2-bit adder: carry and sum logic circuits at 19 GHz clock frequencyin InAlAs/InGaAs HBT technology

Carry and sum circuits for a 2-bit adder used in a pipelined 2N-bit adder-accumulator architecture are reported. To obtain high clock rates in a design with multiple gate delays a novel merged AND-OR-Latch structure using four-level series gated current steering logic is employed. These integrated circuits were fabricated in InAlAs/InGaAs transferred substrate heterojunction bipolar transistor technology and operate up to 19 GHz clock rate     

A 200-MFLOPS 100-MHz 64-b BiCMOS vector-pipelined processor (VPP)ULSI

The first single-chip 64-b vector-pipelined processor (VPP) ULSI is described. It executes vector operations indispensable to high-speed scientific computation. The VPP ULSI attains a 200-MFLOPS peak performance at a 100-MHz clock frequency. This extremely high performance is made possible by the integration on the VPP of a 64-b five-stage pipelined adder/shifter, a 64-b five-stage pipelined multiplier/divider/logic operation unit, and a 40-kb register file. Various new high-speed circuit techniques have been also developed for 100-MHz operations. The chip, which was fabricated with a 0.8-μm BiCMOS and triple-layer metallization process technology, has a 17.2-mm×17.3-mm area and contains about 693 K transistors. It consumes 13.2 W at a 100-MHz clock frequency with a single 5-V power supply     

Design techniques for high-throughput BiCMOS self-timed SRAMs

Design techniques for a high-throughput BiCMOS self-timed SRAM are described. A new BiCMOS read circuit using a pipelined read architecture and a BiCMOS complementary clocked driver (BCCD) are proposed to reduce the operating cycle time. A 8192×9-b dual-port self-timed SRAM designed using the proposed techniques achieved a clock cycle time of 3.0 ns, that is, a 333-MHz operating frequency, by SPICE simulation on model parameters for 0.8-μm BiCMOS technology. A high-speed built-in self-test (BIST) circuit was studied and designed for the 3.0-ns cycle SRAM. It is confirmed that the BIST circuit allows the 3.0-ns cycle SRAM to test at its maximum operating frequency     

A 1.5 V 10-b 30-MS/s CMOS pipelined analog-to-digital converter

A 1.5 V 10-b 30MS/s CMOS pipelined analog-to-digital converter (ADC) is described. Low-voltage techniques are proposed for pipelined analog-to-digital converter that avoids the use of low-threshold voltage process, on-chip clock voltage doubler, bootstrapped switch, or switched-opamp technique. At the front-end, a low-voltage S/H circuit with cross-coupled input sampling switch is employed to eliminate the input signal feedthrough and enhance the dynamic performance of the pipelined ADC. Multiplying digital-to-analog converter (MDAC) with cross-coupled configuration also provides an effective common-mode feedback to overcome the problem of common-mode accumulation. The prototype chips have been fabricated and experimental results confirm the feasibility of this new technique.     

A 500 MHz, 32 bit, 0.4 μm CMOS RISC processor

A 500 MHz, 32 bit RISC microprocessor has been experimentally developed using an 8-stage pipelined architecture and high-speed circuits, including a 500 MHz 1 kilobyte double-stage pipelined cache, a 1.8 ns register file, a double-stage binary look-ahead carry (BLC) adder circuit, and a 500 MHz phase locked loop (PLL) frequency multiplier. Newly developed circuit-integrating techniques include a stacked power-line structure, which serves as a noise shield and also provides low bounce, a low voltage-swing interface circuit with on-chip adjustable termination resistors, a small-skew clock distribution method, and a clock synchronization circuit which provides small-skew clock among LSI chips. About 200000 transistors are integrated into a 7.90 mm×8.84 mm die area with 0.4 μm CMOS fabrication technology. Power dissipation is 6 W at a 500 MHz operation and 3.3 V supply voltage     

