Text preview for : 5990-9250EN S-parameter Series_ Transforming Oscilloscope Acquisitions for De-Embedding_252C Embeddi part of Agilent 5990-9250EN S-parameter Series Transforming Oscilloscope Acquisitions for De-Embedding 252C Embeddi Agilent 5990-9250EN S-parameter Series_ Transforming Oscilloscope Acquisitions for De-Embedding_252C Embedding & Simulating c20140812 [15].pdf
Back to : 5990-9250EN S-parameter S | Home
Keysight Technologies
Transforming Oscilloscope Acquisitions
for De-Embedding, Embedding and
Simulating Channel Effects
Application Note
S-parameter Series
Introduction to Signal Acquisition and Theory
Realizing accu- This paper is the first of 6 papers whose intent is to guide designers and validation engineers of high-
rate and reliable speed digital systems through the details of channel element measurement and modeling for the
analysis of signal purpose of manipulating oscilloscope acquisitions to reveal the signals that the engineer wants to see
data on today's and measure. This extra processing is required because the actual signal cannot be probed (for vari-
high-speed digital ous reasons) or because the probed location is different than is desired, or because the measurement
systems requires a elements themselves (i.e. probes) affect the measurement result and need to be removed from the
thorough under- reported measurement.
standing of the
theories of analysis About ten years ago, the electronics industry departed from the traditional parallel bus interface and
of data acquisi- embraced a serial topology. At first look this seemed a little strange. Not only did the bus clock rate
tion. Armed with have to multiplied by at least a factor of 8 to get the same effective throughput, but it meant that
an understanding digital designers would have to become expert at new blocks such as phase lock loops; tradition-
of the theory, and ally an analog designers area. The issue at hand was that integrated circuit was becoming IO bound;
knowledge of the it wasn't the internal capabilities but getting signals to and from the device. Additionally, the lowly
tools, the designer printed circuit board made of FR-4 had become, in a sense, a precious commodity--in order to keep
can accurately profit margins, the low cost material had to be used and running eight bit parallel buses was a waste
and efficiently of routing real estate. The result of this technological departure was a headlong rush into higher data
achieve precise rates and encumbent microwave principles applied in design. We now call this area of engineering,
models of device Signal Integrity.
behavior. This
knowledge The problem that Signal Integrity practitioners address is the myriad of issues that come into play in
allows the engineer the conveyance of digital data from a digital transmitter through a channel to a receiver. At billions of
to develop bits per second (Gbs), everything in the transmission path affects the signal to some degree. De-
optimum designs velopers and designers must learn what the most significant contributors are and how to model and
the first time, measure them. When these are understood they can be entered into simulators or other measurement
maximizing equipment to assist in the understanding and optimization of system performance. The objective of
efficiency and the an engineer working in the signal integrity space is correlation of simulations to measured results.
optimizing their
value proposition. Engineers working in the high-speed digital area are necessarily concerned about signal integrity is-
sues, but may not be signal integrity experts. The validation and design engineer, for instance, needs
to understand how to solve the problem of specifying a signal characteristic at a remote location, but
not necessarily be worried about the humidity dependence of the loss tangent of FR-4. They also
would like to run some system scenarios using measured or simulated models to see the effects of
varying one elements characteristics or even many elements simultanously. These elements may be
printed circuit transmission lines, vias on the boards, connectors, resistors, inductors , capacitors,
and even the measurement probes themselves. At times, an element might even be another design
whose sole purpose is allow access to measure the transmitted signal at some point. These are called
`fixtures' and though they may be of excellent quality they will have an effect and must be considered.
In the enterprise of evaluating these systems an oscilloscope is ultimately used to view the signals.
Even if the signal location is not accessible or is understood to be affected by system elements at the
time of acquisition, or is derived from added element; the system can be modeled and an accurate
simulation of the real signal can be derived. Keysight Technologies, Inc. has developed a series of tools
that can be effectively used to measure the elements, derive models from simulations, filter measured
data for more accuracy, and ultimately accurately characterize the this signal of interest.
2
Hardware-based Whether measurements are made with probes or test fixtures, the task of validation leads
inexorably to the process of co-simulation. Co-simulation, the combination of simulation
co-simulation and measurement, can show what a simple measurement alone cannot. It can show what
comes of age a measured signal would look like if it were affected by additional circuitry that wasn't
present in the physical measurement. It can also show the opposite - what a measured
signal would have looked like if it were not affected by extraneous circuitry (parasitics)
that are present in the physical measurement path. It can even show what a signal looks
like at a different location than the physical location that was actually measured.
Co-simulation has been used for decades by extracting measured waveforms from oscil-
loscopes and importing them into EDA simulation tools. Although affective, it can be very
cumbersome and time consuming. Modern high-performance oscilloscopes now have the
ability to perform co-simulations directly inside the scope on live measured waveforms.
These co-simulations are simply voltage transformations applied to the measured wave-
forms. A linear filter, which may or may not have gain, is applied to the measured wave-
form that renders a different view