Relevance of Physics
(Originally published in EDN Magazine, May 1, 2003)
I recently returned from a semester as a visiting fellow at Oxford University. The trip was a sabbatical assignment in which I would participate in undergraduate tutoring, hobnob with professors, and study. Yes, I still study, as does anyone who wants to keep doing a job they love to do.
The purpose of the sabbatical (besides the sheer fun of living with my family in Oxford for a semester) was to discover what major universities are currently teaching their undergraduate engineering students.
I was quite impressed with the tutorial system at Oxford. In contrast to the established practice at American universities with which I am familiar, Oxford students receive virtually unlimited access to highly qualified tutors. In the engineering department, tutors sit with groups of two or three students at a time for weekly tutorial sessions on electrical engineering. The university expects the students to attend the lecture, do their supplementary reading, and come to the tutorial session prepared with specific questions about their homework assignments.
The engineering curriculum for first-year students at Oxford includes a good amount of basic physics-despite the trend at other universities to de-emphasize that subject. In my session, students computed the mean drift velocity of electrons in a metallic conductor (very slow) and compared it with the average magnitude of the velocity of those same electrons as they bang around between the atoms (very fast). This discussion led to a fundamental understanding of Ohm's law for conductors and pointed out how it doesn't work at very high current densities (the conductor explodes) or at very low current densities (which are overwhelmed by thermal noise).
On another day, we discussed how the electrons banging around in a conductor cause it to heat up, which leads to the idea that resistors dissipate power. To illustrate that concept, we considered what happens when you short out a car battery with a wire. If the wire is too small, it melts. If the wire is sufficiently large, the wire doesn't melt, but you boil off the electrolytic fluid within the battery in a sudden and explosive way. (Do not try this at home!)
Lest you think these macroscopic-sounding examples don't matter for chip design, let me point out that it is the density of current that matters, not the absolute quantity. As our industry continues to press toward ever-smaller geometries, it is rapidly approaching the limits of the current-carrying capability of the conductors. For example, at tiny geometries, the fast-moving electrons within a conductor actually knock aside the atoms of the metallic lattice in which the current flows, a problem called electro-migration.
Regarding power dissipation and signal transmission, the highest performance chips that designers create today are running into the same fundamental limitations that affected ac-power generation and transmission equipment a century ago. Those limits are:
- The power that you can dissipate within a small region, and
- The power that you can rapidly transmit through long RC-mode transmission structures.
As the electronics industry moves toward higher speeds and denser circuitry, a firm understanding of power dissipation and signal transmission becomes increasingly important.
If someone you know is planning a career in custom-chip or ASIC design, encourage their study of physics. It's still relevant.