A dual 10-b 200-MSPS pipelined D/A converter with DLL-based clock synthesizer

A dual 10-b/200-MSPS pipelined digital-to-analog converter (DAC) suitable for communication applications is here presented. Prior implementation limitations have been overcome through circuit techniques. A prototype has been designed using a 4-metal-levels 3.3-V 0.5-/spl mu/m BiCMOS technology and operates on a 3-phase clock synthesized by an on-chip delay-locked loop (DLL). The DAC shows 9.7 effective bits and 70 dB of spurious free dynamic range for a synthesized sine wave of 2 V/sub pp/ at 34 MHz and output rate of 200 MSPS. Altogether, the two DACs, their reference, and the DLL occupy an active area of 2.28 mm/sup 2/ and consume 693 mW at full speed.     

A 700-MHz 24-b pipelined accumulator in 1.2-μm CMOS forapplication as a numerically controlled oscillator

To accomplish timing recovery/synthesis in high-speed communication systems, a 24-b numerically controlled oscillator (NCO) IC using a circuit design technique called true single-phase clock (TSPC) pipelined CMOS has been fabricated in a standard 1.2-μm CMOS process. The device achieves a maximum tested input clock rate of 700 MHz, which results in an output frequency tuning range from DC up to 350 MHz with a 41.7-Hz tuning resolution and a peak-to-peak phase jitter of 1.4 ns. The 1.7-mm×1.7-mm IC dissipates 850 mW with a single 5-V supply, which is substantially lower than similar ECL and GaAs devices     

A 12-b 600 ks/s digitally self-calibrated pipelined algorithmic ADC

This paper discusses fully digital error correction and self-calibration which correct errors due to capacitor mismatch, charge injection, and comparator offsets in algorithmic A/D converters. The calibration is performed without any additional analog circuitry, and the conversion does not need extra clock cycles. This technique can be applied to algorithmic converter configurations including pipelined, cyclic, or pipelined cyclic configurations. To demonstrate the concept, an experimental 2-stage pipelined cyclic A/D converter is implemented in a standard 1.6-μm CMOS process. The ADC operates at 600 ks/s using 45 mW of power at ±2.5 V supplies. The active die area excluding the external logic circuit is 1 mm². Maximum DNL of ±0.6 LSB and INL of ±1 LSB at a 12-b resolution have been achieved     

All-N-logic high-speed true-single-phase dynamic CMOS logic

In this paper, a novel all-N-logic single-phase high speed dynamic CMOS logic is introduced and analyzed. The circuits achieve high speed by eliminating the need for the low-speed P-logic blocks. The use of all-N-logic allows the speed of the proposed circuits to be two to three times the speed of conventional CMOS dynamic circuits. An 2:1 frequency divider, using proposed ANL2 circuits, is simulated using 0.8 μm CMOS technology with the operating clock frequency reaching as high as 1.5 GHz. A pipelined 8-b carry generator of five-stacked NMOS transistors, which operates at a clock rate of over 710 MHz, has also been simulated. Experimental results show that the proposed circuits operate over 910 MHz implemented in a 1.2 μm CMOS technology     

Low-voltage dynamic BiCMOS CLA circuit with carry skip using novelfull-swing logic

This paper presents novel low-voltage dynamic BiCMOS logic gates and an improved carry look-ahead (CLA) circuit with carry skip using these new dynamic BiCMOS topologies. The well-known “MOS clock feedthrough effect” is used to achieve full swing with substantially reduced low-to-high evaluation delay in the logic gates, thus, resulting in reduced carry propagation/bypass delay in the cascaded CLA array. Simulations at clocking frequency of 100 MHz, using 2-μm BiCMOS process parameters and supply voltage in the range of 2-4 V displays lower gate delay and lower power dissipation compared to other recent dynamic BiCMOS logic topologies. The circuit has no dc power dissipation, race, or charge redistribution problems. An 8-b CLA with 5-b carry skip was achieved in 2.917 ns. This speed is significantly higher than other recent dynamic BiCMOS CLA designs. In addition, the new CLA circuit is more compact compared to previous dynamic BiCMOS CLA designs. A tiny chip was fabricated using the MOSIS Orbit Analog 2-μm V-well CMOS process for the experimental verification of the new low-voltage dynamic BiCMOS topologies     

