R.C. Williamson

Optically sampled analog-to-digital converters

Optically sampled analog-to-digital converters (ADCs) combine optical sampling with electronic quantization to enhance the performance of electronic ADCs. In this paper, we review the prior and current work in this field, and then describe our efforts to develop and extend the bandwidth of a linearized sampling technique referred to as phase-encoded optical sampling. The technique uses a dual-output electrooptic sampling transducer to achieve both high linearity and 60-dB suppression of laser amplitude noise. The bandwidth of the technique is extended by optically distributing the post-sampling pulses to an array of time-interleaved electronic quantizers. We report on the performance of a 505-MS/s (megasample per second) optically sampled ADC that includes high-extinction LiNbO/sub 3/ 1-to-8 optical time-division demultiplexers. Initial characterization of the 505-MS/s system reveals a maximum signal-to-noise ratio of 51 dB (8.2 bits) and a spur-free dynamic range of 61 dB. The performance of the present system is limited by electronic quantizer noise, photodiode saturation, and preliminary calibration procedures. None of these fundamentally limit this sampling approach, which should enable multigigahertz converters with 12-b resolution. A signal-to-noise analysis of the phase-encoded sampling technique shows good agreement with measured data from the 505-MS/s system.

The Use of Surface-Elastic-Wave Reflection Gratings in Large Time-Bandwidth Pulse-Compression Filters

A new type of surface-wave device has been developed which uses the reflection of surface elastic waves to achieve a desired transfer function. A series of experiments on the reflection of surface waves at normal and oblique incidence from periodic arrays of grooves and overlayer stripes provided guidelines for the choice of the type of reflector, the reflection angle, and the depth of grooves. A prototype pulse-compression filter with a time-bandwidth product of 1500 (T=30 /spl mu/S, /spl Delta/f=50 MHz) has been developed. The grooves were etched into LiNbO/sub 3/ by a neutralized argon ion beam in a manner which provided precise depth control and a desired amplitude response. This reflective-array compressor (RAC) has proved to be relatively free of spurious signals and second-order effects and, as a result, large capacities have been obtained. In the prototype device, rms phase errors were 3.5 deg or less and, as a result, the compressed-pulse sidelobe structure was near ideal. A compression ratio of 1500 was demonstrated. The same device, when operated over a wider bandwidth, yielded a compression ratio of about 4000 with only a modest sacrifice in the level of the time sidelobes.

Wide-band electrooptic guided-wave analog-to-digital converters

Progress in the development of high-speed electrooptic A/D converters is reviewed. A/D converters of this type have been operated at 1 gigasample/second (GS/s) in 2- and 4-bit structures for 500-MHz analog bandwidth. The converter consists of an array of LiNbO3guided-wave interferometric modulators that function as an analog amplitude analyzer, pulsed lasers for optical sampling, and high-speed monolithic comparators/demultiplexers to generate the digital levels and slow the data to ECL-compatible rates. The operational principles of the converter are summarized and a performance analysis presented. The analysis indicates that with currently attainable components, conversion in the 4- to 6-bit range at rates from 1 to 3 GS/s is feasible. Experimental results for several converters are summarized, including a description of beatfrequency tests for analog signals with frequency content up to 500 MHz that indicate the analog bandwidth capabilities of this device. The electrooptical technology is compared to competing high-speed A/D technologies in Si, GaAs, and superconducting materials and the relative merits analyzed. It is found that the electrooptic approach eliminates some of the fundamental and severe problems of conventional converters (e.g., sampling pickup and large numbers of comparators). Finally, application of this converter to wide-band signal-processing problems is described. It is noted that there are numerous applications where a moderate number of bits at a high (gigahertz) sampling rate is attractive.

Properties and applications of reflective-array devices

The use of reflective arrays in surface-wave devices has yielded new devices such as resonators in addition to an enlarged parameter range for devices such as radar pulse expanders and compressors. A number of different grating geometries, substrates, and types of reflectors have been employed in resonators, band-pass filters, filter banks, oscillators, and dispersive delay lines. Simple, yet accurate, models have been developed for design and analysis of reflection gratings in these applications. The major application of reflective arrays has been in radar pulse compressors wherein low spurious levels and precise phase and amplitude response have been achieved in devices with time-bandwidth products as large as 104. Other signal-processing applications such as chirp-z transforms can exploit the large time-bandwidth products available in reflective-array devices.

