High-speed lightwave component analysis to 20 GHz

A new family of instruments-analyzer, test set, sources, receivers, and modulator-characterizes electrical, electrooptical, and optical components of fiber optic communications systems at modulation rates to 20 GHz.

FIBER OPTIC COMMUNICATIONS systems that have emerged over the last decade compete very well with more traditional communications systems as a cost-effective means for information exchange. These systems typically operate at hundreds of megabits per second, and pilot systems being installed work at gigabit-per-second rates. Research laboratories throughout the world are developing devices capable of operating at modulation bandwidths of many tens of gigahertz, and are developing building blocks that would bring reality and practicability to the 10-gigabit-per-second fiber optic communication system.

However, as the technology advances, system speeds exceed many gigabits per second, and the trend toward lowering the effective cost per bit and mile continues, the component designer needs to maximize the performance of each system block and minimize or eliminate adverse interactions among system blocks to optimize overall system performance. New techniques and tools are needed to help in the design, manufacture, and support of these components and systems. A knowledge of the latest concepts in characterizing such components, and of the modern measuring techniques used to quantify a device's performance, simplifies the tasks of design, evaluation, and characterization for the engineer.

The typical communication system (Fig. 1) can be partitioned into several subsystems such as the transmitter, the receiver, and the propagation medium. Individual network components include amplifiers, filters, modulators, and demodulators. All of these need to meet certain performance levels for the overall system to meet its objectives. It is customary to characterize a digital system's overall performance in terms of bit error rate (BER). However, although a bit error rate of 1 x 10-'9 (one error per billion bits transmitted) is a meaningful system specification, it is rather difficult to derive from it a modulator or-filter-specification or characteristic to which the component can be designed. Furthermore, if the system does not meet the given BER specification, such a measurement of error rate provides very little diagnostic information for isolating a faulty assembly or component.

Derivatives of the BER measurements such as eye patterns are more powerful when it comes to detecting signal-to-noise degradations, but these cannot replace the fundamental component measurements like gain, bandwidth, or frequency response. When the bit rates are so high that most of the baseband signal energy is in the microwave frequency range, then reflections and multipath transmission cause major additional performance degradations, which have to be specified and characterized.1

Lightwave Component Analysis

In the ultrahigh-speed systems being designed in research laboratories, the modulated signals contain spectral components reaching well into the microwave region. In digital systems, where the information source is an encoded bit stream of high and low states, the information-carrying signal is usually a pulse sequence. The required bandwidth to transmit this sequence is derived from the pulse width of the signal, or conversely, the system's bandwidth defines the pulse shape, rise time, and overshoot of the applied time-domain signal.

With ultrafast gigabit systems, most of the signal energy is no longer propagated by electrons in conductors but rather in electromagnetic fields along transmission lines. Signal scattering caused either by mismatches (reflections) or by multipath transmission significantly degrades system performance.

The component designer needs to have tools to measure and characterize each high-speed device in terms of its modulation frequency response (bandwidth), its microwave impedance, its optical return loss, its propagation time or equivalent electrical length, and its microwave and optical interactions with other high-speed devices. The making of these measurements on electrical, electrooptical, and optical devices is called lightwave component analysis. Typical devices are shown in Table I.

The lightwave component analyzer measurement concept is shown in Fig. 2. The lightwave component analyzer system measures a modulation transfer function of a device under test, providing the amplitude and phase characteristics. The information source is a sine wave signal incident to the transmitter. The sine wave modulation signal is placed on the light carrier to create an amplitude modulated light signal, which is transmitted through an optical device such as an optical fiber. The receiver demodulates the modulated signal after it has been operated upon by the device under test. The signal processing unit forms the complex ratio of the device's input and output (i.e., the modulation transfer function amplitude and phase) and applies the appropriate corrections for the device being measured.

20-GHz Lightwave Measurement Needs

In years past, selected lightwave laboratories have developed their own measurement systems, which typically consisted of a directly modulable laser diode (bandwidths 10 to 15 GHz), a photodiode detector, and a microwave network analyzer. Typically, these systems had limited dynamic range because of small signal levels either into the laser or from the photodiode, and they required extensive knowledge of the measurement block diagram and extraordinary care when connecting and measuring devices. This alternative of building one's own system did not exist except for a small number of laboratories, for there weren't then nor are there now any commercially available laser diodes with 15-GHz to 20-GHz modulation bandwidths.

To develop 20-GHz lightwave component analysis capability, our investments were focused on the optical modulator and on optical isolator technologies. One of the primary challenges in the 20-GHz lightwave program was to develop a lightwave source capable of 20-GHz modulation rates. More information on the optical modulator and isolator technology developments can be found in the articles on pages 41 and 45, respectively.

