F. Mailly

Modeling the influence of etching defects on the sensitivity of MEMS convective accelerometers

In this paper, a behavioral model that includes the influence of etching defects on the sensitivity of MEMS convective accelerometers is presented. Starting from an existing behavioral model, new physically-based expressions have been derived to introduce etching defects in the simulation of thermal conduction in the sensor. In addition, a semi-empirical model has been introduced for thermal convection. Finally, a very good agreement is obtained between the behavioral model and FEM simulations.

Investigations on electrical-only test setup for MEMS convective accelerometer

In this paper, we investigate potential solutions for the development of an electrical-only test procedure for MEMS convective accelerometers. The objective is to define an alternative low-cost test procedure applicable at wafer-level. Simple electrical test measurements are analyzed and a behavioral model allowing fault injection is developed. Simulation results show that most of the parametric faults that affect the device specifications can be detected with electrical measurements. Only faults that affect the convective behavior escape this alternative test. A preliminary study of convective effects using FEM simulation is then conducted to identify the key parameters that should be included in the model for the subsequent definition of adequate electrical test parameters.

An electrical test method for MEMS convective accelerometers: Development and evaluation

In this paper, an alternative test method for MEMS convective accelerometers is presented. It is first demonstrated that device sensitivity can be determined without the use of physical test stimuli by simple electrical measurements. Using a previously developed behavioral model that allows efficient Monte-Carlo simulations, we have established a good correlation between electrical test parameters and device sensitivity. Proposed test method is finally evaluated for different strategies that privilege yield, fault coverage or test efficiency.

Rejection of power supply noise in wheatstone bridges: Application to piezoresistive MEMS

This paper deals with the design of MEMS using piezoresistivity as transduction principle. It is demonstrated that when the sensor topology doesnpsilat allow a perfect matching of strain gauges, the resolution is limited by the ability of the conditioning circuit (typically a Wheatstone bridge) to reject power supply noise. As this ability is strongly reduced when an offset voltage is present at the output of the bridge, the proposed architecture implements a feedback loop to control MOS transistors inserted in the Wheatstone bridge to compensate resistor mismatches. This feedback exhibits a very good offset cancellation and therefore a better resolution is achieved.

A CMOS-MEMS Inertial Measurement Unit

This paper introduces the first monolithic integration of a 3-axis accelerometer, a 2-axis magnetometer and a full-custom front-end electronics. This complete Inertial Measurement Unit (IMU) has been designed on a single die of 7.5mm2 in a 0.35°m CMOS technology. The five electromechanical sensors are batch-manufactured using a single-step wet etching of the CMOS die. Compared to existing Inertial Measurement Units, assembly costs and common issues associated with sensor misalignments are suppressed. Sensors make use of resistive transduction, either mechanically or thermally induced. The on-chip electronic has been designed to cancel parasitic effects such as offset and to optimize the signal-to-noise ratio on each measurement axis. The achieved performance (8-bits on Earth magnetic field measurement, 8-bits and 6-bits on x/y and z gravity acceleration respectively) makes the device suitable for numerous applications in consumer electronics.

A wide-bandwidth, wide dynamic-range thermal ΣΔ architecture for convective accelerometers

This work describes the implementation and the modeling of a convective accelerometer with 1st-order thermal ΣΔ readout. This closed-loop system is obtained by the hybrid combination of a CMOS sensor prototype which includes an amplifier, and discrete electronic devices. A high comparator precision is reached by dynamic offset cancellation of the amplifier. After a modeling of the Johnson noise and of the quantization-noise, the parameters of the modulator are optimized to get a good linearity, an improved gain precision and a wide dynamic range.

A micro-power high-resolution ΣΔ CMOS temperature sensor

This paper introduces a simple and compact CMOS temperature sensor which sensing principle relies on the measurement of integrated polysilicon resistances. The architecture makes use of two resistive layers of opposite temperature coefficients, both being available in the CMOS process. In order to tackle power consumption issues, usually related to resistive transduction and Wheatstone bridge conditioners, resistors are here placed in an original stage featuring gain even at very low-biasing current (2µA in this case). This analog front-end is used into a 1st order ΣΔ modulator, providing a digital output (i.e. a bitstream) with little additional silicon surface. The paper describes the design of the circuit and provides both simulation and experimental results. Experimental data are obtained from silicon prototypes, over a - 40°C to 100°C temperature range. High resolution (below 0.1°C) is observed.

A CMOS Multi-sensor System for 3D Orientation Determination

This paper presents a unique combination of magnetic and inertial MEMS sensors on a fully monolithic CMOS chip. Pitch and roll are measured using two identical heat transfer based tilt sensors while yaw is derived from an original electromechanical compass. For complete 3-D positioning, the use of an additional z-axis piezoresistive accelerometer is also investigated. Contrary to systems based on gyroscopes, the proposed device outputs the absolute values of the three Eulerpsilas angles. All sensors (i.e. the magnetometers, the thermal tilt sensors, and the z-axis accelerometer) are fabricated simultaneously using a cheap one-step auto-aligned post process (front side bulk micromachining), thus addressing consumer markets for very low-cost applications. Prototypes have been designed and characterized for each sensor. The measured resolution is about 1.7deg for pitch and roll with a time response of 30 ms and 2deg for yaw with a time response of only 10 ms. The targeted overall surface is around 10 mm while power consumption is approximately 50 mW when operated continuously. Such sensors association has numerous identified applications in the field of mobile devices: compass with tilt compensation; pointing devices, new functionalities in mobile phones to mention few.

An approach to integrate MEMS into high-level system design flows

MEMS integration into standard design flows is still an open topic after several years of research and investigations. Facing the diversity of CAD tools, modeling languages and associated simulators, the choice of a particular design flow seems to be a matter of culture. The work presented demonstrates how, in a microelectronic environment, it is possible to focus on system level simulation taking into account MEMS integration. The study is illustrated on a convective accelerometer fabricated on a CMOS technology. Various issues concerning the design of such system are addressed, from the low-level modeling of the sensing device to system level design of a delta-sigma modulator. Example of MEMS integration into the Cadencereg and Matlabreg environments are provided.

Design-for-manufacturability of MEMS convective accelerometers through adaptive electrical calibration strategy

In this paper, we explore the use of an adaptive electrical calibration strategy in the context of design-for-manufacturing for MEMS convective accelerometers. The calibration principle relies on the adjustment of the heater power level such that sensitivity is set to a given target value. The idea is to define multiple sensitivity targets in order to improve production yield and to insert a criterion on power consumption. Different device binning can then be achieved depending on test limit settings. Results obtained from Monte-Carlo simulation are presented to demonstrate potentialities of the technique.