Sensors and Actuators A 123–124 (2005) 185–193 A method for cross-sensitivity and pull-in voltage measurement of MEMS two-axis accelerometers P. Bruschi a , A. Nannini a , D. Paci a,b , F. Pieri a,∗ a Dipartimento di Ingegneria dell’Informazione, Universit` a di Pisa, Via Caruso, 56122 Pisa, Italy b IEIIT-Sezione di Pisa, CNR, Via Caruso, 56122 Pisa, Italy Received 13 September 2004; received in revised form 18 March 2005; accepted 26 April 2005 Available online 22 June 2005 Abstract An electronic circuit for the characterization of two-axis accelerometers is presented, implementing a non-linear method which reduces the effect of parasitic signals. Single-ended and differential measurements were used to extract a value of 0.15 for the cross-sensitivity of the device. Other parameters, like the pull-in voltage, the quality factors and the resonance frequencies, are extracted as well. Curves of the frequency response at different pressure levels, and with different dc bias voltages are presented. © 2005 Elsevier B.V. All rights reserved. Keywords: MEMS accelerometers; Electronic interfaces; Cross-sensitivity; Pull-in voltage 1. Introduction Characterization methods for accelerometers are of paramount importance both for research purposes, and for industrial testing. They are especially convenient if they can be implemented with an automated, computer controlled system. Inertial testing methods based on shakers or turn- tables [1,2] and capacitive sensing are the obvious choice for the measurement of several accelerometer parameters, even if they require specific hardware. Methods based on interferometry [3] or image processing [4] provide extensive information on the dynamical behaviour of the device, but they probably are of a more limited use for industrial testing because of, among other reasons, higher cost of instrumen- tation and more complex measurement set-ups. Electrostatic techniques, based on the capacitive actuation of the moving mass with specifically designed electrodes or with the actual sensing electrodes, are widely used [5,6], especially for built-in self test [7,8], even if it has been recently suggested that electrostatic driving cannot identify every possible defect [9]. Electric methods have the advan- ∗ Corresponding author. Tel.: +39 050 2217661; fax: +39 050 2217522. E-mail address: f.pieri@iet.unipi.it (F. Pieri). tage of being based on readily available instrumentation, already used in other testing applications. Electric testing can be routinely performed at the wafer level, reducing costs and improving reliability. Finally, plenty of commercial hard- ware and software automation solutions are available. The development of methods to measure specific parameters of a MEMS accelerometer with an all-electrical method is thus justified. A difficulty with electric testing of two-axis MEMS accelerometers is that electrostatic driving along one of the channels (used to mimic the effect of an external acceler- ation) may produce a deflection of the moving mass along the orthogonal direction even if the mechanical structure is symmetrical, because of possible asymmetries in the driving electrodes. In this paper we propose a method to discrimi- nate the various components of the cross-axis output signal. Our approach allows the measurement of the cross-axis sen- sitivity of the accelerometer, defined as the ratio between the outputs of the two channels when acceleration is present along one of the two orthogonal directions, without the need of actually exciting it with external accelerations. The method also exploits the intrinsic non-linearity of electrostatic actu- ation to reduce the effect of parasitic capacitances [10]. The measurements are performed with an experimental set-up, 0924-4247/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.sna.2005.04.028