Home | Articles | Forum | Glossary | Books |
AMAZON multi-meters discounts AMAZON oscilloscope discounts
1 INTRODUCTION Snubber networks (usually dissipative resistor-capacitor diode networks) are often fitted across high-voltage switching devices and rectifier diodes to reduce switching stress and EMI problem during turn-off or turn-on of a switching device. When bipolar transistors are used, the snubber circuit is also required to give "load line shaping" and ensure that secondary breakdown, reverse bias, and "safe operating area" limits are not exceeded. In off-line flyback converters, this is particularly important, as the flyback voltage can easily exceed 800 V when 137-V (ac) voltage-doubled input rectifier circuits or 250-V (ac) bridge rectifier circuits (dual input voltage circuits) are used. 2 SNUBBER CIRCUIT (WITH LOAD LINE SHAPING) FIG. 1a shows the primary of a conventional single-ended flyback converter circuit P1, driven by Q1, and a leakage inductance energy recovery winding and diode P2, D3. Snubber components D1, C1, and R1 are fitted from the collector to the emitter of Q1. FIG. 1b shows the voltage and current waveforms to be expected in this circuit. If load line shaping is required, then the main function of the snubber components is to provide an alternative path for the inductively maintained primary current IP as Q1 turns off. With these components fitted, it is now possible to turn off Q1 without a significant rise in its collector voltage during turn-off. (The actual voltage increase on the collector of Q1 during the turn-off edge depends on the magnitude of the diverted current Is, the value of the snubber capacitor C1, and the turn-off time t1 to t2 of Q1.) Without these components, the voltage on Q1 would be very large, defined by the effective primary leakage inductance and the turn-off di/dt. Because the snubber network also reduces the rate of change of collector voltage during turn-off, it reduces RFI problem.
3 OPERATING PRINCIPLES During the turn-off of Q1, under steady-state conditions, the action of the circuit is as follows. As Q1 turns off, starting at t1 ( FIG. 1b), the primary and leakage inductances of T1 will maintain a constant current IP in the transformer primary winding. This will cause the voltage on the collector of Q1 to rise (t1 to t2), and the primary current will be partly diverted into D1 and C1 (Is) (C1 being discharged at this time). Hence, as the current in Q1 falls, the inductance forces the difference current Is to flow via diode D1 into capacitor C1. If transistor Q1 turns off very quickly (the most favorable condition), then the rate of change of the collector voltage dVC/dt will be almost entirely defined by the original collector current IP and the value of C1. Hence… With Q1 off, the collector voltage will ramp up linearly (constant-current charge) until the flyback clamp voltage (2VDC) is reached at t3, when D3 will conduct. Shortly after this (the delay depends on the primary-to-secondary leakage inductance), the voltage in the output secondary winding will have risen to a value equal to that on the output capacitor C2. At this point, the flyback current will be commutated from the primary to the secondary circuit to build up at a rate controlled by the secondary leakage inductance and the external loop inductance through D2, C2 (t3 to t4). In practice Q1 will not turn off immediately; hence, if secondary breakdown is to be avoided, the choice of snubber components must be such that the voltage on the collector of Q1 does not exceed Vceo before the collector current has dropped to zero. FIG. 2a and b shows the relatively high edge dissipation and secondary break down load line stress when snubber components are not fitted. FIG. 2c and d shows the more benign turn-off waveforms obtained from the same circuit when optimum snubber values are fitted. (a) Turn-on and turn-off voltage, current, and dissipation stress without load line shaping. (b) Active load line imposed on "reverse base safe operating area" (RBSOA) limits without load line shaping. (Note secondary breakdown stress.)
4 ESTABLISHING SNUBBER COMPONENT VALUES BY EMPIRICAL METHODS Referring again to FIG. 1a, unless the turn-off time of Q1 is known (for the maximum collector current conditions and selected drive circuit configuration), the optimum choice for C1 will be an empirical one, based upon actual measurements of collector turn-off volt ages, currents, and time. The minimum value of C1 should be such as to provide a safe voltage margin between the Vceo rating of the transistor and the actual measured collector voltage at the instant the collector current reaches zero at t2. A margin of at least 30% should be provided to allow for component variations and temperature effects. The design of the drive circuit, collector current loading, and operating temperatures have considerable influence on the switching speed of Q1. A very large value of C1 should be avoided, since the energy stored in this capacitor at the end of the flyback action must be dissipated in R1 during the first part of the next "on" period of Q1. The value of R1 is a compromise selection. A low resistance results in high currents in Q1 during turn-on. This gives excessive turn-on dissipation. A very high resistance, on the other hand, will not provide sufficient discharge of C1 during a mini mum "on" period. Careful examination of the voltage and current waveforms on the collector of Q1, under dynamic loading conditions, is recommended. These should include initial turn-on at full load and maximum input voltage, wide and narrow pulse conditions, and output short circuit. The selection of R1 and C1 for this type of snubber network must always be a compromise. 