Philippe Dondon

Implementation of a feed-forward Artificial Neural Network in VHDL on FPGA

Describing an Artificial Neural Network (ANN) using VHDL allows a further implementation of such a system on FPGA. Indeed, the principal point of using FPGA for ANNs is flexibility that gives it an advantage toward other systems like ASICS which are entirely dedicated to one unique architecture and allowance to parallel programming, which is inherent to ANN calculation system and one of their advantages. Usually FPGAs do not have unlimited logical resources integrated in a single package and this limitation forcesrequirement for optimizations for the design in order to have the best efficiency in terms of speed and resource consumption. This paper deals with the VHDL designing problems which can be encountered when trying to describe and implement such ANNs on FPGAs.

Optimization of a BiCMOS integrated transducer for self-compensated capacitive pressure sensor

Two new pressure sensor demonstrators have been designed and mounted using silicon/Pyrex capacitive sensing cells and analog-digital BiCMOS transducers. A ratiometric scheme has been used to self-compensate thermal drifts and nonlinearities. The first demonstrator implemented with four chips is characterized by a relative sensitivity close to -1.85%/bar, a nonlinearity of the order of 1.1%FS and an offset thermal coefficient smaller than 20 ppm//spl deg/C. The second one has been implemented with two integrated chips. It consists, on the one hand, of two pairs of converters and counters integrated in the same chip, and on the other hand, of a reference and measurement capacitors into the same sensing cell. This configuration permits one to reduce stray capacitors by a factor of two but also provokes interference between oscillators. Characterization and PSPICE modeling have permitted understanding of localized causes and then proposes some solutions to avoid and/or minimize these problems.

An original approach for the design of a Class D power switching amplifier-an audio application

The Class D amplifier is not well known in audio applications. Its excellent power ratio (greater than 90%) is the most important advantage. The MOS transistors switching power stage is able to drive a useful power up to 100 W to the loudspeaker. However, designing such an amplifier is more difficult than designing a Classical Class A or AB power amplifier. The distortion level is also slightly higher than in AB structures. A method to design a Class D power switching amplifier as easily as possible is presented here. The experimental results are given to illustrate the design method.

Thermal transient characterization of electronic assemblies by infrared thermography and Delta Vbemeter methods

The severe operating environment of electronic systems, means that the design of electronic boards requires continual investigations, involving not only electronics but also thermal and thermomechanical studies. We present here two complementary experimental methods to characterize the thermal transient behaviour of electronic assemblies and we compare the results obtained. Then, we conclude by presenting the fields of applications for such test equipment.

Performances and potentialities of a very simple self-compensated pressure sensor demonstrator

A pressure sensor demonstrator has been designed and mounted using a simple analog-digital BiCMOS converter and a capacitive silicon-Pyrex sensing cell. The measurements as a function of pressure and temperature have enabled the authors to evaluate the efficiency of a ratiometric scheme to self-compensate thermal drifts and nonlinearities. The best demonstrator is characterized by a thermal coefficient of offset smaller than 20 ppm/K and a nonlinearity lower than or equal to /spl plusmn/1.2% FS. The essential features of the sensor have been modeled by a simple analytical expression. The model specifies the potential gains in pressure sensitivity and nonlinearity. The best design is based on a quasisymmetrical architecture in which stray capacitors are minimized. This study demonstrates the feasibility of self-compensated capacitive sensors characterized by an overall accuracy of the order of a few percent of the measurement scale over a temperature range of 100 K at very low cost.