Simulating MicroElectroMechanical Systems

Transcription

1 Simulating MicroElectroMechanical Systems David Bindel UC Berkeley, CS Division Simulating MicroElectroMechanical Systems p.1/30

2 Overview Simulation: motivations and approaches Introducing SUGAR Using SUGAR Ongoing work Conclusion Simulating MicroElectroMechanical Systems p.2/30

3 Simulation: What? Process simulation (technology CAD) Device simulation System simulation Simulating MicroElectroMechanical Systems p.3/30

4 Simulation: Why? Test what if ideas and initial designs Design verification, fine tuning, and parasitic discovery Different objectives mean different speed and accuracy requirements Simulating MicroElectroMechanical Systems p.4/30

5 Simulation: How? Solve continuum equations (finite elements, finite differences, boundary elements) Solve simplified equations of beam and plate theory (finite elements, etc) Solve network equations (e.g. modified nodal analysis) These approaches are not mutually exclusive! Simulating MicroElectroMechanical Systems p.5/30

6 SUGAR: System simulation Goal: Be SPICE to the MEMS world Fast enough for early design stages Simple enough to attract users Capable of simulating interesting coupled physics Simulating MicroElectroMechanical Systems p.6/30

7 SUGAR group Faculty Grad students Undergrads A. Agogino D. Bindel W. Kao Z. Bai J.V. Clark A. Kuo J. Demmel D. Garmire E. Zhu S. Govindjee B. Jamshidi M. Gu R. Kamalian K.S.J. Pister S. Lakshmin J. Nie N. Zhou Simulating MicroElectroMechanical Systems p.7/30

8 Related work Nodas from CMU Coventorware SPICE, CADENCE, HDL-A,... ANSYS, ABAQUS, FEAP,... Simulating MicroElectroMechanical Systems p.8/30

9 What can we model? Simulating MicroElectroMechanical Systems p.9/30

10 SUGAR architecture Netlist Solvers System assembly Models Results Transient analysis Steady state analysis Static analysis Sensitivity analysis Matlab Web Library Interfaces Simulating MicroElectroMechanical Systems p.10/30

11 Netlist syntax mybeam beam2d p1 [A B] [l=100u w=2u] mybeam is the element name (optional) beam2d specifies the 2-d beam model p1 specifies the properties of the POLY1 MUMPS layer A and B are the beam end-nodes [l=100u w=2u] specifies the length (100 microns) and width (2 microns). Simulating MicroElectroMechanical Systems p.11/30

12 Parameterization param nfingers = 10 nfingers can be set by user at load time If no value explicitly provided, use ten fingers Simulating MicroElectroMechanical Systems p.12/30

13 Subnets and heirarchical design Subnets are parameterized components Subnet calls look like built-in model calls Matches design heirarchies Can put commonly-used subnets in a library Simulating MicroElectroMechanical Systems p.13/30

14 Subnet example subnet XMass [A B] [finger_len=*] [ b1 beam3d parent [A b1] [l=25u w=50u h=6u oz=-deg(90)] b2 beam3d parent [b1 B ] [l=25u w=50u h=6u oz=-deg(90)] b3 beam3d parent [b1 b2] [l=finger_len w=2u h=6u oz= deg(0) ] b4 beam3d parent [b1 b3] [l=finger_len w=2u h=6u oz=deg(180)] ] Simulating MicroElectroMechanical Systems p.14/30

15 Arrays XSusp p1 [c(1)] [susp_len=200u] for k=1:nfingers [ mass(k) XMass p1 [c(k) c(k+1)] [finger_len=100u] ] XSusp p1 [c(11)] [susp_len=200u oz=pi] Suspensions at either end nfingers comb-finger units as defined above Simulating MicroElectroMechanical Systems p.15/30

16 Result: simplified ADXL-50 x Y vertical [m] X horizontal [m] x 10 4 Simulating MicroElectroMechanical Systems p.16/30

17 Types of analysis Transient response Static equilibrium (DC analysis) Steady-state response Mode shape determination Simulating MicroElectroMechanical Systems p.17/30

18 Example: Modal analysis 2 x x 10 4 Mode 3, frequency Hz Y vertical [m] 0 Y vertical [m] X horizontal [m] x X horizontal [m] x 10 4 Dogbone suspension for comb drive resonator Simulating MicroElectroMechanical Systems p.18/30

19 Using the Matlab interface 0 x x Y vertical [m] dq = []; for k=1:12 end param.v = k; X horizontal [m] x 10 4 net = cho_load( beamgap2b.net, param); dq = cho_dc(net, dq); tip(k) = cho_dq_view(dq, net, c, y ); Simulating MicroElectroMechanical Systems p.19/30

20 SUGAR 3.0 SUGAR 2.0: Matlab core, C add-ons SUGAR 3.0: C core, Matlab interfaces Release target: some time this month? Web service using 3.0 should be more robust Can use SUGAR 3.0 even without Matlab Simulating MicroElectroMechanical Systems p.20/30

21 M&MEMS: SUGAR on the Web Simulating MicroElectroMechanical Systems p.21/30

22 M&MEMS: Pros and cons Pro: Only need a web browser with Java Pro: Fixes can be applied rapidly Pro: Potential parallelism Con: Less flexibility than Matlab environment Con: Too many failure points in initial version Con: Poor response time in initial version Simulating MicroElectroMechanical Systems p.22/30

23 M&MEMS: Lessons from EE245 Used in Introduction to MEMS course, Fall 01 (Pister) Problems with both hardware and software Matlab too heavyweight in this environment Students used it anyhow! Simulating MicroElectroMechanical Systems p.23/30

24 Matisse: Integrating measurement Set of optical measurement tools Make available on the web as a virtual lab Matisse team: R. Kant, R. Muller, C. Rembe, M. Young Simulating MicroElectroMechanical Systems p.24/30

25 Matisse: Closing the design loop Simulating MicroElectroMechanical Systems p.25/30

26 Design synthesis and optimization Genetic algorithms to evolve new designs Specializing designs from a library N. Zhou, B. Zhu, A. Agogino, and K. Pister: "Evolutionary Synthesis of MEMS (Microelectronic Mechanical Systems) Design" (ANNIE 2001). First Runner-up for Novel Smart Engineering System Design Award. Simulating MicroElectroMechanical Systems p.26/30

27 Model reduction Used to create macromodels from detailed device simulations Also used to reduce large system models Need to handle nonlinear effects Z. Bai is the local expert on ROM Simulating MicroElectroMechanical Systems p.27/30

28 Improved models Geometric nonlinear beams Torsion-bending coupling for single-crystal Si Contact More coupling effects: electrothermal, thermomechanical, electromechanical,... Plates, hinges, other mechanics Improved damping models Improved thermal models Simulating MicroElectroMechanical Systems p.28/30

29 Improved numerics Incorporate current technology: SuperLU, DASSL and IDA, etc. Improved time integration Scaling and improved error measures Continuation for improved static analysis and study of bifurcations Sensitivity analysis Incorporate parallel techniques where appropriate Simulating MicroElectroMechanical Systems p.29/30

30 Conclusions SUGAR is already a useful design / education tool Lots of activity to extend SUGAR and improve robustness Friendly users appreciated! Simulating MicroElectroMechanical Systems p.30/30