|
What technologies are needed to create and maintain the information superhighway?
Certainly fiber optics will be one of them. Fiber-optic communications systems
have the ability to transmit vast streams of data including voice (telephone
conversations), video (TV channels), and data (computer networks). In fact,
multimedia applications, in which voice, video, and data will be transmitted
and displayed at the same time, will be highly desirable in the near future.
As these services become more commonplace, the required network capacity will
expand and current fiber-optic systems operating at data rates of hundreds
of megabits per second will be superseded by those with rates near ten gigabits
per second. Nationwide networks will give way to one that is much more international
in scope. Currently, a large number of fiber-optic networks, operating at 155
megabits per second and 622 megabits per second, are in operation in the U.S.A.,
Europe, and Japan and new installations of 2.4-gigabit-per-second networks
are near completion.
To support this high-speed, long-distance traffic, new technologies are
rapidly evolving. Optical amplifiers based on rare-earth-doped optical fibers-for
example, erbium-doped fiber amplifiers, or EDFAs-make it possible to overcome
fiber loss and construct transparent undersea optical links thousands of kilometers
long without electronic regenerators. These amplifiers can also be used as
excellent noise sources to test optical components such as photodetectors (page
6). In terrestrial systems,
several signals at different wavelengths can be sent over the same fiber to
increase capacity (wavelength division multiplexing), and their intensity can
be boosted in a single EDFA. Accurately testing the amplifier performance is
critical in these systems. The HP 81600 EDFA test system (page
13) makes calibrated
gain and noise figure measurements as a function of wavelength and optical
power. This test system uses several high-performance lightwave products such
as the HP 8168C tunable laser source and the HP 71450A optical spectrum analyzer
in addition to a sophisticated software algorithm to measure the characteristics
of optical amplifiers. These algorithms were designed and tested by a team
consisting of engineers from California and Germany. To achieve a wide tuning
range, the HP 8168C uses a new semiconductor laser chip that has been developed
at HP Laboratories (Page
20) using a quantum-well
indium gallium arsenide (InGaAs) structure. This device also makes it possible
to obtain sufficiently high optical input power to test amplifier saturation.
Long fiber spans using optical amplifiers with no pulse reshaping compound
subtle signal distortion in fiber. Particularly at high data rates, fiber polarization
effects can increase dispersion problems. Therefore, it is essential to be
able to measure very accurately the state of polarization of optical signals
and the polarization-mode dispersion of optical components. It is also highly
desirable to be able to do these measurements in real time. The HP 8509B lightwave
polarization analyzer (page 27) is used in these
cases. In addition, in high-performance systems, requirements on optical back-reflections
are more severe. Fundamental fiber-optic test equipment like the HP 8156A optical
attenuator (page 34) must be designed
to these more exacting tolerances. To measure low back-reflection levels, the
HP 8504B precision reflectometer (page
39) can be used. A high-performance
light-emitting diode (page 43) makes it possible
for the reflectometer to measure -80-dB back-reflections-as little as 1 picowatt
of reflected power-with a spatial resolution of 50 micrometers. At high data
rates, system timing accuracy becomes increasingly important. The HP 71501B
jitter and eye diagram analyzer (page 49) is capable of making
jitter measurements at rates up to 10 gigabits per second.
Monitoring network integrity becomes more complex in higher-capacity systems.
The HP 81700 remote fiber test system (page
57) helps guard against
breaks in the fiber causing outages. The system can monitor many fibers at
once, even at different sites, while they are carrying live traffic. If a fiber
break occurs, the system generates signals and alarms indicating the location
of the fault.
The development of lightwave test and measurement products requires a solid
photonics technology base. In addition to the quantum-well lasers and LEDs
mentioned above, HP Laboratories is working on other devices and subsystems
for future products. A highly stable, miniature laser using YVO (yttrium orthovanadate)
crystals operating at a center frequency of 282 terahertz (page
63) is an example of
excellent low-noise optical sources. The recent development of surface emitting
lasers (page
67) is very promising
for a variety of applications such as optical interconnect systems for data
communications and generating visible light using complex laser structures
(page
72).
The technologies and products described in this issue show HP's strength and
core competency in the area of photonics and lightwaves. A key element of the
company's strength in lightwave test and measurement is the strong and healthy
coupling between the research teams at HP Laboratories and the manufacturing
divisions: the Lightwave Operation at Santa Rosa, California and the Boblingen
Instruments Division in Germany.
Waguih Ishak, Manager Photonics Technology Department, HP Laboratories
Roger Jungerman, Engineer/Scientist, Lightwave Operation
Also in this Issue
Automatic test systems for digital components, boards, and systems generally
contain a test pattern sequencer, a module that applies the test signals to
the device under test and receives the device's responses. However, the traditional
sequencer architecture has proved inadequate for testing many devices that
operate on serial bit streams. One of the authors of the article on page
76 recalls writing a
test for such a device using a traditional sequencer that required three months
and 13,000 lines of source code and tested only a fraction of the device's
functionality. Realizing that a new sequencer architecture was needed for testing
serial-oriented devices, engineers at HP's Manufacturing Test Division first
developed a generic model of a serial communication system. Based on their
analysis of the model, they then designed a new serial test sequencer architecture
for the HP 3070 family of board test systems. The architecture features modules
called reconfigurable bit processors, which the engineers have dubbed circuitware because they're neither hardware nor software. A new Serial Test Language was
written to simplify test programming. The article on page
76describes all of
these developments and presents several case studies of customer applications
that show dramatic improvements in test development time, test coverage, throughput,
and equipment requirements.
Four papers in this issue are from the 1994 HP Design Technology Conference,
a forum for the exchange of ideas, best practices, and results among HP engineers
involved in the development and application of integrated circuit design technologies.
The theme for the conference was "Accelerating Integration." * To shorten the
time it takes to put high-performance ASICs (application-specific integrated
circuits) into production, one design group developed coding guidelines and
a process for generating wire load models (page
91). The coding guidelines
head off later problems and the wire load models are conservative enough to
make routing easy without sacrificing performance. ** As IC features become
smaller and chips become more densely packed, on-chip connections have an increasing
impact on delay times and therefore on performance. Advanced interconnect modeling
(page 97) is a framework
that allows designers to model, optimize, and scale circuit delays, including
both gates and interconnections.
* While thorough delay testing is necessary
for high-performance designs, not all circuit paths can be tested for delay.
The article on page 105 proposes an algorithm
for synthesizing 100% delay testable circuits. The algorithm uses a method
called cube partitioning. ** The article on page
110 compares two fault
diagnosis methods to determine which does a better job in CMOS circuits. They
conclude that a bridging fault model with a simple diagnosis algorithm is better
than a simple stuck-at-0 or stuck-at-1 fault model with a complex algorithm.
R.P. Dolan
Editor
|
|