A BiCMOS dynamic carry lookahead adder circuit for VLSIimplementation of high-speed arithmetic unit

A BiCMOS dynamic carry lookahead circuit that is free from race problems is presented. A 16 b full-adder test circuit, which has been designed based on a 2 μm BiCMOS technology, shows a more than five times improvement in speed as compared to the CMOS Manchester carry lookahead (MCLA) circuit. The speed advantage of the BiCMOS dynamic carry lookahead circuit is even greater in a 32- or 64-b adder     

A pipelined multiplier-accumulator using a high-speed, low-powerstatic and dynamic full adder design

This paper proposes a new pipelined full-adder circuit structure for the implementation of pipelined arithmetic modules. With both static and dynamic structures, it has the advantages of high operational speed, smallest transistor count, and the low power/speed ratio. The adder cell is then used to design a pipelined 8×8-b multiplier-accumulator (MAC). In this MAC, a special pipelined structure is designed to reduce the latency. The MAC is fabricated in a 0.8-μm single-poly-double-metal CMOS process. The post-layout simulation shows that the pipelined 1-b full adder can work up to 350 MHz with a 3 V power supply. The whole MAC chip that contains 4200 transistors is measured to operate a 125 MHz using 3.3 V power supply     

An 8-b 85-MS/s parallel pipeline A/D converter in 1-μm CMOS

A new architecture consisting of a time-interleaved array of pipelined analog-to-digital converters (ADCs) is presented. A prototype has been designed consisting of four switched-capacitor (S/C) multistage pipelined ADCs in parallel. Hardware cost is minimized by sharing resistor strings, bias circuitry and clock generation circuitry over the array. Digital error correction is employed to ease comparator accuracy requirements. Techniques are employed to minimize the effect of mismatches across the array. A key circuit issue is the design of a high-speed sample-and-hold (S/H) amplifier: a fully differential, mostly NMOS, non-folded-cascode operational-amplifier topology is used. An experimental chip was implemented in 1-μm CMOS and 8-b resolution at a sample rate of 85 megasamples per second (MS/s) was obtained. Signal-to-noise plus distortion (S/(N+D)) was 41 dB for an input sinusoid of 40 MHz     

An efficient charge recovery logic circuit

Efficient charge recovery logic (ECRL) is proposed as a candidate for low-energy adiabatic logic circuit. Power comparison with other logic circuits is performed on an inverter chain and a carry lookahead adder (CLA). ECRL CLA is designed as a pipelined structure for obtaining the same throughput as a conventional static CMOS CLA. Proposed logic shows four to six times power reduction with a practical loading and operation frequency range. An inductor-based supply clock generation circuit is proposed. Circuits are designed using 1.0-μm CMOS technology with a reduced threshold voltage of 0.2 V     

一种40 MSPS 10位流水折叠分级式A/D转换器

探讨和研究基于流水线(Pipelined)技术的折叠分级式A/D转换器(ADC),理论分析了它的原理和一般结构,给出了一个具体结构的ADC框图和具体的折叠电路,并得出了实际制作的ADC的测试图。该折叠分级式ADC的输入频率可达到1 MHz,2级折叠电路产生的高2位加上子ADC产生的8位,使A/D转换器可达到10位的分辨率,采样率最大为40 MSPS。     

