CMOS Integrated Switching Power Converters

In this chapter, the design space exploration is exposed and explained as the main tool to get an optimized design of a fully integrated switching power converter. The converter output voltage ripple is used to constraint the design space, while the merit figure to be optimized is defined as the power efficiency versus the occupied silicon area ratio. Additionally, an overall structure of the design space exploration including particular optimizations for each component design is proposed. Then, simplified models for the converter integrated components (i.e. inductor, capacitor, power drivers and power transistors) are presented, as well as an analytical output ripple model that takes into account the output capacitor Equivalent-Series-Resistance (ESR). Finally, the design space exploration optimization procedure is exemplified by means of the design of a classical Buck converter, where the simplified components models are used. 2.1 Main Concepts and Design Procedure In the first steps of a switching power converter design (once the required topology has been determined), there are many questions and decisions to answer regarding input and output voltage ranges, output current, switching frequency, the reactive components values (mainly inductors and capacitors), transformer input–output voltage ratio (if any is required by the design), even some dynamic performance parameters such as bandwidth. If different technological options are considered to implement the power switches and their corresponding drivers, the design options can become overwhelming. In order to develop an appropriate design procedure, it is interesting to classify the design variables or parameters into different categories to identify their impact upon the design characteristics or performance. In the following, groups proposed herein are presented. Application parameters. Those parameters that are completely defined by the application needs, like input and output voltage ranges, output current capability and output voltage ripple. Usually, application parameters appear as hard constraints in the design space. Additionally, other less conventional G. Villar Piqué, E. Alarcón, CMOS Integrated Switching Power Converters, DOI 10.1007/978-1-4419-8843-0_2, C © Springer Science+Business Media, LLC 2011 9 10 2 The Design Space Exploration: Simplified Case parameters or constrains can be defined: radiated interferences spectrum, maximum input current, etc. . . Static design variables. Conventional design variables, such as inductance and capacitance value, or the voltage conversion ratio of a transformer. We call them ‘static’ because although they offer degrees of freedom in the design stage, they become fixed once the design is implemented. Dynamic design variables. The main dynamic design variable that appears in the design of a conventional switching power converter is the switching frequency, since although it can be determined during the design stage, it offers an additional degree of freedom once the converter is implemented, that allows to better fit to application parameter changes. Performance factor. This category includes those factors that indicate the power converter quality, once application parameters are satisfied or accomplished. Most common performance factors are energy efficiency in power conversion, the bandwidth of the output response versus the control signal, the volume and area occupation, its weight. They can appear as soft constraints, since they might not invalidate the design despite it could be interesting to maximize or minimize them. In fact, depending on the application, they could be totally or partially defined as application parameters (i.e. in case where a minimum power efficiency is required, or a maximum weight is allowed). Technological information. This refers to information related to the physical implementation of the converter components, specially in case of the power switches and their corresponding drivers. Although first order approach designs do not require detailed information about the technology used to implement any component, information about non idealities and parasitics is needed to take into account some performance factors (e.g. energy efficiency). In Table 2.1, the previous classification is summarized. In this case, only those variables that are particularly relevant for the case of integrated switching converters are shown. Obviously, the categories and what they include could change if non conventional designs are considered, i.e. multiple conversion ratio transformers, variable capacitors, etc. . . From the previous classification a design procedure is derived, which is described in the following. 1. First of all, taking into account the application parameters, some converter topologies are selected to fit specifications, and the corresponding required ideal analytical expressions for voltages and currents are obtained. In this design step, system-level simulations can be used to confirm that the selected topologies (it could be interesting to consider more than one topology, in order to select the optimum one afterwards, from the performance factors evaluation) suit the application parameters. 2.1 Main Concepts and Design Procedure 11 Table 2.1 Design variables and parameters classification used in the thesis Category Parameters/Variables Application parameters Input voltage range Output voltage range Output current range Output voltage ripple Static design variables Inductance value Capacitance value Dynamic design variables Switching frequency Performance factors Energy efficiency Occupied area Technological information Equivalent parasitic resistances for each component Parasitic capacitors for each component Capacitive density (capacitance per area) Inductive density (inductance per area) 2. Then, the design variables (static and dynamic) impact on the design are determined by means of ideal expressions or model, and some performance factors can be evaluated (i.e. maximum inductor current). At this point, converter topologies options should be shrunk down to a few ones that suit the application; and some design variables range of values should be partially delimited, by checking some application parameters fulfillment (e.g. maximum output voltage ripple). Power switches are considered ideal and no parasitic components or non idealities are considered. 3. The following design step implies a further increase in the model complexity. It is necessary to decide the technological implementation of the different components of the switching power converter. In case of integrated monolithic converters, this requires to select the microelectronic process used to manufacture the design. The foundry provides information about what is feasible or unfeasible to implant in the selected process, and detailed models for components that lead to parasitic characterization. The addition of all this information generates a notable increase in the converter model complexity, and circuit-level simulations become fundamental (although complex system-level simulation are still interesting). The result of this design stage is the evaluation of performance factors such as energy efficiency and volume (or area) occupation, which should lead to an optimized design. It should be noted that steps 2 and 3 of such design procedure should be carried out for any different selected topology, to choose the one that provides the best performance factors. Furthermore, it is possible that, in spite of the accomplishment of the application parameters, the evaluation of the performance factors results in so poor quality or performance that another topology or design must be considered from the initial point (Fig. 2.1 summarizes the whole process explained above). Therefore, once the design variables have been identified and their corresponding range of values have been constrained by the application parameters criteria, any combination of design variables values should be explored to evaluate the selected 12 2 The Design Space Exploration: Simplified Case Application parameters and general information Ideal expressions for relevant currents and voltages: iL = ... ; vC = ... Design variables identification : {L, C, f, ...} Application parameter checking Design variables range of values reduction Technological information 0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40 –8 –6 –4 –2 0 2 4 6 8 10 Optimized design S ep 1