A 10-Gb/s receiver with series equalizer and on-chip ISI monitor in 0.11-/spl mu/m CMOS

A 10-Gb/s receiver is presented that consists of an equalizer, an intersymbol interference (ISI) monitor, and a clock and data recovery (CDR) unit. The equalizer uses the Cherry-Hooper topology to achieve high-bandwidth with small area and low power consumption, without using on-chip inductors. The ISI monitor measures the channel response including the wire and the equalizer on the fly by calculating the correlation between the error in the input signal and the past decision data. A switched capacitor correlator enables a compact and low power implementation of the ISI monitor. The receiver test chip was fabricated by using a standard 0.11-/spl mu/m CMOS technology. The receiver active area is 0.8 mm/sup 2/ and it consumes 133 mW with a 1.2-V power supply. The equalizer compensates for high-frequency losses ranging from 0 dB to 20 dB with a bit error rate of less than 10/sup -12/. The areas and power consumptions are 47 /spl mu/m /spl times/ 85 /spl mu/m and 13.2 mW for the equalizer, and 145 /spl mu/m /spl times/ 80 /spl mu/m and 10 mW for the ISI monitor.     

A CMOS 0.25-/spl mu/m continuous-time FIR filter with 125 ps per tap delay as a fractionally spaced receiver equalizer for 1-gb/s data transmission

This paper presents a CMOS 0.25-/spl mu/m continuous-time 6-tap FIR filter that is used as a fractionally spaced receiver equalizer for 1-Gb/s data transmission. Each tap of the FIR filter delay line is realized with a second-order low-pass filter. Simulations show that the tap delay can be tuned from 100 ps to 300 ps while keeping a constant group delay within the bandwidth of 2.1 GHz and 800 MHz correspondingly. Experimental results show that the FIR filter can successfully recover a 1-Gb/s differential digital signal that has been transmitted over a 220-inch PCB trace which causes -31.48-dB attenuation at the symbol rate frequency of 1 GHz. The measured bit error rate after equalization is less than 10/sup -12/ over a 750-ps sampling range, compared to a 10/sup -2/ bit-error rate before equalization. Also presented are the measurement results comparing the horizontal and the vertical openings of the signals before and after equalization for PCB traces with different length. The chip dissipates 45 mW from a 2.5-V supply and occupies 0.33/spl times/0.27 mm/sup 2/ in a 0.25-/spl mu/m CMOS process.     

Parallel-optical interconnects >100 gb/s

A parallel-optical interconnect with 12 channels operating at 8.5 Gb/s giving an aggregate data rate of 102 Gb/s is demonstrated, to the authors' knowledge, for the first time. The paper describes and demonstrates 13 /spl times/ 16-mm cross-section 12-channel parallel-optic transmitter and receiver modules with each channel operating at a data rate of 8.5-10 Gb/s. This was achieved using bottom-emitting 990-nm vertical-cavity surface-emitting lasers and bottom-illuminated InGaAs-InP photodetectors flip-chip bonded directly to 12-channel transmitter and receiver integrated circuits, respectively. In addition, 102-Gb/s link results are demonstrated over 100 m of 50-/spl mu/m-core standard multimode ribbon fiber. A bit-error ratio of <10/sup -13/ was measured on a single channel after transmission through 100 m of multimode fiber at a data rate of 8.5 Gb/s with all 12 channels operating simultaneously.     