The objectives of the 20-GHz lightwave program were to provide lightwave component analysis measurements to customers who may and may not already own HP microwave network analyzers, and to provide lightwave technology to customers in the form of instruments and accessories. Fig. 3 shows the products that have resulted from the 20-GHz lightwave program. They are:

* HP 8703A lightwave component analyzer

* HP 83420A lightwave test set

* HP 83421A lightwave source

* HP 83422A lightwave modulator

* HP 83423A lightwave receiver

* HP 83424A lightwave CW source

* HP 83425A lightwave CW source.

Analyzer Design Objectives

The HP 8703A is the integrated system that combines the 20-GHz microwave and lightwave technologies to provide measurement solutions for high-speed component design and characterization. The key objectives were to provide 20-GHz lightwave component analysis measurement capabilities similar to earlier products,2 provide accurate, repeatable, and calibrated measurements, to provide the greatest possible measurement dynamic range with the technology used, and to provide a measurement system that was easy to use from both calibration and measurement standpoints. This system is designed for customers who do not already have HP microwave network analyzers.

The integrated system configuration allows more freedom to optimize system dynamic range by designing microwave components that drive the lightwave components at their preferred levels and optimize the lightwave and microwave interfaces. Fig. 4 shows the HP 8703A system, which consists of three sections. The top section performs the signal processing and display functions and the user interface function. The middle and lower sections perform the microwave and lightwave test set functions, respectively. They condition and route the appropriate signals through the appropriate RF switches to provide accurate, repeatable, and fast measurements.

The lightwave test set consists of a modulatable optical source, a 3-dB directional coupler, and an optical receiver. The six RF switches in the middle and bottom sections allow a much simplified calibration and measurement process that minimizes the effects of RF cable nonrepeatablity and RF mismatches. A more detailed discussion of the HP 8703A system design issues, trade-offs, and solutions appears in the article on page 13.

Test Set Design Objectives

For those who already have an HP microwave network analyzer but want to add basic 20-GHz lightwave measurement capability, the HP 83420A lightwave test set is the most cost-effective approach. The design objectives for this project were similar to those of the HP 8703A, except that the test set would work with existing HP 8510B, HP 8720B, and HP 8757C network analyzers.

Fig. 5 shows a lightwave component analyzer system configured with the HP 83420A lightwave test set, the HP 8510B network analyzer, and an HP 9000 Series 300 computer acting as system controller. Because the block diagrams of the network analyzers were already defined and fixed, the test set definition had its challenges and constraints. The most severe constraint, fixed network analyzer interfaces, precluded incorporating an RF switch matrix into the measurement system. Consequently, users must effect a few more cable connections and disconnections than are necessary to perform the same device measurements with the HP 8703.

The optical measurement dynamic range of the system of Fig. 5 typically approaches that of the HP 8703A within a few dB. The system provides all basic lightwave device measurements, and the external controller provides the same guided calibration and measurement setup instructions as the integrated system. More detailed information on the lightwave test set and its development are given in the article on page 23.

Measurements

Table II lists the basic transmission and reflection measurements most useful for the design, evaluation, and characterization of high-frequency components." Fig. 6 shows the HP 8703A connections for the four types of devices. A few of the more unusual measurements and their applications will be covered in this article. 1550-nm Chromatic Dispersion Fiber Measurement

Chromatic dispersion in standard single-mode optical fiber crosses through the zero dispersion point in the 1300nm wavelength band and reaches values of approximately 16 ps/km-nm at 1550 nm. Fig. 7a illustrates the effects of chromatic dispersion on a signal. Consider an intensity modulated signal with upper and lower sidebands spaced f.sub.1 from the carrier. As the signal propagates along the fiber, each component travels at a slightly different velocity, resulting in time delays from one component to another. For a given length of fiber, there exists a set of modulation frequencies at which the sidebands are phase shifted to the extent that the AM component vanishes and nulls occur in the baseband AM response of the fiber. A closed-form expression predicts the frequencies at which this will occur and allows the fiber dispersion value to be calculated."

The baseband AM response of a 17.7-km single-mode fiber is shown in Fig. 7b. The fiber was measured in the optical transmission configuration shown in Fig. 6. The source was a 1550-nm DFB (distributed feedback) laser. The measurement was done in step sweep mode to provide the most accurate signal detection despite the long transmission delay. It was performed twice, at different values of HP 8703A optical modulator bias voltage, to eliminate the effect of residual phase modulation. The null frequencies of the two measurements are averaged. The dispersion determined in this manner is 16.2 ps/km-nm.

