Control by PWM

It is necessary to modify the step-down SMPS if it is to operate at relatively high voltage in the continuous conduction mode. The problem with continuous conduction occurs when the MOSFET is turned on with the diode still in conduction. To explain this aspect, Fig. 4 is presented. When the MOSFET is turned on, the diode can act as a short circuit till it recovers. To overcome this problem, the circuit in Fig. 4 is to be modified.

The modifed circuit is presented in Fig. 5. This circuit, when designed properly, would well whether the conduction is continuous or discontinuous. This circuit contains a few additional components. The components added are an additional diode marked as D1, an RC snubber circuit for D1, a capacitor labeled as C2. Diode D3, inductor L2 and capacitor C3 are the same components, marked as D, L and C respectively in Fig. 4. The operation of the circuit in Fig. 5 is explained now.

The conduction path that exists when the MOSFET is ON is shown in red colour in Fig. 6. The current flow is through the MOSFET and the inductors.

When the MOSFET is turned off, the set of components in conduction varies. In mode 1 following immediately after the turn-off of the MOSFET, diode D1, inductor L1, capacitor C1, resistor R1 and components L2 C3 and RL are in conduction. The RC circuit in series with D1 conducts, the time constant associated with C1 and R1 being very small compared with the on-off periods of the MOSFET. During this mode, inductor L1 discharges its energy to R1 and C1. The components in conduction in mode 1 are shown in Fig. 7.

When mode 1 is over after inductor L1 has discharged its energy, mode 2 follows. Here only D2 continues to conduct, but L1 and D1 are not in conduction. This mode is illustrated in Fig. 8. During this phase, C2 would discharge any energy it may have acquired into R2. After the end of the period corresponding to the switching frequency, the MOSFET is turned ON again and the circuit reverts to the state shown in Fig. 9.

The circuit lasts in the state shown in Fig. 9 till the current through L1 becomes equal to that through L2. After that, the circuit reverts to the state shown in Fig. 6.

When the MOSFET is switched on, current through L1 rises gradually and current through D2 falls gradually. If inductor L1 is sufficiently large, there would be hardly any reverse recovery transient due to diode D2. Typically L1 should be such that the rate of rise current is less than 25% of the maximum rate of fall specified for diode D2. Inductor L1 can even be an air-core inductor, whereas inductor L2 has a ferrite-core with an air gap.

The fourth applet presented below simulates the circuit in Fig. 5. It is assumed that the turn-on delay of diodes is negligible. The puul-down menu contains two items that are not shown explicitly in Fig.5. They are the equivalent-series resistance of capacitor C3 and the internal resistance of inductor L2. In practice, even with a fixed duty cycle, a fixed switching frequency and a steady-input voltage, the ripple content in output tends to be higher than the calculated value, mainly due to the ESR of C3. The default value of ESR of C3 is set to be zero. To see its effect, the type of response should be set as Statistics. Then when the program is run, the peak-to-peak ripple in output voltage is displayed. The program has to be run a few times in this mode before the peak-to-peak ripple in output settles down to its periodic value. For example, when ESR of C3 is increased to 0.1 W ,it can be seen that the peak-to-peak ripple that results is much higher.

The internal resistance of L2 reflects the winding resistance and the core loss and the losses in inductor L2 tend to reduce the efficiency of this converter. The default value of internal resistance has been set to be zero. Its realistic value can be calculated by assuming a quality factor of about 100 at the switching frequency. For the value of inductor L2, an appropriate value of internal resistance of L2 can be set to be about 0.5 W. To see its effect, the type of response should be set as Statistics in the program.

When the type of response is Statistics, the program should be run a few times before the results become repetitive. Even then, the sum of output power and all the losses may not add upto the input power, due to errors in modelling and the presence of energy storage elements. The loss calculation of diode D1 in particular is quite incorrect, because the turn-on losses of a diode have been ignored. Even though the average current of D1 turns out to be small, its turn-on losses are quite significant and it is advisable to use the same diode selected to be used as D2. These diodes should be fast-switching diodes with a low reverse period of about 50 ns.

click here to open the applet

APPLET FOR PRACTICAL BUCK CONVERTER IN FIG. 5

The modified circuit shown in Fig. 5 is better than the ideal circuit, since it tries to fix the problem that arises due to the reverse recovery transient in diode D2. The modified circuit in Fig. 5 works well so long as inductor L1 is large enough to prevent diode D2 from going into reverse recovery transient process. It can be seen from the applet for reverse recovery transient in a diode that there is hardly any reverse recovery current if the test inductance is large. If L1 shown in Fig. 5 is not large enough, the modified circuit's behaviour is unpredictable. When the MOSFET turns ON, current through L1 builds up and current through D2 falls, with hardly any change in current through L2. When the current through D2 becomes zero too soon, there would stille be charge carriers trapped inside D2 and it would continue to conduct. The rise in reverse recovery current would be restricted by L1, but when D2 snaps into off condiction, the current in L1 could be larger than that in L2 and then we have a circuit with a potential danger. If there be nod path for excess current in L1, there would be a voltage transint due to L1, which in turn can lead to damage of either the MOSFET or diode D2 or both. This problem can be averted by repositioning D1, R1 and C1 as shown in Fig. 10. This circuit behaves exactly in the same way as the circuit in Fig. 5, except that it has a built-in mechanism to handle the problem described above. In case when the current through L1 is higher than L2 when diode D2 snaps off, the excess current can free-wheel though D1, R1 and C1. The circuit in Fig. 10 is the recommended modified circuit.