Realization of multigigabit channel equalization and crosstalk cancellation integrated circuits

In this paper, we present integrated circuit solutions that enable high-speed data transmission over legacy systems such as short reach optics and electrical backplanes. These circuits compensate for the most critical signal impairments, intersymbol interference and crosstalk. The finite impulse response (FIR) filter is the cornerstone of our architecture, and in this study we present 5- and 10-Gsym/s FIR filters in 2-/spl mu/m GaAs HBTs and 0.18-/spl mu/m CMOS, respectively. The GaAs FIR filter is used in conjunction with spectrally efficient four-level pulse-amplitude modulation to demonstrate 10-Gb/s data throughput over 150 m of 500 MHz/spl middot/km multimode fiber. The same filter is also used to demonstrate equalization and crosstalk cancellation at 5 Gb/s on legacy backplane. The crosstalk canceller improves the bit error rate by five orders of magnitude. Furthermore, our CMOS FIR filter is tested and demonstrates backplane channel equalization at 10 Gb/s. Finally, building blocks for crosstalk cancellation at 10 Gb/s are implemented in a 0.18-/spl mu/m CMOS process. These circuits will enable 10-Gb/s data rates on legacy systems.     

Burst-mode clock and data recovery in optical multiaccess networks using broad-band PLLs

Broad-band phase-locked loops (PLLs) are proposed for burst-mode clock and data recovery in optical multiaccess networks. Design parameters for a charge-pump PLL-based clock and data recovery (CDR) with fast phase acquisition are derived using a time-domain model that does not assume narrow loop bandwidth or small phase errors. Implementation in a half-rate CDR circuit confirms a clock phase acquisition time of 40 ns, or 100 bits at 2.488-Gb/s rate, and data recovery at 1.244-Gb/s rate with a bit-error rate of 1/spl times/10/sup -10/ (2/sup 14/-1 pseudorandom binary sequence with Manchester-encoding). The CDR was fabricated in complementary metal-oxide-semiconductor 0.18-/spl mu/m technology in an area of 1/spl times/1 mm/sup 2/ and consumes 54 mW of power from a 1.8-V supply.     

A four-channel 3.125-Gb/s/ch CMOS serial-link transceiver with a mixed-mode adaptive equalizer

This work presents a quad-channel serial-link transceiver providing a maximum full duplex raw data rate of 12.5Gb/s for a single 10-Gbit eXtended Attachment Unit Interface (XAUI) in a standard 0.18-/spl mu/m CMOS technology. To achieve low bit-error rate (BER) and high-speed operation, a mixed-mode least-mean-square (LMS) adaptive equalizer and a low-jitter delay-immune clock data recovery (CDR) circuit are used. The transceiver achieves BER lower than <4.5/spl times/10/sup -15/ while its transmitted data and recovered clock have a low jitter of 46 and 64 ps in peak-to-peak, respectively. The chip consumes 178 mW per each channel at 3.125-Gb/s/ch full duplex (TX/RX simultaneous) data rate from 1.8-V power supply.     

RF transmission over multimode fibers using VCSELs-comparing standard and high-bandwidth multimode fibers

Fiber-optic radio-frequency links have been assembled using oxide-confined vertical-cavity surface-emitting lasers (VCSELs) and multimode fibers. Links with single and multimode VCSELs and with standard and high-bandwidth fibers have been evaluated and compared in the frequency range of 0.1-10 GHz. The best results were obtained for links with a multimode VCSEL and a high-bandwidth fiber. For a 500-m-long link, a spurious free dynamic range of 104 dB/spl middot/Hz/sup 2/3/ at 2 GHz and 100 dB/spl middot/Hz/sup 2/3/ at 5 GHz were obtained while allowing for a VCSEL-fiber misalignment of /spl plusmn/12 /spl mu/m. Corresponding numbers for the intrinsic link gain and noise figure are -29 and -33 dB, and 39 and 42 dB at 2 and 5 GHz, respectively. Inferior performance was observed for the standard fiber link due to a larger variation in modal group velocities. This paper also presents a detailed link analysis to identify performance limitations and to suggest modifications for improved performance.     

