Signals & Noise: February 15, 1996
to send a letter to signals & noise via email: |
Add the following microprocessor to the 16-bit chips in EDN’s 1995 µP/µC Directory (Sept 14, pg 33): the Harris Semiconductor 80286. The vendor manufactures commercial, military, and industrial CMOS versions of the 80286. Prices are $8 and $12.70 for the 16- and 25-MHz (commercial) versions, respectively. You can reach Harris Semiconductor (Melbourne, FL) by phone at (800) 442-7747.
Regarding "High-speed-connector systems" (EDN, Oct 26, 1995 pg 40), a time-domain relectometry (TDR) waveform does not accurately represent the impedance profile (impedance vs distance) of the path being measured, except in the case of a single, simple discontinuity. When a device such as a connector causes complex reflections, superimposed ringing from the device can cause the TDR waveform to become muddled. This ringing gives a misleading representation of what is occurring on the other side of the connector.
For example, in Figure 7 (pg 50), I presume that the second discontinuity is somewhat larger than the 25-psec waveform indicates. The second discontinuity has been smeared and attenuated by the effect of the first (that is, the TDR-to-daughtercard connection). A correct characterization of the impedance profile can be important for Spice and other simulator models.
The relationship between a TDR waveform and the true impedance profile is rather complicated in the case of a nonuniform profile with multiple discontinuities. However, a surprisingly simple algorithm transforms a TDR waveform into its true impedance profile and mathematically unravels the complex interactions (see "TDRs profile impedances of backplanes and pc-boards," Electronics, July 14, 1981, pg 113). A number of years ago, I used this algorithm in an application (backplane wiring and printed-circuit impedance profiles). Designers have used the algorithm in several applications ranging from power lines to microwave waveguides. The algorithm requires roughly N2 floating-point multiplications for a TDR waveform digitized into N points, making it possibly fast enough for real-time display. I do not believe that algorithm is commonly known (in particular, within the TDR industry), and I thought it might interest your readers.
Norman D Megill
Production Services Corp
Belmont, MA
| Sound off |