Modeling and numerical methods for power electronic devices

Reiman, Chloe Michelle

Permalink

https://hdl.handle.net/2142/105674 Copy

Description

Title

Modeling and numerical methods for power electronic devices

Author(s)

Reiman, Chloe Michelle

Issue Date

2019-07-11

Director of Research (if dissertation) or Advisor (if thesis)

Rosenbaum, Elyse

Doctoral Committee Chair(s)

Rosenbaum, Elyse

Committee Member(s)

• Krein, Philip
• Hanumolu, Pavan
• Banerjee, Arijit

Department of Study

Electrical & Computer Eng

Discipline

Electrical & Computer Engr

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Date of Ingest

2019-11-26T20:35:02Z

Keyword(s)

• silicon controlled rectifier
• SCR
• PNPN
• latch-up
• 14-nm
• switched-mode power supply
• SMPS
• flyback
• boost
• state vector simulation
• event detection

Abstract

A scalable I-V model for latch-up in non-collinear PNPN devices is adapted from a previous model for collinear SCR devices. The model is applied to 14-nm FinFET test structures. Layout scaling trends for key latch-up metrics, such as holding and trigger voltage, are captured by the model in circuit simulation. TCAD simulation is used to gain physical insight into the behavior of non-collinear PNPN devices. The dynamic behavior of switched-mode power supplies is simulated by adopting a state-space representation. Piecewise linear models are used to represent the nonlinear switching devices within the power supplies. With state-space representation models, averaging techniques can be used to speed up simulation time. A reduced-order averaged model is used to predict the dynamic turn-on behavior of a flyback converter. A correction factor is added to the model to account for the effect of the snubber circuit. An Elementary Effects algorithm and a Bayesian inference routine are used to fit the averaged model to a more expensive netlist model. State-space models can also be used with the sampled-data method for state vector simulation. This approach is more accurate than the averaged model, but more computationally expensive. Computation time is reduced by calculating the matrix exponential using a decomposition method. With an efficient means of computing the matrix exponential, switching instances are updated reliably from previous computed values, providing a very quick means for event-detection state vector simulation.

Graduation Semester

2019-08

Type of Resource

text

Permalink

http://hdl.handle.net/2142/105674

Copyright and License Information

Copyright 2019 Chloe Reiman

Owning Collections