Brian Elfman is President & CEO of Certified Devices Inc. (CDI) in San Leandro, California. He is the founder of a public Silicon Valley technology company (CAERE Corp. NASDAQ). His engineering design background spans the era of discrete transistor logic to the transition of ICs. In each of his companies, he has filled the roles of President, leading-edge market visionary, and principal product designer. He has raised both venture capital and private financing for his companies. In the main, he generally has designed and built products first in their field. Click here to read Mr. Elfman's complete bio. If you would like to ask Brian a question or simply give comment, click here.

Keeping Hubble Happy—A Bit on the Advanced Camera Upgrade Following its historic mission and installation of $172 million of retrofit and upgrade gear, Columbia landed routinely on March 12, 2002. The third-generation Advanced Camera for Surveys (ACS) camera itself is a marvel of astrophysics photography. To achieve all this, among other things, took highly disciplined opto-electronic hardware and software engineering. ACS represents over eight years years of design, testing, and characterization.   Click here to read the complete article.   Time Interval Measurements — Getting Down to a Picosecond Single Event Measuring single-event time intervals (TIM) to a fraction of a nanosecond (ns) once was almost exclusively either an advanced weapons systems design need or a high-energy physics experimental function. Very specialized markets. Even today, measuring single-event time intervals of 1 Ns and less is not exactly duck soup. With gigahertz speeds now commonplace, jitter rates and stability factors in general require that accurate single-event subnanosecond time interval measurements be at hand. See the sidebar below for more information on how to measure picosecond time intervals.   Click here to read the complete article.

Time Interval Measurements — Getting Down to a Picosecond Time interval measurements are made in a repetitive or a continuous mode, a fixed-period mode, and a single-event mode. Mainly, the resolution is mode-related. It will be apparent that the best resolution runs with accumulated successive measurements. Conversely, the least resolution is a single-event mode. A couple—three editions—ago I described the function of a measurement comprising a target (or unknown) value, a measured value, and a true value. These same attributes apply to time measurements as well. To demonstrate this, let's use a hypothetical case. We'll use microseconds for illustration. Assume we're going to measure a 1 (one) microsecond interval. And let's establish a clock derived from a 1 MHz oscillator. We'll further set the clock at two pulses per cycle, or two-to-one. What's more, let's assume aging and temperature coefficients to be of no impact on the measurement. Perfect stability. Now when START (of the TIM) is enabled, the point of enablement will be arbitrarily at a point between clock transitions. The same for when STOP (of the TIM) is enabled. So we have a pulse train of 2 MHz (the duty cycle is not important) to totalize our now measured time interval. Did we measure 1 µs? Nope. It was not 1 µs. Here's why. START was enabled at an arbitrary point between clock transitions. So was STOP. These two increments cannot be counted. (We'll stay away from the different schemes used to account for parts of these two increments. See SRS Application Note #2 — Making Measurements with the SR620.) So there's no chance we would have totalized even one increment. The most we can say is that the resolution of this hypothetical case is 2 µs. In 2 µs we'll be guaranteed two full increments. In other words, as the time interval measured approaches resolution, the error increases towards the period of one clock increment. This error is introduced by the time variance of one increment approaching a full interval between transitions. Remember that the functions of a measurement comprise a target (or unknown) value, the measured value, and the true value. This should help us to better understand the SRS specs in my TIM article (Time Interval Measurements -- Getting Down to a Picosecond Single Event). Note where I discuss the Stanford Research Systems (SRS) SR620. You'll see that resolution is stated in the SRS spec sheet at 25 ps, the relative accuracy is stated as ± 50 ps, and the absolute accuracy is ± 500 ps. This is explained further in the SRS Application Note, and if you're doing TIM you should consult the application note. (You can download the catalog that contains the application note, or you can download the manual, or you can request on-line info: SRS Application Note #2 — Making Measurements with the SR620.) Click here for more information on measuring picosecond single events.

EDA Tools: Use the EDA Tools directory to find companies and their products within specific EDA disciplines.   Design Center Reference Library: This is your access point for lots of good information covering applications, design tools, consultants, intellectual property, trade shows, hot ICs and standards.   Standards Watch: Check here for news on proposed standards from IEEE, IEC, etc., and ad hoc industry groups.

Digital/Analog Product Design Archive

We need a few good EE Experts to present useful information and perspective to other members of the EE community. Your audience will be working engineers much like you, who value suggestions, directions, and timely solutions to problems of the workplace. To qualify as an EE Expert, you'll have to demonstrate knowledge of your subjects of interest. Click here and become an EE Expert today!

Guides and Experts   Analog Avenue   EDA Tools   PLD   DSP   EDA   Embedded Systems   Power   Test