An LTSpice - MATLAB Interface for Mitigating Convergence Problems in Circuit Optimization with SPICE

Publikation: Bidrag til bog/antologi/rapport/konference proceedingKonferenceartikel i proceedingForskningpeer review

13 Downloads (Pure)

Abstrakt

With the recent emergence of wide-bandgap semiconductors the power electronics design process moves gradually towards the digital space. However, SPICE software that is a dominant tool for accurate transient analysis still lacks in both functionality and robustness needed in order to explore large design spaces unveiled by the digital twin models. In this work, a SPICE circuit solver has been interfaced with MATLAB programming environment. Compared to the literature, main focus has been put on mitigating convergence errors inherent to SPICE solvers without altering the solver algorithm itself. Convergence becomes a crucial problem when SPICE is used for optimization within a big solution space, therefore it is necessary to ensure it. This was achieved by implementing parameter variation based on circuit element tolerance, solver parameter adaptation and initial condition inheritance. The effects of each approach are compared for the case of Double-Pulse Test circuit, by the means of Not-a-Number (NaN) results proportion and runtime.
OriginalsprogEngelsk
Titel2022 IEEE International Workshop on Integrated Power Packaging (IWIPP)
Antal sider5
UdgivelsesstedGrenoble, France
Publikationsdato24 aug. 2022
StatusUdgivet - 24 aug. 2022
Begivenhed2022 IEEE International Workshop on Integrated Power Packaging (IWIPP) - Grenoble, France, Grenoble, Frankrig
Varighed: 24 aug. 202226 aug. 2022

Konference

Konference2022 IEEE International Workshop on Integrated Power Packaging (IWIPP)
LokationGrenoble, France
Land/OmrådeFrankrig
ByGrenoble
Periode24/08/202226/08/2022

Fingeraftryk

Dyk ned i forskningsemnerne om 'An LTSpice - MATLAB Interface for Mitigating Convergence Problems in Circuit Optimization with SPICE'. Sammen danner de et unikt fingeraftryk.

Citationsformater