Uncooled high-temperature (130/spl deg/C) operation of InGaAs-GaAs multiple quantum-well lasers at 20 Gb/s

Short-haul fiber-optic communication systems require high-speed semiconductor lasers that can operate uncooled over a wide temperature range. In this letter, we describe high-speed short-cavity InGaAs-GaAs multiple-quantum-well lasers operating at 1.1-/spl mu/m wavelength. The Fabry-Perot lasers were fabricated in a triple-mesa geometry suitable for on-wafer probing. With 3/spl times/200 /spl mu/m/sup 2/ ridge-waveguide lasers, which showed the best compromise between high-temperature and high-speed performance, a 3-dB modulation bandwidth of 14.5 GHz at 130/spl deg/C was achieved. Uncooled 20-Gb/s operation of these lasers is presented over a wide-temperature range from 25/spl deg/C to 130/spl deg/C without automatic power control.     

A 20-Gb/s 0.13-/spl mu/m CMOS serial link transmitter using an LC-PLL to directly drive the output multiplexer

A 20-Gb/s transmitter is implemented in 0.13-/spl mu/m CMOS technology. An on-die 10-GHz LC oscillator phase-locked loop (PLL) creates two sinusoidal 10-GHz complementary clock phases as well as eight 2.5-GHz interleaved feedback divider clock phases. After a 2/sup 20/-1 pseudorandom bit sequence generator (PRBS) creates eight 2.5-Gb/s data streams, the eight 2.5-GHz interleaved clocks 4:1 multiplex the eight 2.5-Gb/s data streams to two 10-Gb/s data streams. 10-GHz analog sample-and-hold circuits retime the two 10-Gb/s data streams to be in phase with the 10-GHz complementary clocks. Two-tap equalization of the 10-Gb/s data streams compensate for bandwidth rolloff of the 10-Gb/s data outputs at the 10-GHz analog latches. A final 20-Gb/s 2:1 output multiplexer, clocked by the complementary 10-GHz clock phases, creates 20-Gb/s data from the two retimed 10-Gb/s data streams. The LC-VCO is integrated with the output multiplexer and analog latches, resonating the load and eliminating the need for clock buffers, reducing power supply induced jitter and static phase mismatch. Power, active die area, and jitter (rms/pk-pk) are 165 mW, 650 /spl mu/m/spl times/350 /spl mu/m, and 2.37 ps/15 ps, respectively.     

A 10-Gb/s CMOS clock and data recovery circuit with a half-rate binary phase/frequency detector

A 10-Gb/s phase-locked clock and data recovery circuit incorporates a multiphase LC oscillator and a half-rate phase/frequency detector with automatic data retiming. Fabricated in 0.18-/spl mu/m CMOS technology in an area of 1.75/spl times/1.55 mm/sup 2/, the circuit exhibits a capture range of 1.43 GHz, an rms jitter of 0.8 ps, a peak-to-peak jitter of 9.9 ps, and a bit error rate of 10/sup -9/ with a pseudorandom bit sequence of 2/sup 23/-1. The power dissipation excluding the output buffers is 91 mW from a 1.8-V supply.     

40-Gb/s WDM transmission with virtually imaged phased array (VIPA) variable dispersion compensators

We have demonstrated variable dispersion compensation by using a virtually imaged phased array (VIPA) to overcome the small dispersion tolerance in 40-Gb/s dense wavelength-division multiplexing (WDM) transmission systems. By utilizing the periodical characteristics of VIPA compensators, we performed simultaneous dispersion compensation in a 1.28-Tb/s (40-Gb/s/spl times/32 ch; C band) short-haul transmission and confirmed that only two VIPA compensators and one fixed dispersion-compensating fiber are required for a large transmission range of 80 km. This performance can greatly reduce the cost, size, and number of compensator menus in a 40-Gb/s WDM short-haul transmission system. In addition, we achieved 3.5-Tb/s (43-Gb/s/spl times/88 ch; C and L bands) transmission over a 600-km nonzero dispersion-shifted fiber by using VIPA compensators. Although channel-by-channel dispersion compensation is required due to the larger residual dispersion slope in long-haul transmission, the periodical characteristics of the VIPA compensators offer the advantage of considerably reducing the number of different modules required to cover the whole C (or L) band. An adequate optical signal-to-noise ratio, which was the same for all channels, was-obtained by using distributed Raman amplification, a gain equalizer, and a preemphasis technique. We achieved a Q-factor of more than 11.8 dB; (BER<10/sup -17/ with forward-error correction) for all 88 channels.     

