Many seasoned power supply design engineers who were educated in the "old school" prior to when computer science became a major part of the electrical engineering curriculum of the universities are resistant to using circuit simulation programs for exercising and evaluating their designs. Many of their arguments are quite sound. Two such arguments are discussed:

The most prevalent argument is related to the stability of the power supply. The seasoned designer most likely took a course in control system theory and applications and has since designed many feedback control systems, of which, power supplies are a very simple example. The seasoned power supply designer can, by experience, determine the worst case conditions for input line and load and very quickly determine the stability margins. But the stability margins are also dependent on the EMI input filter and the impedance of the power line. To solve that stability equation, the experienced power supply designer refers to hallmark tutorial papers where the input filter issues are defined and treated.

Another argument involves determining the dynamic switching performance of the power supply circuit. The seasoned designer will often complain that the simulated circuit does not usually contain enough of the stray components to be a reasonable facsimile of the actual circuit to be constructed. For that reason, the simulation circuit must become overly complex with a large number of stray components connected to the circuit. For instance, the capacitance between the power transistors, rectifiers and the heat sink is very important. The inductance of critical wire leads and passive components as well as the self resonant frequencies of critical passive components are also very important. Other common mode and differential mode affects as well as mutual inductive coupling must be considered.

The experienced circuit designer would then assert the simple fact that regardless of the results of a simulation of the circuit, no smart design engineer would release the circuit for manufacture prior to building a breadboard reflective of most or all of the stray conditions. The testing of the circuit is necessary in any case so why invest time and money in a modeling step if it will not eliminate any of the other steps involved?

The real question should be "Is it possible that the simulation can provide information related to the performance of the circuit that is virtually impossible to obtain from the breadboard?" There are design verification areas that the simulation can improve.

During the recent simulation of a very simple buck converter, the circuit of which is illustrated in Figure 1, I questioned the waveshape of the actual gate drive signal of the power MOSFET, X1. I knew that it was being driven by a 10 Volt pulse common with the source of X1. However, there was a series resistor, R3, betwe en the pulse source and the gate of X1.

The gate to source of a common drain transistor is very difficult to measure. Attempts have been made using differential inputs and matched oscilloscope probes. The problem with that type of measurement is that the common mode rejection of an oscilloscope plug-in worsens as a function frequency. With the sharp rise and fall times of switching mode power supplies, the differential oscilloscope method is useless. Special, very expensive differential probes are also available but they suffer the same degredation as the high frequency components rise.

Another attempt has been made by floating the power supply and connecting the source of the MOSFET to ground. While utilizing that method, an accurate measure of the gate to source waveform is easily made; however, since the circuit has to be literally turned upside down to perform that measurement, the stray component conditions are all very different from actual. So the measurement, although quite accurate, is usually meaningless.

Another method is to simulate the circuit and store the waveforms for both the gate and source Voltages of the MOSFET and then subtract one waveform from the other. That simulation was performed. Then the source to ground Voltage was subtracted from the gate to ground Voltage resulting in the gate to source Voltage of the MOSFET. Figure 2 illustrates the V2 pulse Voltage, the gate to source Voltage and the source to ground Voltage, all of the MOSFET. Notice that during the time that the MOSFET's source Voltage is rising, the Voltage on its gate is limited to a plateau resulting from the Miller capacitance within the MOSFET and the series resistor, R3. Once the source reaches the 24 Volt battery Voltage, the gate begins to rise to reach the 10 Volt pulse level. Upon turn off, the miller capacitance working with R3 again comes into play causing a widening of the source Voltage output pulse when compared to the width of the Gate to source pulse.