Abstract
Most mobile phones today have capacitive micro-electro-mechanical systems (MEMSs) microphones that use either single or dual diaphragms. Methods to detect failures easily and non-invasively have become of critical importance for microphones mobile phone manufacturers as a basis for built-in self-test (BIST) and self-repair (BISR) strategies. In that regard, a four-layer framework is presented that includes lumped element modeling (LEM), failure mode simulation, failure mode discrimination, and recovery. The frequency response of the microphone is taken as the main output to analyze. To experimentally validate this framework, this article provides a failure mode induction method based on bias voltage sweeping and four new techniques, based solely on acoustic measurements to discriminate the states of electrostatic capture for single diaphragm capacitive MEMS microphones. These include 1) analysis of an acoustic signature that is unique to electrostatic capture based on cosine similarity analysis; 2) −3 dB point measurement; 3) +3 dB point measurement; and 4) cluster analysis. Measurement of pull-in voltage and snapback voltage ranges is further demonstrated based on sensitivity measurements in laboratory conditions and response magnitude and noise power measurements in non-laboratory conditions. Up to 100% success rate in detecting electrostatic capture of diaphragm is reported for this type of device.
Highlights
THE detection, characterization and understanding of the time behaviour of system failures is critical for the diagnosis and prognosis of health of assets in all industrial sectors
In the case of micro-electro-mechanical systems (MEMS), this procedure is documented in various JEDEC standards such as Assuming that voltage values can be varied, two voltage thresholds can be established: the pull-in voltage, Vpi, which refers to the voltage for which the plates snap together, and the snapback voltage, Vsb, that describes the voltage for which the plates are released from the captured state
While lumped element modelling does not possess the accuracy of finite element modelling (FEM), it proved to be sufficient for highlighting the trends in failure mode induced deviations
Summary
THE detection, characterization and understanding of the time behaviour of system failures is critical for the diagnosis and prognosis of health of assets in all industrial sectors. Obtaining empirical data can be challenging if there are insufficient faulty devices at hand as is the case for returns from some customer products. A typical remedial approach to acquire data is to artificially induce the faults. In the case of micro-electro-mechanical systems (MEMS), this procedure is documented in various JEDEC standards such as. The physics related to this phenomenon has been intensively studied and results obtained have been used to avoid membrane collapse for Manuscript received XX XX, 2021; accepted XX XX, 2021. Date of publication XX XX, 2021; date of current version XX XX, 2021. Recommended for publication by XXXXX XXXXX upon evaluation of reviewers’ comments.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