High-performance and power-efficient CMOS comparators

Several design techniques for high-performance and power-efficient CMOS comparators are proposed. First, the comparator is based on the priority-encoding (PE) algorithm, and the dynamic circuit technique developed specifically for the priority encoder can be applied. Second, the PE function and the subsequent logic functions are merged and efficiently realized in the multiple output domino logic (MODL) to result in a shortened logic depth. The circuit in MODL CMOS is also compact and power efficient because few transistors are needed. Third, the multilevel look-ahead technique is used to shorten the path of priority-token propagation. Finally, the circuit is realized with a latch-based two-stage pipelined structure, and the comparison function is partitioned into two parts, with each part executed in each half of the clock cycle in a delay-balanced manner. Post-layout simulation results show that a 64-b comparator designed with the proposed techniques in a 3-V 0.6-/spl mu/m CMOS technology is 16% faster, 50% smaller, and 79% more power efficient as compared with the all-n-transistor comparator, which is the fastest among the conventional comparators. Measurement results of the test chip conform with simulation results and prove the feasibility of the proposed techniques.     

A cascaded sigma-delta pipeline A/D converter with 1.25 MHz signalbandwidth and 89 dB SNR

A low-noise multibit sigma-delta analog-to-digital converter (ADC) architecture suitable for operation at low oversampling ratios is presented. The ADC architecture uses an efficient high-resolution pipelined quantizer while avoiding loop stability degradation caused by pipeline latency. A 16-b implementation of the architecture, fabricated in a 0.6-μm CMOS process, cascades a second-order 5-b sigma-delta modulator with a four-stage 12-b pipelined ADC and operates at a low 8X oversampling ratio. Static and dynamic linearity of the integrated ADC are improved through the use of dynamic element matching techniques and the use of bootstrapped and clock-boosted input switches. The ADC operates at a 20 MHz clock rate and dissipates 550 mW with a 5 V/3 V analog/digital supply. It achieves an SNR of 89 dB over a 1.25-MHz signal bandwidth and a total harmonic distortion (THD) of -98 dB with a 100-kHz input signal     

Implementation of a low power 16-bit radix-4 pipelined SRT divider using a modified Split-Path Data Driven Dynamic Logic (SPDL) structure

In this paper a 16-bit radix-4 pipelined divider implemented in a modified version of SPD³L family structure (SPCD³L: Split-Path Clock-Data driven Dynamic Logic) is presented. Through the modification, the clock signal is also used to pre-charge some critical parts of the circuit. Performance of the circuit is evaluated at different simulation corners. The results show that, compared with Domino structure, the proposed circuit has lower power consumption and higher speed. Latency of the divider is equal to 10 half clock cycles. The design is simulated using HSPICE in a 1.8-V TSMC_180 nm CMOS process.     

A 200-MHz 16-bit super high-speed signal processor (SSSP) LSI

A 200-MHz 16-b BiCMOS super high-speed signal processing (SSSP) circuit has been developed for high-speed digital signal processor (DSP) LSIs. In order to produce extremely fast LSI circuits, several novel techniques have been combined for integration of the SSSP. They include a redundant binary convolver architecture, a double-stage pipelined convolver architecture, and submicrometer BiCMOS drivers with large capacitive load drivability. The SSSP performs 200-MHz addition. The chip, which was fabricated with 0.8-μm BiCMOS and triple-layer metallization technology, has an area of 5.87 mm×5.74 mm and contains 20150 transistors. It operates at a clock frequency of 200 MHz with a single 5-V power supply and typically consumes 800 mW     

A 30-b integrated logarithmic number system processor

The authors describe an integrated processor that performs addition and subtraction of 30-b numbers in the logarithmic number system (LNS). This processor offers 5-MOPS performance in 3-μm CMOS technology, and is implemented in a two-chip set comprising 170 K transistors. Two techniques are used to achieve this precision in a moderate circuit area. Linear approximation of the LNS arithmetic functions using logarithmic arithmetic is shown to be simple due to the particular functions involved. A segmented approach to linear approximation minimizes the amount of table space required. Subsequent nonlinear compression of each lookup table leads to a further reduction in table size. The result is that a factor of 285 reduction in table size is achieved, compared to previous techniques. The circuit area of the implementation is minimized by optimizing the table parameters, using a computer program that accurately models ROM area. The implementation is highly pipelined, and produces one result per clock cycle using a ten-stage pipeline