MURAL - Maynooth University Research Archive Library



    Non-linear Dynamic Transformer Modelling and Optimum Control Design of Switched-mode Power Supplies


    Vu, Tue T. (2014) Non-linear Dynamic Transformer Modelling and Optimum Control Design of Switched-mode Power Supplies. PhD thesis, National University of Ireland Maynooth.

    [img]
    Preview
    Download (8MB) | Preview


    Share your research

    Twitter Facebook LinkedIn GooglePlus Email more...



    Add this article to your Mendeley library


    Abstract

    With recent advances in semiconductor manufacturing and computational technology, digital control systems have grown to a relatively mature stage, and will soon become a viable replacement for their analogue counterparts in the design of isolated and non-isolated DC-to-DC converters in general, and yback converters in particular. Inspired by this possibility, the thesis adopts the first-ever digital control design in the field for wide-operating range yback converters, based on a low-cost microcontroller. Accurate transformer modelling is a necessary exercise for the study of the yback converters as well as for model-based controller design. Therefore, a non-linear dynamic model, which allows an accurate representation of both linear dynamics and non-linear core behaviour in a practical transformer, is proposed. The parameters of the proposed transformer model are obtained using time-domain system identification based on experimental data. In order to reduce the round-of error typically occurring in the collected time-domain data, a method which is based on adjusting the value of the current sensing resistor is also adopted. To facilitate control design, a control-oriented model is developed based on the full converter model through a simplification step. As demonstrated in the thesis, the control-oriented model is able to preserve the bulk of the full converter model fidelity, critical for a control design step, while at the same time requiring a significantly shorter execution time for simulation when compared with the full converter model. For the purpose of implementing isolated-feedback control within a low-cost microcontroller, a magnetic sensing principle, which can operates in both continuous and discontinuous conduction modes of the yback converter, is developed. The proposed sensing principle is also based on the bias winding voltage of the yback transformer to estimate the converter output voltage; however, the sampling instant is chosen at the point where the secondary current is known, instead of the knee point where the secondary current is zero. The implementation of the proposed sensing technique, based on analogue circuitry and a microcontroller, is also studied. Finally, optimum digital control for a wide-operating range yback converter is developed and implemented. The control architecture is purposely designed to perform a variety of tasks, including efficiency optimisation, magnetic sensing, and valley switching operation, in addition to the main task of regulating the output voltage. Three different methods for synthesizing optimum compensators, based on mixed-sensitivity H1 robust control theory, gain-adaptive predictive functional control (GAPFC) theory, and gain-adaptive quantitative feedback theory (GAQFT), are also studied. In order to improve the performance of the robust controllers, parametric variations of the yback converter models are minimized before applying the robust control. Two possibilities for reducing converter parametric model uncertainty, based on adapting the converter open-loop gain and varying the sampling rate of the digital controller, are also demonstrated.

    Item Type: Thesis (PhD)
    Keywords: Non-linear Dynamic Transformer Modelling; Optimum Control Design; Switched-mode Power Supplies;
    Academic Unit: Faculty of Science and Engineering > Electronic Engineering
    Item ID: 5615
    Depositing User: IR eTheses
    Date Deposited: 15 Dec 2014 12:03
    URI:

      Repository Staff Only(login required)

      View Item Item control page

      Downloads

      Downloads per month over past year