Search

Circuit modeling is an essential tool in the design of power electronic applications. The widespread use of these models, however, depends upon the availability of parametric data for devices and the proper trade-off of the accuracy and simplicity of the model. This paper addresses these two areas with regard to the forward on-state voltage of thyristors and diodes. As has been described in previous papers, forward on-state forward voltage drop can be modeled by the now classical ABCD parameters. This paper discusses the problems and strengths of the ABCD model and describes how the ABCD parameters can be quickly and efficiently calculated by a 3rd order, multiple, nonlinear regression in a new STARS Visual module. This paper proposes an alternate mathematical model to the ABCD model, labeled the MNOP model, and describes the advantages of this model. This paper also describes one of the main limitations of the ABCD model, which is the fact that it is typically used to describe the maximum forward voltage drop on the data sheet, which is NOT an isothermal curve. This, in turn, does not permit the change in the on-state forward voltage drop as a function of junction temperature to be modeled. Isothermal models of diodes and thyristors will be discussed and a new simplified model proposed for surge and overload conditions.

For circuits using SCRs and diodes, both the on-state forward voltage drop and transient thermal impedance are complex characteristics, for which adequate mathematical models been developed. The on-state forward voltage drop can be modeled by both the classical ABCD and the new MNOPQ...parameters. The transient thermal impedance can be well represented via four or five exponential terms representing the significant transient thermal time constants of the device. There is, however, one complication which is not represented in the ABCD and MNOPQ...models. This is the dependence of the on-state voltage on the instantaneous junction temperature. This effect, termed COD or VFIT, requires that the model be electro-thermal, meaning that this important dependence must be accounted for in a microsecond by microsecond manner during the simulation. This paper describes the basic device modeling equations and the techniques required to use these models in "analog" simulation programs such as PSPICE/sup TM/. Parameter extraction from empirical test data is described and the electro-thermal model accuracy evaluated. The final and new result is an electro-thermal model for power electronic application using SCRs and diodes which includes the dynamic effect of instantaneous junction temperature on the on-state voltage. The resulting PSPICE/sup TM/ electro-thermal model is expected to be an important design tool in multi-megawatt power electronic applications, providing the detailed analysis the power electronic engineer needs to quickly evaluate an existing or proposed power electronic application.

Circuit modeling is an essential tool in the design of power electronic applications. The widespread use of these models, however depends upon the availability of parametric data for devices and the proper trade-off of accuracy and simplicity of the model. This paper addresses these two areas as well as describing another modeling subject, the spatial temperature distribution during serge and overloads of thyristors and diodes. The conversion from the physical ladder thyristor (diode) thermal model to the much simpler series equivalent thermal circuit model was described previously. This paper describes a unique multiple regression method to automate the calculation of the parameters for the simplified model. This permits the manufacturer to provide the necessary parameter values for the power electronic design engineer to hard code and/or use SPICE circuit modeling to obtain the waveform of virtual junction temperature verses time in a proposed or existing application. An important characteristic of thyristors is the ability to withstand surge and overloads significantly above their steady state ratings. When considering this ability of the device to absorb and remove heat we are forced to extend our analysis to another dimension, namely, the physical or spatial distance through the device. The spatial temperature-time distribution provides a millisecond by millisecond account of the temperature in each of the device components during surge and overload conditions.