo am plitude

P 219 some Avago phemt devices sure behave like enhancement jfets

P 265 might mention c-load opamps, like LM8261

P 294 not a mistake, but a BFT25 makes an incredible low-c, low leakage diode

P 343 Opinion: the 34401A is a mediocre DVM, especially on AC.

P 452 Could mention coaxial ceramic resonators. And YAGs.

P 690 The zener + SCR crowbar is not robust if it has to discharge a lot of capacitance. SCRs can fail if soft triggered, sort of a second breakdown effect.

How many printings will Cambridge accept corrections for? Wiley cuts them off at the 4th printing.

Cheers

Phil Hobbs

The thyristor self-triggers once IGT threshold is crossed. Underdriving IGT only makes a difference when there is not enough anode current to cause the conduction to spread across the entire device cross-section. How could that be the case with a crowbar?

GT only makes a difference when there is not enough anode current to cause the conduction to spread across the entire device cross-section. How could that be the case with a crowbar?

It's the dI/dt limit: the thyristor cannot deliver full current in its firs t microseconds, the conduction region must spread from the gate electrode across the area of th e cathode. In a power device, that can be a few millimeters, and take a lot of microse conds. There's fast turn-on for photoSCRs, however.

see sheets 5 and 6.

It was a pretty well known problem, back when linear supplies were more common. Special chips e.g. the MC3423 are available to fix it, by triggering the SCR hard even as the supply voltage is collapsing.

The I**2 t handling capability of the SCR depends on the trigger characteristics, because the triggering happens at the contact and propagates across the die. (Centre-contact SCRs are better than edge-contact ones for that reason, see the MC3423 datasheet.) In the zener circuit, the SCR starves itself of gate drive as it's firing, which slows down the propagation. The I**2 t capacity goes down basically as the fraction of the die area that's actually fired, AIUI.

Cheers

Phil Hobbs

Yeah, well, when you've got it as error free as the 2nd edition is, let us know so we can all go out and buy it. :-)

Too late, everybody's ordered one already. ;)

It's amazing how you can go over and over a manuscript, and then as soon as the finished product arrives, you flip it open and start seeing mistakes. I found about half of the errata in my first edition myself, which took me down a peg, for sure.

Cheers

Phil Hobbs

I finally got notice today from amazon that mine shipped. (the free shipping option probably puts me last in line.) Not that I'm in any hurry, any "mistakes" I find will likely be my own, not H&H's. :^)

George H.

It works with software too, and hardware RTL designs being sent off to the fab, and I think a similar principle applies to interpersonal relations as well (dropping letters in the mailbox, hitting "Send" on the email, etc.)

There's something about taking a non-reversable step of commitment, which manages to un-mask numerous forms of selective blindness.

Real Life is lamentably short on "Undo!" buttons which actually work.

Personally, I use a TL(V)431 with a 2N3906 common emitter inverter. Precise threshold, accommodates hysteresis, fast yet modest gate drive.

Tim

I'm not aware that they have any cutoff. In the 2nd edition we stopped when corrections dropped to nearly zero. I suspect it's pretty much up to the dedication of our editor, which is awesome.

Cambridge University Press is awesome, in general, because even tho they publish relatively-obscure monographs, etc., they don't discontinue books when the sales rate drops. But then, of course, our sales rate never really dropped all that much.

::-))

Thanks John.

How exactly do you use it? When the Ref input is below 1.24V, what does the Cathode pin do? How low does it go (unspecified). Low enough, below Vbe, to run an inverting transistor, EF to gnd? We need some more details.

Example:

The top 100 ohm sets gate current. The bottom 100 ohm sinks gate leakage to ground, and sets positive feedback into the sense divider R1-R2. The top 220 bleeds away bias current, and the bottom 220 prevents the TL431 from ripping the BJT apart (TL431s are rated for 100mA, though I admit I haven't measured what they typically self-limit at -- it could be much more?).

It's been a little while since I measured it, so I don't remember exactly which order it's in (Vref or Va doing what), but what I remember is that a TL431 acts something like a BJT with very well defined Vbe and very high Gm (at DC), at the expense of a somewhat leaky sub-threshold region (so to speak). Like, it starts turning on (uA to

That chip was used mainly to obtain a precision trip threshold and transien t overvoltage immunity. There is no advertisement whatsoever about driving the gate hard even as the supply is collapsing.

I2t in the case of a crowbar applies to fuse selection and not withstanding current surges. The current surge rating for a switching application is co vered by ITSM, the maximum single shot switch current that can be handled w ithout causing a damaging junction temperature rise.

The regenerative current from the anode is way more than anything you're go ing to inject into the gate. If the voltage across the SCR is starting coll apse then you already have a pretty hefty anode current and the SCR is not likely to abruptly turn off. Larkin is always running his components outsid e safe operating area limits so it's no surprise to hear he was blowing up his crowbars.

"Fred",

Do you actually design anything, or just rehearse your grudges on Usenet?

The zener circuit has been notorious for blowing up SCRs needlessly, for at least 30 years that I know about. That has zilch to do with personalities--yours, mine, John's, or anyone else's.

And if your views are so transparently just, why not use your real name, like a grown-up?

Cheers

Phil Hobbs

I thought it was a good sign that he at least said something electronics related.

George H.

Well, a better sign, anyway. ;)

Cheers

Phil Hobbs