Picosecond InP optoelectronic switches

Proton bombardment is used to increase the response speed of InP optoelectronic switches. Photoconductivity measurements indicate response times following bombardment of <100 ps, with the electron mobility estimated to be ⩾600 cm2/Vs. This mobility is over an order of magnitude larger than that observed in similar high‐resistivity devices of comparable speed fabricated in germanium or silicon‐on‐sapphire.

Photorefractive effects in LiNbO3 channel waveguides: Model and experimental verification

Photorefractive effects in Ti‐indiffused LiNbO3 channel waveguides at λ=0.85 μm are modeled and the theoretical predictions are experimentally verified. This model allows, for the first time, the quantitative prediction of the magnitudes and time evolution of photorefractive effects in LiNbO3 channel waveguide devices and provides a consistent method of characterizing the photorefractive behavior of LiNbO3 channel waveguides.

Effects of crosstalk in demultiplexers for photonic analog-to-digital converters

Time interleaving of samples digitized by a parallel array of analog-to-digital (A/D) converters provides a means of increasing the sampling rate beyond that possible with a single A/D converter. For time-interleaved photonic A/D converters, optical demultiplexers can be used to advantage. Both time-division and wavelength-division demultiplexers must yield low crosstalk between the parallel output channels in order to yield accurate A/D conversion. An analysis predicts the level and form of the resulting errors. The analytical results compare well with experiment.

Large‐numerical‐aperture microlens fabrication by one‐step etching and mass‐transport smoothing

Precision f/1 microlenses have been fabricated in GaP by smoothing a multiple‐mesa structure etched with a designed width and length variation. High‐resolution lithography and ion‐beam‐ assisted etching were used for mesa definition and resulted in accurate lens profiles after mass‐transport smoothing at 900–1070 °C. This much simplified fabrication technique is highly promising for efficient, diffraction‐limited micro‐optical elements.

Measurement of mode-locked laser timing jitter by use of phase-encoded optical sampling

The phase-noise characteristics of a harmonically mode-locked fiber laser are investigated with a new measurement technique called phase-encoded optical sampling. A polarization-maintaining ring laser is mode locked by use of the short-pulse electrical output of a resonant-tunneling diode oscillator, enabling it to produce 30-ps pulses at a 208-MHz repetition rate. The interferometric phase-encoded sampling technique provides 60-dB suppression of amplitude-jitter noise and allows supermode phase noise to be observed and quantified. The white-noise pulse-to-pulse timing jitter and the rms supermode timing jitter of the laser are measured to be less than 50 and 70 fs, respectively.

Experimental Exploration of the Limits of Achievable Q of Grooved Surface-Wave Resonators

Abstract : An important basic question that remains to be answered for the surface-wave resonator is that of maximum achievable Q. Intrinsic material loss imposes a limit of Q approximately equal to 100,000 at 150 MHz on LiNbO3 and quartz. However, devices produced to date have been limited by grating reflection loss to Q approximately equal to 20,000. This paper compares two alternative approaches to the problem of minimizing the grating reflection loss and hence maximizing the Q of the grooved Fabry-Perot resonator. In one approach, the number of elements (grooves) in the reflecting arrays is modest, and increasing array reflectivity (and hence Q) is obtained by increasing the element reflectivity (deepening the grooves). Experimental measurements show that Q can be increased in this way until a groove depth of roughly 0.025 Lambda (or a step reflectivity of about 1%) is attained on a 300-groove YZ LiNbO3 resonator, at which point excessive losses due to bulk-wave scattering preclude the existence of a resonance. In the second approach, the element reflectivity (groove depth) is kept at a minimum and increasing reflectivity and Q are obtained by increasing the number of elements in the array. In either approach, element reflectivity is increased by a closer control of groove width-to-period ratio to values of 0.5 or less. Measurements of Q for various devices on LiNbO3 and ST quartz are compared with theory. In addition, the maximum Q values measured will be compared with the theoretical maximum Qs set by intrinsic material losses. (Author)