Electronic Design

Bob's Mailbox

Dear Bob: I understand your views about Spice. I'm sure you're aware that companies periodically appear to promise better analog EDA, but most seem to fail. (There's a promising firm called Barcelona, but it's not far along yet.) Here's my question for the master (meaning you, of course): Do you think that the process of analog circuit design will ever be "automated" to the degree that digital circuit design is now "automated"?

(Absolutely not. Anybody who claims that he or she can automate the design of any analog function is thinking of subtrivial analog functions—that is, analog functions that nobody would want to buy. See below at "CHD." Even the layout of analog functions on a pc board, or on a chip, is usually nontrivial, and it typically can't be done by automation or by computer. In any high-performance system (40 milli-in.2 or 10 in.2), there are usually serious constraints on where different blocks get put or how they are connected. As you well know, "ground" is NOT "ground." In many circuits, there is no ground that can be treated as "ground." Hot components heat other components. Wild signals get transmitted to nearby components. Also, if you're not careful, they will be transmitted to the outside world as RFI.

Moreover, anybody who says that his computerized system, or EDA, can do most of the layout does not understand the situation. He might lay out somebody's circuit, but not one that's high performance. You are familiar with the phrase in which CAD stands for "computer-aided design." If you get involved with this, then you might appreciate how the phrase "CHD" means "computer-hindered design." We see it all the time. /rap)

If so, do you have a time frame when this (analog EDA) might take place?

  • Geoffrey James (via e-mail)
  •  

  • Pease: Do you have any schedule for when Hell is expected to freeze over? About as soon, I'll say.
  • Dear Bob: You did it again! I read your articles about the K2-W (Electronic Design, Jan. 6, p. 20 and Feb. 3, p. 22), and sentimental feelings overcame me. This tells you about my age, and it was/is interesting to see the tremendous advances in the electronics field.

    I had the fortune and opportunity to take a tiny part in the process. I am talking about the 1950s, when I was working at the Allen B. Dumont company in Passaic, N.J. I was a design engineer fabricating radar tubes and circuitry for the Navy. One of the early experiments was a huge contraption to compute elementary math.

    The storage media was a vidicon-size vacuum tube called Radechon with a fine nickel mesh plus an MgF2 layer evaporated on it as the storage surface. It was essentially a scan converter—one could read out the data at any arbitrary scan system and speed. This tiny tube was serviced by four 9-ft tall racks with several hundred flip-flop and other tubes. We had to arrange an extra three-phase power feed with about 15-kW capacity to fuel the few kilobyte storage process! It came in "handy" at winter times. We would turn on the racks and warm our frosty hands with the hot air emanating out of the tubes.

    At any rate, I just wanted to say that I enjoy your wisdom while I'm repairing and restoring ancient vacuum-tube radios! Best regards.

  • Paul Lukas (via e-mail)
  • Pease: What a wild old machine! Obviously it was an experiment that worked, but it took too much maintenance and was surpassed by less (not so desperately) complicated schemes! I just built an old 1943-design op amp using two 6SL7GTs, and it worked darn well! A pair of them would make good mitten-warmers!
  • Dear Bob: In the March 3 issue of Electronic Design, the recipe for Unbaked Bourbon Balls appeared. It calls for "Karo Syrup." Do you use the dark or light variety of Karo Syrup when you make the cookies?

  • Kevin Kocher (via e-mail)
  • Pease: I think I usually use the light stuff, but I use whatever is around. There's not enough of it in there to make much difference!
  • Hide comments

    Comments

    • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

    Plain text

    • No HTML tags allowed.
    • Web page addresses and e-mail addresses turn into links automatically.
    • Lines and paragraphs break automatically.
    Publish