D. C. Sessions makes an excellent point when he questions what we
mean by output impedance of a driver. He has given three examples
of what we might mean. An IO driver will give three very different
values for these three definitions of output impedance.
Several years ago, it was common to identify a driver by it's
current rating (ie 4mA or 24mA driver). This made sense back in
the TTL days when we had a pull-up resistor, but is pretty
meaningless with CMOS drivers.
It has become popular to rate drivers with an impedance number,
possibly to indicate the impedance of a line that it can drive.
But, as Mr Sessions has pointed out, we are dealing with a very
nonlinear IV curve (N-FET or P-FET Id vs Vds curve). Both the
ratio of V/I and dV/dI take on very different values along this
curve.
In my opinion, the IBIS data sheet (sometimes called a model) is
the best way to describe the output characteristics of a driver.
The I vs V information and ramp rate provide enough information
for a simulator to get accurate results. I have compared the
results of IBIS simulation to that of full HSPICE model simulation
and hardware measurements for several different transmission line
problems. IBIS model simulation compares very well to both.
regards,
Larry Smith
Sun Microsystems
> Date: Wed, 01 Apr 1998 10:55:14 -0700
> From: "D. C. Sessions" <dc.sessions@tempe.vlsi.com>
> MIME-Version: 1.0
> To: si-list@silab.eng.sun.com
> Subject: Re: [SI-LIST] : Output Impedance
> Content-Transfer-Encoding: 7bit
>
> alaa alani wrote:
>
> > What is the simplest way to measure the output impedance of a buffer
> > given just a spice netlist?
>
> Since CMOS buffers are generally very nonlinear, the first
> problem is to decide what you mean by "output impedance."
>
> Example: Zout is the value of pull-up resistance to Vddq
> that the driver can pull down to Vddq/2
>
> Example: Zout is the slope of the V/I curve at the supply
> intercept.
>
> Example: Zout is Zline minus the series resistance that
> gives a first-incidence amplitude of Vddq/2.
>
> For all of these the answer is pretty clear as soon as you
> choose the definition; just simulate the output under the
> given conditions and constraints.
>
> --
> D. C. Sessions
> dc.sessions@tempe.vlsi.com