A 3.125-Gb/s clock and data recovery circuit for the 10-Gbase-LX4 Ethernet

A 3.125-Gb/s clock and data recovery (CDR) circuit using a half-rate digital quadricorrelator frequency detector and a shifted-averaging voltage-controlled oscillator is presented for 10-Gbase-LX4 Ethernet. It can achieve low-jitter operation and improve pull-in range without a reference clock. This CDR circuit has been fabricated in a standard 0.18-/spl mu/m CMOS technology. It occupies an active area of 0.6 /spl times/ 0.8 mm/sup 2/ and consumes 83 mW from a single 1.8-V supply. The measured bit-error rate is less than 10/sup -12/ for 2/sup 7/ - 1 PRBS 3.125-Gb/s data. It can meet the jitter tolerance specifications for the 10-Gbase-LX4 Ethernet application.     

A 40-Gb/s CMOS clocked comparator with bandwidth modulation technique

A differential comparator that can sample 40-Gb/s signals and that operates off a single 1.2-V supply was designed and fabricated in 0.11-/spl mu/m standard CMOS technology. It consists of a front-end sampler, a regenerative stage, and a clocked amplifier to provide a small aperture time and a high toggle rate. The clocked amplifier employs a bandwidth modulation technique that switches the feedback gain to reduce the reset time while keeping the effective gain high. We confirmed that the comparator receives a 40-Gb/s data stream at a toggle rate of 10 GHz with bit error rate less than 10/sup -12/ by laboratory measurements.     

A 10-Gb/s CDR/DEMUX with LC delay line VCO in 0.18-/spl mu/m CMOS

A monolithic 10-Gb/s clock/data recovery and 1:2 demultiplexer are implemented in 0.18-/spl mu/m CMOS. The quadrature LC delay line oscillator has a tuning range of 125 MHz and a 60-MHz/V sensitivity to power supply pulling. The circuit meets SONET OC-192 jitter specifications with a measured jitter of 8 ps p-p when performing error-free recovery of PRBS 2/sup 31/-1 data. Clock and data recovery (CDR) is achieved at 10 Gb/s, demonstrating the feasibility of a half-rate early/late PD (with tri-state) based CDR on 0.18-/spl mu/m CMOS. The 1.9/spl times/1.5 mm/sup 2/ IC (not including output buffers) consumes 285 mW from a 1.8-V supply.     

Compact wavelength-selective switch for gigabit filtering in access networks

A compact [200/spl times/200 /spl mu/m/sup 2/] wavelength-selective switch based on thermally tunable SiO/sub 2/--Si/sub 3/N/sub 4/ microring resonators has been designed and realized. The switch supports gigabit filtering applications in access networks. Spectral measurements show an ON-OFF ratio of 12 dB and a channel separation of 20 dB. The 10-Gb/s measurements on a single ring show no degradation of the modulated signal and a theoretical BER (bit-error rate) <10/sup -12/.     

Interdigitated Ge p-i-n photodetectors fabricated on a Si substrate using graded SiGe buffer layers

We report an interdigitated p-i-n photodetector fabricated on a 1-/spl mu/m-thick Ge epitaxial layer grown on a Si substrate using a 10-/spl mu/m-thick graded SiGe buffer layer. A growth rate of 45 /spl Aring//s/spl sim/60 /spl Aring//s was achieved using low-energy plasma enhanced chemical vapor deposition. The Ge epitaxial layer had a threading dislocation density of 10/sup 5/ cm/sup -2/ and a rms surface roughness of 3.28 nm. The 3-dB bandwidth and the external quantum efficiency were measured on a photodetector having 1-/spl mu/m finger width and 2-/spl mu/m spacing with a 25/spl times/28 /spl mu/m/sup 2/ active area. At a wavelength of 1.3 /spl mu/m, the bandwidth was 2.2, 3.5, and 3.8 GHz at bias voltages of -1, -3, and -5 V, respectively. The dark current was 3.2 and 5.0 /spl mu/A at -3 and -5 V, respectively. This photodetector exhibited an external quantum efficiency of 49% at a wavelength of 1.3 /spl mu/m.     

