GR-06 - ELECTRONIC DESIGN AUTOMATION - EDA
Creation of a generator for a gain boosted Operational Transconductance Amplifier (OTA) with AI features
Tesi esterna in azienda Tesi all'estero
Riferimenti DANIELE JAHIER PAGLIARI
Riferimenti esterni Rocco Calabro: firstname.lastname@example.org
Matteo Dalla Longa: email@example.com
Riccardo Fontanini: firstname.lastname@example.org
Aurelio Teliti (POLITO).
Descrizione Thesis to be carried out at Infineon Austria (Villach). The internship period will last 6/9 months, starting January 2024, and will be compensated.
The automatization of repetitive tasks is a key-point in modern electronic industry to unlock the potentiality of their employees and speed-up the design of analog circuits. The Anagen project (Analog Generators project) aims to automatize the implementation of widely used analog design blocks by transferring part of the knowledge of the mixed signal analog designers into a scalable and accessible platform. This should not be seen as a simple repository of electrical schematics, but as a framework where analog designers interact and share their experience providing the fundamental building blocks of Infineonís chips. These blocks should be able to morph their features, adapt to different use cases and be robust in terms of process and temperature variations for different working conditions. Consequently, the design of an analog block generator is challenging, and we are looking for a motivated master student for this task.
The candidate is expected to work on the following tasks:
1. Face Infineonís analog design and simulation tools (spice simulators and Cadence environment)
2. General analysis of a gain-boosted OTA
3. Exploration of the optimal sizing of OTA components
4. Implementation of a Neural Network / Genetic algorithm to perform an initial sizing of the circuit
5. Embedding the initial sizing platform in an anagen generator empty shell
6. Optimization of the working point through gradient based algorithms
7. Development of a graphical interface for the generator
Conoscenze richieste Basic analog design skills (analysis and design of simple analog circuits, model linearization, gain analysis etc..), basic python scripting knowledge.
Plus: PyTorch for neural network implementation, CUDA and parallel programming skills for acceleration purposes, git file versioning framework.
Note Thesis to be carried out at Infineon Austria (Villach). The internship period will last 6/9 months, starting January 2024, and will be compensated.
Scadenza validita proposta 05/10/2023 PROPONI LA TUA CANDIDATURA