Table II

Lightwave Component Analysis Measurements' Transmission

Insertion Loss/Gain Frequency Response

Modulation Bandwidth

Flatness

Slope Responsivity (a) Time-Domain Analysis

Rise Time

Pulse Dispersion Delay Length Insertion Modulation Phase Group Delay (as applied to the information) [E.sub.r] or n (index of refraction) Reflection Sensitivity E/O devices) Reflection Return Loss impedance (electrical/RF) Reflectometry

Electrical Time-Domain Reflectometry

Optical Frequency-Domain Reflectometry (OFDR) Delay Length (a) Conversion efficiency of electrical/optical or optical/electrical devices only

Apply to electrical/electrical (E/E), electrical/optical (E/O), optical/electrical O/E), and optical/optical (O/O) devices unless specified otherwise.

Laser FM Response Measurement

Frequency-shift keying (FSK) systems shift the frequency of an optical carrier in response to a modulating signal. Since DFB lasers produce carrier frequency shifts of 100 MHz to 2 GHz per milliampere of bias current, they are often chosen as sources in FSK systems. Very high FM rates are achieved by the mechanism of carrier modulation. Thus, there is interest in measuring FM laser response as a function of bias current.

Coherent detection FSK systems typically use a local oscillator laser at the receiving end. Alternatively, an optical frequency discriminator can be used to convert the incoming FM signal to an AM signal, allowing the use of conventional direct detection schemes, which are well understood. The discriminator may be a Mach-Zehnder or Michelson interferometer or a ring resonator. The concerns in such systems are the FM response of the DFB laser and the bandwidth of the receiver electronics.

FM response tests require some form of optical frequency discriminator. As an example, assume that a Mach-Zehnder interferometer is used as the discriminating element. When operated at quadrarure, the point halfway between full output and minimum output, the interferometer produces an output optical intensity that varies linearly with changes of carrier frequency. Using an FM discriminator of known sensitivity with a lightwave component analyzer (HP 8703A or HP 83420A + network analyzer), the DFB laser's AM and FM frequency responses can be measured.4

Fig. 8 shows measurements of the AM and FM responses of a 1300-nm DFB laser. The two responses are constructed by adding and subtracting measurements performed at adjacent quadrature points of the discriminator, that is, at discriminator operating slopes of opposite sign but identical magnitude. The relative levels of AM and FM depend upon the differential path length of the discriminator. This measurement can be calibrated with the use of an external modulated optical power reference. Note that the measurement reveals a null in the FM response at 3.7 GHz. Receiver Optical Launch Measurement

One common application for optical reflection measurements is the design and characterization of optical launches onto a photodiode receiver. Of interest to the designer would be not only the overall return loss of the launch including the optical connector and launch optics, but also the location and return loss value of each individual reflection.

By using the lightwave component analyzer technique and its time-domain capabilities, selective reflection measurements-the separation of optical reflections in the distance domain-are easily made on a wide variety of optical sensors, lightwave components, subsystems, and systems. An optical mismatch can be measured by separating the incident test signal from the reflected one by means of a directional coupler. The ratio of reflected to incident signals yields the reflection coefficient magnitude at the optical frequency and can be measured by an optical power meter. However, using the lightwave component analyzer technique and applying a modulated test signal has the advantage that the origin of the reflected signal can be determined by computing the propagation delay and distance from the envelope (modulated) phase shift of the reflected signal. Furthermore, it is possible to apply the same technique to mutliple reflections, thus separating them in the distance or time domain. This method of measuring optical reflections is far superior to the power meter approach, which yields only an average reflection coefficient with no indication of the number of reflections or their amplitudes or positions.1 Fig. 9 shows a selective reflection measurement of a photodiode receiver optical launch using a GRIN (graded-index) lens. Marker 2 shows a return loss of 35 dB created by the interface of a fiber to the first surface of the GRIN lens with index matching compound. Marker 1 shows the combination of the reflection from the second GRIN lens surf ace, which is antireflection-coated, and the photodiode surface (25 dB return loss). The two responses are spaced about 14 mm apart, which corresponds to the length of the GRIN lens used.

The lightwave component analyzer system has a maximum modulation frequency range of 20 GHz. The inverse of the maximum frequency range, in this case 50 ps, determines the theoretical limit for resolving two closely spaced and equal reflections. In other words, the system can resolve two closely spaced reflections whose roundtrip signal travel time is equal to or greater than 50 ps, which corresponds to two reflections spaced approximately 5 mm apart in optical fiber. This limit is often referred to as the two-event resolution or response resolution of the system.