A 1.8-V 10-Gb/s fully integrated CMOS optical receiver analog front-end

A fully integrated 10-Gb/s optical receiver analog front-end (AFE) design that includes a transimpedance amplifier (TIA) and a limiting amplifier (LA) is demonstrated to require less chip area and is suitable for both low-cost and low-voltage applications. The AFE is fabricated using a 0.18-/spl mu/m CMOS technology. The tiny photo current received by the receiver AFE is amplified to a differential voltage swing of 400 mV/sub (pp)/. In order to avoid off-chip noise interference, the TIA and LA are dc-coupled on the chip instead of ac-coupled though a large external capacitor. The receiver front-end provides a conversion gain of up to 87 dB/spl Omega/ and -3dB bandwidth of 7.6 GHz. The measured sensitivity of the optical receiver is -12dBm at a bit-error rate of 10/sup -12/ with a 2/sup 31/-1 pseudorandom test pattern. Three-dimensional symmetric transformers are utilized in the AFE design for bandwidth enhancement. Operating under a 1.8-V supply, the power dissipation is 210 mW, and the chip size is 1028 /spl mu/m/spl times/1796 /spl mu/m.     

A 1.2-V 1.5-Gb/s 72-Mb DDR3 SRAM

A 1.2-V 72-Mb double data rate 3 (DDR3) SRAM achieves a data rate of 1.5 Gb/s using dynamic self-resetting circuits. Single-ended main data lines halve the data line precharging power dissipation and the number of data lines. Clocks phase shifted by 0/spl deg/, 90/spl deg/, and 270/spl deg/ are generated through the proposed clock adjustment circuits. The latter circuits make input data sampled with an optimized setup/hold window. On-chip input termination with a linearity error of /spl plusmn/4.1% is developed to improve signal integrity at higher data rates. A 1.2-V 1.5-Gb/s 72-Mb DDR3 SRAM is fabricated in a 0.10-/spl mu/m CMOS process with five metals. The cell size and the chip size are 0.845 /spl mu/m/sup 2/ and 151.1 mm/sup 2/, respectively.     

Reduced surface sidewall recombination and diffusion in quantum-dot lasers

We examine the surface recombination rate in quantum-dot semiconductor lasers and determine the diffusion length (1.0 /spl mu/m) and, for the first time, provide a value for surface recombination velocity (5/spl times/10/sup 4/ cm/s) in quantum-dot material. As a result of strong carrier confinement in the dots, these values are much lower than in comparable quantum-well lasers (5/spl times/10/sup 5/ cm/s and 5 /spl mu/m, respectively) allowing the creation of narrow (2-3 /spl mu/m wide) lasers with comparable threshold currents to those of broad area devices.     

Minimal group refractive index dispersion and gain evolution in ultra-broad-band quantum cascade lasers

The group refractive index dispersion in ultra-broad-band quantum cascade (QC) lasers has been determined using Fabry-Perot spectra obtained by operating the lasers in continuous wave mode below threshold. In the wavelength range of 5-8 /spl mu/m, the global change of the group refractive index is as small as +8.2 /spl times/ 10/sup -3/ /spl mu/m/sup -1/. Using the method of Hakki and Paoli (1975), the subthreshold gain of the lasers has furthermore been measured as a function of wavelength and current. At the wavelength of best performance, 7.4 /spl mu/m, a modal gain coefficient of 16 cm/spl middot/kA/sup -1/ at threshold and a waveguide loss of 18 cm/sup -1/ have been estimated. The gain evolution confirms an earlier assumption that cross-absorption restricted laser action to above 6 /spl mu/m wavelength.