MEMS in closed-loop

The validation of a MEMS electronic loop (SENSOR + ASIC) can easily become a nightmare without appropriate means for ASIC specification and validation.

The Dummy MEMS is an electrical emulator of the mechanical sensor behavior in closed loop. The main issue for specifying it is to assess with the right accuracy each of its domains of equivalence with the sensor.

Key Objectives

  • Performing system prototyping with the capability to emulate the physical signal being sensed
  • Optimizing the ASIC design
  • Validating the ASIC performances in closed loop
  • Getting a MEMS electronic loop working in a single pass
  • Operating over a range where the MEMS displays a linear behavior

Equivalence Domains for an accelerometer

  • Electrical input to represent the external acceleration
  • Feedback force consists in any combination of the following (treating separately or combining) : sensor linearization and activation, force Balance and compensation of process variations
  • First resonance frequency in closed loop
  • Sensor sensing capacitance
  • Closed loop gain
  • Second resonance frequency
  • Parasitics resistance
  • MEMS Thermal Noise
  • ...

The best way to design a purely electronic Testbench for the ASIC

  1. Specify the Dummy Sensor
  2. Perform simulations with the electronics
  3. Develop the Dummy Sensor
  4. Perform validation of the electronics with the Dummy Sensor
  5. Check the MEMS electronic loop with the real Sensor

Mechanics to Electronics Translation

The purpose is to translate the behavior of mechanics into the electronics domain through efficient modeling techniques.
The whole process makes sense as soon as the level of mechanical modeling allows a piecemeal construction using the EMBLEM library elements from an analysis of the geometries of the mechanical structure.
Similarly the electronic dummy is synthesized on a library of circuit models.

 

 

SoC design with MEMS

 

MEMS Sensors, coupled with SoCs, are becoming mandatory for many industrial applications targeting high resolution.
How to ensure the fastest Return-on-Investment, whereas the whole system robustness and yield depend mainly on variations of the MEMS Sensor behavior?
Dolphin contributes its multi-domain competencies: behavioral model library EMBLEM for mechanics, EDA solutions - SMASH and SLED - and dedicated Virtual Components of Silicon IP.
The documented top-down technique enables to ensure stability and yield of complete system: interactions between the MEMS sensor and the affordable constraints on the SoC feasibility and performance are explored and quantified. The optimum tradeoff is then set for meeting economical targets.

Black box

Key Benefits

  • Reduced Time to Market through shortening of specification, design and characterization phases
  • Brief latency in the "Sensor + ASIC" close-loop MEMS: the sensor is part of the feedback loop with the analog modulator only
  • Feedback force treating separately or combining Sensor activation, force balance and compensation for fab variations
  • Linearization of sensor behavior by feedback control to limit the range of variation
  • Yield optimization of MEMS fabrication through process tolerance analysis on virtual testbench
  • Ascertainment of equivalence between simulations and silicon measurements thanks to diagnostic modes in characterization phase
  • Simplicity and shortness of the SoC test protocol thanks to electrical control of the Dummy Sensor
  • Ability to perform system calibration and to compensate process variations

A Multi Domain and Multi Level Approach

  1. Specifying a high resolution system
  2. Modeling Techniques for yield
  3. Checking the Virtual Prototype through multi level simulation
  4. Silicon Validation with Dummy Sensor Prototype
  5. Silicon Characterization with Dummy Sensor including the sensor process variations
  6. Embedded Calibration of a MEMS-based system

 

 

A technique for mastery of MEMS yield

experienced by Dolphin Integration

 

loop system

Figure 1: Multi Domain system coupling electronics and mechanics

1 - Specifying a High Resolution closed loop system

Key Features

  • Assignment of global noise budget
  • Definition of the calibration process
  • Compensation of the MEMS process variations
  • Activation with adequate pulse shaping
  • Specification of the DUMMY SENSOR as an electrical emulator of the SENSOR
  • Definition of the testability
    • Reflex for the SENSOR signature
    • Self Test for the ASIC with the DUMMY SENSOR
Modeling for yield

Figure 2 : Virtual Testbench with Behavioral Simulation Models
Figure 3: Yield analysis with σ = 200 Hz | 100 % of circuits get a SNR > 120 dB

2 - Modeling for Yield

Key Features

  • Sensor Modeling with EMBLEM-MECHA library
  • Overall behavioral model with mechanical and electrical couplings
  • Assessment of tolerance margin in each domain
  • Performing loop simulations in reasonable runtime
  • Computing yield analysis
Multi level simulation

Figure 4: Noise Reference Tolerance Template
Figure 5: FFT Curves computed within SMASH – Ideal versus Real

3 - Checking the Virtual Prototype through multi level simulation

Key Features

  • Ascertainment of equivalences between the different level
  • simulation models
  • Jitter simulation with the Jitter Tolerance Template (JT2)
  • Noise simulation with the Noise Reference Tolerance Template
  • (NRT2)
  • Loop Stability verification with the secondary resonance
  • Frequency (SRT2)
  • Dispersion Analysis

4 – Silicon Characterization of an ASIC prototype

Key Features

  • Assessment of loop stability with the DUMMY SENSOR
  • Evaluation of the impact of stress on the interfaces
    • Jitter on input clock
    • Noise on power supply
    • Noise on reference voltage
  • Customization of ASIC compensation to each sensor
  • Ascertainment of equivalence between the simulation results and
  • the silicon measurements
ASIC prototype

Figure 6: Real Testbench for silicon characterization
Figure 7: Illustration of correlation between simulation and measurements

5 –Embedded Calibration of a MEMS-based system

Calibration process as part of the design

  • Determination of the sensitivity of each circuit
  • Dynamic programming of the architecture to reach the stability and the performance in noise floor
  • Synchronization of emulations
  • Noise reference
  • Jitter
  • Power Supply
  • Customization of signal conditioning (first stage of the ASIC)
MEMS-based system

Figure 8: MAPLE as calibration station