Test intégré pseudo aléatoire pour les composants microsystèmes

Abstract : The growing use of MEMS in life-critical applications has accelerated the need for robust test methods. MEMS have complex failure mechanisms and device dynamics that are most often poorly understood. This is due to their multi-domain nature which makes them inherently complex for both design and test. Manufacturing is in addition complicated by the need of new fabrication steps in particular when System-in-Package (SiP) techniques are used. These packaging techniques enable to have a module that contains highly heterogeneous IP blocks or chips, giving important benefits in terms of time-to-market shortening and miniaturization. However, this poses many test problems. In this area, BIST techniques for analog and mixed-signal circuits have attracted considerable industrial interest for helping reduce increasing test related difficulties. In this thesis we propose a pseudorandom (PR) functional BIST for MEMS. Since the test control is necessarily electrical, electrical test sequences must be converted to the energy domain required by the MEMS. Thus, we propose the use of pseudorandom electrical pulses that have the advantage of being easily generated on-chip and the conversion to the actual energy domain has been demonstrated for different types of MEMS. We show how different types of PR sequences can be exploited within a BIST approach for both linear and nonlinear MEMS. In general, we show that two-level PR sequences are sufficient for testing both linear and nonlinear MEMS. In addition, while two-level PR sequences are sufficient for characterizing linear MEMS, we describe how the use of multilevel PR sequences is necessary for the characterization of nonlinear MEMS. The number of needed levels depends on the order of nonlinearity of the MEMS under test. The output test response is digitized using an existing on-chip self-testable ADC and a digital circuit performs some simple digital signal processing to extract Impulse Response (IR) samples for linear MEMS, or Volterra kernel samples for nonlinear MEMS. Next, these samples (called test signature) are compared with their tolerance ranges and a pass/fail signal is generated by the BIST. We use Monte Carlo simulations to derive the test signature tolerance ranges out of the specification tolerance ranges. Monte Carlo simulations are also used to form the test signature after a sensitivity analysis, and to inject parametric variations to calculate the test metrics and to optimize BIST design parameters, such as the length of the LFSR and the bit precision of digital circuitry. We have applied the PR BIST for MEMS like commercialized accelerometers and microbeams that we have fabricated. Satisfactory experimental results have been obtained.
Complete list of metadatas

Cited literature [171 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00135916
Contributor : Lucie Torella <>
Submitted on : Friday, March 9, 2007 - 3:50:45 PM
Last modification on : Wednesday, May 16, 2018 - 6:30:04 PM
Long-term archiving on : Tuesday, April 6, 2010 - 10:08:24 PM

Identifiers

  • HAL Id : tel-00135916, version 1

Collections

TIMA | UGA

Citation

A. Dhayni. Test intégré pseudo aléatoire pour les composants microsystèmes. Micro et nanotechnologies/Microélectronique. Institut National Polytechnique de Grenoble - INPG, 2006. Français. ⟨tel-00135916⟩

Share

Metrics

Record views

232

Files downloads

823