FOIL Method Comes In Handy
I just read your article about the continued relevance of paper and pencil, and I couldn't agree more heartily \["Just Being New Doesn't Make It Better Technology," March 4, p. 22\].
Consider the following anecdote from a few years ago when my sons were still in high school. One of the boys had a math problem, in which, among other things, he had to perform a multiplication (something like 53 times 74). While he searched for a calculator, I cranked the answer out in my head using the "FOIL" method.
Clearly, for that situation, my method--born of the 1950s and 1960s when I went to school--was superior. Certainly, I would opt for the calculator if I had a dozen such problems, or if I needed to multiply accurately a couple of 10-digit numbers. But a lot of life situations don't require that type of repetitive action or accuracy. My belief is that the rising generation has been short-changed on both the tools and the common sense to pick the right tool for a specific situation.
I have met people who seemingly cannot write a letter without e-mail. I would be the first to admit that e-mail is often convenient, and text processing allows one to edit, reread for just the "right" tone of address, etc. But sometimes just a hen-scratch on a piece of parchment is faster and more convenient, and it gets the job done.
So far I have not found it urgent in my business or personal life to carry a portable computer of any type (PDA, laptop, and the like). It would help on occasion, but it isn't worth the hassle to carry around for the rare convenience that it offers.
Besides, one needs to consider that to use such a device, there are issues like the need for a telephone line, power, and so on. There are restrictions on use in air travel, and the bother of security. There is always the concern for loss and theft. (I recently read of a huge number of PDAs left behind at airport security screens!) Additionally, the programs available don't always do the very simple tasks that I can readily use a paper and pencil to perform.
My wife is (or has at times been) an AOL addict. She loves to get online and do the "instant messenger" thing. On several occasions I have pointed out that she could use the regular telephone and get about three times as much mileage over the century-old voice technology as over the modem, especially because neither of us types very fast.
The cost of long-distance phone service is one of today's genuine bargains, and a phone call is always more personal than a hastily typed message. The time or two that I have tried "instant messenger," I have found that the other party had already moved on to a different topic, while I was still trying to reply to the last comment.
Faxing, PDAs, e-mail, and the Internet all have appropriate niches. But none will ever be the perfect solution for everything, and frequently I just need a paper and a pencil!
A Bright Future
Your article "Will Analog Circuits Continue Scaling With CMOS?" \[April 15, p. 28\] spurred me to write. If it weren't for innovation like what Impinj is doing, analog would be deprived of the benefits of Moore's Law due to the fact that analog usually requires bigger-area transistors for matching to achieve good analog performance.
However, bigger also is slower. With this kind of innovation, I think that we have broken through that barrier, at least for a while. We will still be able to make progress in separate analog and digital chips. The bigger problem is with mixed-signal ICs. The close proximity of massive fast digital to very sensitive analog circuits makes really high-dy-namic-range systems nearly impossible.
Still, I'm sure that innovation will find a way around that. In fact, on-chip so-lutions might be easier to do than at the board level. With good on-chip isolation, and proper control of grounding, power, and trace routing that can be extended off chip, it may be possible to do an even better job of containing the nasty aspects of fast digital interference. This will allow the analog circuits to perform their best.
Fully differential analog circuits with high common-mode rejection at high frequency can help. Low noise, LVDS (differential) off-chip interfaces may help, too. Plus, having enough on-chip processing so that no intermediate results are brought off chip for further noisy processing can help. I think that our future is bright.
Finer CMOS Process Geometries
Of course everybody would like analog circuits to scale with finer and finer CMOS process geometries \["Will Analog Circuits Continue Scaling With CMOS?" April 15, p. 28\]. Techniques mentioned in your article like calibration and self-adaptive silicon promise to help, but in noise-limited situations, the kT/C requirement imposes a square relationship between capacitive load and dynamic range reduction with the power-supply voltage.
So to maintain speed, more power will have to be consumed in a lower-supply-voltage situation whenever noise is a limiting factor. This is a fundamental (and somewhat nonintuitive) result of basic noise analysis.
Got an opinion? Comment? Suggestion?
We want to hear it. Send your letters to [email protected], to specific editors (see the masthead), or to:
611 Route 46 West
Hasbrouck Heights, N.J. 07604
Include a phone number or e-mail address we can use for verification. Letters may be edited and may be used in print and electronic media.