[SI-LIST] : Determining CMOS Cin with TDR methodology

Dr. Edward P. Sayre ([email protected])
Wed, 15 Apr 1998 10:13:24 -0400

Mike & SI List colleagues:

I would like to amplify Mike Jenkins' (LSI Logic) comments regarding use of
the TDR for capacitance measurements. He points out that there are . In
fact, any discontinuity or time effect which is proportional to the first
time derivative (inductors, forward x-talk) has the invarience of an
integral of the time waveform. There are well documented time domain
methodologies based on integration of the waveform which are constants and
independent of risetime and found in the "old TDR" literature from TEK and
HP. With proper scope waveform software, it is possible of both computing
the area under the curve as well as exporting the waveform that you note
below into a SPICE environment. (See our web site for a "closing the design
process" paper that describes the process). By utilizing proper and
accurate SPICE simulation (including the fixturing), it is possible to
determine the desired capacitance very accurately. Use of the TDR
measurement for Zo and the first approximation for C based on the
integration of the waveform sharpens the first SPICE guess considerably.

At 12:43 PM 4/13/98 -0700, you wrote:
>If you don't mind, I'd like to recommend an alternative method for
>measuring pin capacitance. L-C meters may be fine, (potentially)
>accurate instruments, but, as you've seen, they're not much good
>for telling you what's going on.
>My preferred method uses a TDR with a scope capable of waveform
>math (subtraction and integration).
> ____________________ ___
> | * | ___________
> without | * | ___________()--- +
> cap -->| * <-- with cap unit Zo |
> | * step C ===
> _________|* _|_ _|_
> ///
>Without the cap, the TDR shows the classical open-circuit doubling
>reflection. With the cap added, the reflection adopts an RC time
>constant. The time constant is Zo * C.
>Of course, real waveforms don't look like this -- ringing and other
>ugly stuff. However, the Zo*C time constant is still the integral
>of the difference between the "without" and "with" waveforms. This
>is exact, even with arbitrary incident step waveforms and even with
>inductance in series with the cap. And the TDR trace lets you see
>the effect of such inductance. The accuracy is as good as your
>equipment, probably limited by fixturing (including knowledge of Zo).
>I'll get off the soapbox, now.
>ps: If you're interested, the proof involves the LaPlace initial and
> final value theorems.
> Mike Jenkins Phone: 408.433.7901 _____
> LSI Logic Corp, ms/G750 Fax: 408.433.2840 LSI|LOGIC| (R)
> 1525 McCarthy Blvd. mailto:[email protected] | |
> Milpitas, CA 95035 http://www.lsilogic.com |_____|

| ------------------------------------- |
| "High Performance Engineering & Design" |
| Dr. Ed Sayre e-mail: [email protected] |
| NESA, Inc. http://www.nesa.com/ |
| 636 Great Road Tel +1.508.897-8787 |
| Stow, MA 01775 USA Fax +1.508.897-5359 |