Paul Jespers

Charge pumping in MOS devices

Gate pulses applied to MOS transistors were found to stimulate a net flow of charge into the substrate. Investigation of this effect revealed a charge-pumping phenomeonon in MOS gate-controlled-diode structures. A first-order theory is given, whereby the injected charge is separated into two components. One component involves coupling via fast surface states at the Si-SiO 2 interface under the gate, while the other involves recombination of free inversion-layer charge into the substrate.

A CMOS 13-b cyclic RSD A/D converter

A 13-b CMOS cyclic A/D converter that does not need trimming nor digital calibration is presented. The effects associated with the error on the gain factor 2 as well as the offset errors are corrected by taking full advantage of the redundant signed digit (RSD) principle. The gain error resulting from mismatches among switched capacitors is corrected by a novel strategy that implements an exact multiplication by four after two cycles. As a result, offset errors do not affect the integral or the differential linearities from the RSD algorithm. The remaining overall shift caused by offsets is reduced under the LSB level by a proper choice of capacitor switching sequence. The converter achieves 1/2 LSB integral and differential linearity at 25 kS/s; harmonic distortion is less than -83 dB. Chip area is 2.9 mm2 in a standard CMOS 3-mu-m technology, including control logic and the serial-to-parallel output shift register. Power consumption is 45 mW under +/-5-V supplies.

A new carry-free division algorithm and its application to a single-chip 1024-b RSA processor

A carry-free division algorithm is described. It is based on the properties of redundant signed digit (RSD) arithmetic to avoid carry propagation and uses the minimum hardware per bit, i.e. one full adder. Its application to a 1024-b RSA (Rivest, Shamir, and Adelman) cryptographic chip is presented. The features of this new algorithm allowed high performance (8 kb/s for 1024-b words) to be obtained for relatively small area and power consumption (80 mm/sup 2/ in a 2- mu m CMOS process and 500 mW at 25 MHz). >

Design of SOI CMOS operational amplifiers for applications up to 300/spl deg/C

Design guidelines using two analog parameters (Early voltage and transconductance to drain current ratio) are proposed for correct operation of silicon-on-insulator (SOI) CMOS operational amplifiers (opamp) at elevated temperature up to 300/spl deg/C. The dependence of these parameters on temperature is first described. A new single-stage CMOS opamp model using only these two parameters is presented and compared to measurements of several implementations operating up to 300/spl deg/C for applications such as micropower (below 4 /spl mu/W at 1.2 V supply voltage), high gain (65 dB) or high frequency up to 100 MHz. Trade-offs among such factors as gain, bandwidth, phase margin, signal swing, noise, matching, slew rate and power consumption are described. The extension to other architectures is suggested and the design methodology is valid for bulk as well as SOI CMOS opamps.

Analog implementation of a Kohonen map with on-chip learning

Kohonen maps are self-organizing neural networks that classify and quantify n-dimensional data into a one- or two-dimensional array of neurons. Most applications of Kohonen maps use simulations on conventional computers, eventually coupled to hardware accelerators or dedicated neural computers. The small number of different operations involved in the combined learning and classification process, however, makes the Kohonen model particularly suited to a dedicated VLSI implementation, taking full advantage of the parallelism and speed that can be obtained on the chip. A fully analog implementation of a one-dimensional Kohonen map, with on-chip learning and refreshment of on-chip analog synaptic weights, is proposed. The small number of transistors in each cell allows a high degree of parallelism in the operations, which greatly improves the computation speed compared to other implementations. The storage of analog synaptic weights, based on the principle of current copiers, is emphasized. It is shown that this technique can be used successfully for the realization of VLSI Kohonen maps.

Fully-Depleted SOI CMOS Technology for Low-Voltage Low-Power Mixed Digital/Analog/Microwave Circuits

This paper demonstrates that fully-depleted (FD) silicon-on-insulator (SOI) technology offers unique opportunities in the field of low-voltage, low-power CMOS circuits. Beside the well-known reduction of parasitic capacitances due to dielectric isolation, FD SOI MOSFETs indeed exhibit near-ideal body factor, subthreshold slope and current drive. These assets are both theoretically and experimentally investigated. Original circuit studies then show how a basic FD SOI CMOS process allows for the mixed fabrication and operation under low supply voltage of analog, digital and microwave components with properties significantly superior to those obtained on bulk CMOS. Experimental circuit realizations support the analysis.

Modelling and application of fully depleted SOI MOSFETs for low voltage, low power analogue CMOS circuits

Transistor models which reproduce the superior device characteristics of fully depleted silicon-on-insulator (SOI) MOSFETs and which are efficient for the design of analogue CMOS circuits are discussed and validated. These analogue models are then used to investigate the significant performance improvement that several basic analogue cells can achieve when optimized in fully depleted SOI CMOS, rather than in bulk CMOS technology. Experimental verifications support this original demonstration of the great potential of fully depleted SOI CMOS for low voltage, low power analogue applications.

Improved synthesis of gain-boosted regulated-cascode CMOS stages using symbolic analysis and gm/ID methodology

A systematic study of the gain-boosted regulated-cascode operational transconductance amplifier (OTA) CMOS stage is presented. Symbolic analysis is used first to describe the pole-zero behaviour and second to propose design criteria for optimal settling time. A synthesis procedure based on the “gm/ID” methodology is considered further on for quick optimization of the architecture based on the dc open-loop gain, transition frequency, and settling time specifications. Practical design cases are finally discussed.

A 10-bit pipelined switched-current A/D converter

A modified RSD algorithm has been implemented in a switched-current pipelined A/D converter. The offset insensitivity of the RSD Converter reduces the effect of several nonidealities proper to current copier cells. Moreover, the benefits resulting from the large tolerances inherent to the RSD algorithm and the pipelined architecture result in an improved conversion rate. Measurements on a first prototype give an integral nonlinearity error less than 0.8 LSB for 10-bit accuracy. Power dissipation is 20 mW and silicon area is 2.5 mm/sup 2/. The measured sampling rate is 550 kS/s. It is an improvement by a factor of twenty compared to known equivalent CMOS switched-current converters. It is nevertheless still well below the predicted conversion rate of 4.5 MHz, which should be obtained once this A/D converter is integrated into an analog front-end. Full compatibility with standard digital technologies makes this kind of converter attractive for low power, medium-fast converters with 10-bit accuracy. >

Neural networks for high-storage content-addressable memory: VLSI circuit and learning algorithm

An implementation of a VLSI fully interconnected neural network with only two binary memory points per synapse is described. The small area of single synaptic cells allows implementation of neural networks with hundreds of neurons. Classical learning algorithms like the Hebbs rule show a poor storage capacity, especially in VLSI neural networks where the range of the synapse weights is limited by the number of memory points contained in each connection; an algorithm for programming a Hopfield neural network as a high-storage content-addressable memory is proposed. The storage capacity obtained with this algorithm is very promising for pattern-recognition applications. >