Temperature Sensor Measurement

In-line Fabry-Perot sensors, currently in development, can be cascaded and are immune to electromagnetic interference, making them ideal candidates for aircraft, ships, and industrial plants. They are built by coating a cleaved fiber with a low-reflectance dielectric mirror, then reattaching the fiber with a fusion splice. The process is repeated some small distance down the fiber to create the short Fabry-Perot interferometer. Light incident on the sensor is partially reflected by both mirrors. Since the sensor is shorter than the coherence length of the light source in the measurement, the two reflections will interfere strongly as the sensor expands and contracts with temperature. Such sensors must be individually calibrated. These sensors can be interrogated using either time-domain or frequency-domain reflectometry techniques.

The frequency-domain reflectometry measurement uses a 1550-nm DFB laser whose linewidth is about 30 MHz. The light from the laser source is coherent over the length of the Fabry-Perot sensor. Measurements are taken at intervals of 5 degrees C, and the data is transformed to the time domain, as shown in Fig. 10. Since the two Fabry-Perot surfaces are separated by less than the distance resolution of the instrument, the combined reflections appear as a single reflection having an amplitude that depends upon temperature.

System Performance

Dynamic range is a common figure of merit for test instrumentation. Typical system performance is dependent upon many factors such as the HP 8703A measurement calibration routine selected (e.g., response, response/isolation, etc.), the signal drive levels used, and the signal processing features selected (e.g., IF bandwidth, signal averaging, signal smoothing). To convey the typical dynamic range performance of the HP 8703A system, three devices were selected: a laser diode E/O device), a photodiode (O/E device), and an optical attenuator (0/0 device), as shown in Figs. 11, 12, and 13, respectively.

Figs. 11 and 12 show the typical slope responsivity or conversion efficiency measurement dynamic ranges for laser/transmitter and photodiode/receiver devices. The dynamic ranges are typically 85 and 105 electrical dB for transmitter and receiver devices, respectively.

Fig. 13 shows the transmission measurement dynamic range for optical devices. An optical step attenuator provided various values of attenuation. The noise floor of the system shown varies from approximately 48 to 44 optical dB over frequency. The response/isolation measurement calibration routine was selected with an IF bandwidth of 10 Hz and an averaging factor of 16 for the 40-dB and 50-dB attenuator steps.

Acknowledgments

Contributions to the development of the light-wave and microwave technologies, hardware components, and system calibration and design that went into the 20-GHz lightwave component analyzer, test set, and accessories came from many people located in many Hewlett-Packard entities. We would like to thank Waguih Ishak, Steve Newton, Bill Chang, Dave Dolfi, and their teams at HP Laboratories and we certainly appreciate their many contributions in lightwave technology and support. Over the course of the project, the codevelopment efforts, cooperation, and support we enjoyed with Bob Bray, Mark Zurakowski, Roger jungerman, and their teams at the Microwave Technology Division were greatly appreciated and much welcomed. It was through the joint efforts of both divisions' teams that the optical modulator was developed from a device of notable scientific curiousity to one capable of meeting manufacturing standards of quality, performance, and robustness. Finally, we would especially like to thank the members of our Network Measurements Division lightwave team for their dedication and teamwork in meeting the challenges and rigors of the aggressively scheduled 20-GHz lightwave program. Additionally, we greatly appreciate the help and support we received from our support groups and other R&D personnel during the development and new product introduction phases of the program. And to Hugo Vifian, a special word of thanks and appreciation for his unwavering support, confidence, and commitment during this development program.

References

1. H. Vifian and R. Wong, "Characterizing High-Speed Lightwave Components," Fiberoptic Product News, Vol. 4, no. 10, October 1989, pp. 22-28 and 39, and Vol. 4, no. 11, November 1989, pp. 24-26.

2. R. Wong, P. Hernday, M. Hart, and G. Conrad, "High-Speed Lightwave Component Analysis," Hewlett-Packard journal, Vol. 40, no. 3, June 1989, pp. 35-51.

3. R. Wong, P. Hernday, and M. Hart, "Technical Innovations for the Design of a New 20-GHz Lightwave Component Analyzer," Hewlett-Packard RF and Microwave and Lightwave Measurement Symposium, October 1989.

4. P. Hernday, R. Wong, and D. Harkins, "Advanced Applications of 20-GHz Lightwave Component Analysis," Hewlett-Packard RF

COPYRIGHT 1991 Hewlett Packard Company
COPYRIGHT 2004 Gale Group