5 ESTABLISHING SNUBBER COMPONENT VALUES BY CALCULATION FIG. 1b shows typical turn-off waveforms when the snubber network D1, C1, R1 shown in FIG. 1 is fitted. In this example, C1 was chosen such that the voltage on the collector Vce will be 70% of the Vceo rating of Q1 when the collector current has dropped to zero at time t2. Assuming that the primary inductance maintains the primary current constant during turn-off, and assuming a linear decay of collector current in Q1 from t1 to t2, the snubber current Is will increase linearly over the same period, as shown. It is assumed that the fall time of the collector current (t1 to t2) is known from the manufacturer's data or is measured under active drive conditions at maximum collector voltage and current. During the collector-current fall time of Q1 (t1 to t2), the current in C1 (Is) will be increasing linearly from zero to IP. Hence the mean current over this period will be IP/2. Provided that the maximum primary current IP and turn-off time t1 to t2 are known, the value of the optimum snubber capacitor C1 may be calculated as follows: (The ½ factor assumes a linear turn-off ramp on the collector current IC such that the mean current flowing into C1 is ½ the turn-off peak value during the turn-off period, as shown in FIG. 1b.) Hence, if the collector voltage is to be no more than 70% of Vceo when the collector current reaches zero at time t2, then... 6 TURN-OFF DISSIPATION IN TRANSISTOR Q1 By the same logic as used above (although the waveform is inverted), C1 and transistor Q1 both see the same mean current and voltage during the turn-off period. Hence, the dissipation in the transistor during the turn-off period t1 to t2 will be the same as the energy stored in C1 at the end of the turn-off period (t2).... 7 SNUBBER RESISTOR VALUES The snubber discharge resistor R1 is chosen to discharge the snubber capacitor C1 during the minimum selected "on" period. The minimum "on" period is given by the designed minimum load at maximum input voltage and operating frequency. The CR time constant should be less than 50% of the minimum "on" period to ensure that C1 is effectively discharged before the next "off" period. Hence ... 8 DISSIPATION IN SNUBBER RESISTOR The energy dissipated in the snubber resistor during each cycle is the same as the energy stored in C1 at the end of the turn-off period. However, the voltage across C1 depends on the type of converter circuit. With complete energy transfer, the voltage on C1 will equal the supply voltage Vcc, as all flyback voltages will have fallen to zero before the next "on" period. With continuous mode operation, the voltage will equal the supply voltage plus the reflected secondary voltage. Having established the voltage across C1 immediately before turn-on (VC), the dissipation in R1 (PR1) may be calculated as follows:... 9 MILLER CURRENT EFFECTS When measuring the turn-off current, the designer should consider the inevitable Miller current that will flow into the collector capacitance during turn-off. This effect is often neglected in discussions of high-voltage transistor action. It results in an apparent collector-current conduction, even when Q1 is fully turned off. Its magnitude depends on the rate of change of collector voltage (dVC/dt) and collector-to-base depletion capacitance. Further, if the switching transistor Q1 is mounted on a heat sink, there may be considerable capacitance between the collector of Q1 and the common line, providing an additional path for apparent collector current. This should not be confused with Miller current proper, as its magnitude can often be several times greater than the Miller current. These capacitive coupling effects result in an apparent collector current throughout the voltage turn-off edge, giving a plateau on the measured collector current. Hence, the measured current can never be zero as the collector voltage passes through Vceo. FIG. 2c shows the plateau current. This effect, although inevitable, is generally neglected in the published secondary breakdown characteristics for switching transistors. Maximum collector dVC/dt values are sometimes quoted, and this can be satisfied by a suitable selection of C1. When power FET switches are used, the maximum dV/dt values must be satisfied to prevent parasitic transistor action; hence, snubber networks must still be used in most high-voltage power FET applications. 10 THE WEAVING LOW-LOSS SNUBBER DIODE As shown previously, to reduce secondary breakdown stress during the turn-off of high voltage bipolar transistors, it is normal practice to use a snubber network. Unfortunately, in normal snubber circuits, a compromise choice must be made between a high-resistance snubber (to ensure a low turn-on current) and a low-resistance snubber (to prevent a race condition at light loads where narrow pulse widths require a low CR time constant). This conflict often results in a barely satisfactory compromise. The "Weaving snubber diode" provides an ideal solution. The circuit for this snubber arrangement is shown in FIG. 3. It operates as follows. Assume that transistor Q1 is on so that the collector voltage is low. Current will be flowing from the supply line through the transformer primary P1, and also from the auxiliary supply through resistor R2 and snubber diode D5 into the transistor collector.
At the end of the "on" period, Q1 will start to turn off. As the collector current falls, the transformer primary leakage inductance will cause the collector voltage to rise. However, when the collector voltage is equal to the auxiliary supply voltage, the primary current will be diverted into the snubber diode D5 (flowing in the reverse recovery direction in D5) and back into the auxiliary supply through D6. This reverse current flow in D5 will continue for its reverse recovery time. During this reverse recovery period, Q1 will continue to turn off, its collector current falling to zero, while the collector stress voltage remains clamped by D5 at a value only slightly above the auxiliary supply voltage. Consequently, Q1 turns off under negligible stress conditions. The reverse recovery time of the snubber diode must be longer than the turn-off time of transistor Q1. Special medium-speed soft recovery diodes are manufactured for this purpose (for example, Philips Type #BYX 30 SN). During the turn-off action, the recovered charge from the snubber diode D5 is stored in the auxiliary capacitor C1, to be used to polarize D5 during the next "on" period; consequently, very little turn-off energy is lost to the system. When Q1 turns on again, very little charge will be extracted from the cathode of D5 during the turn-on edge, because the diode depletion layer is wide and the capacitance low (the normal variable-capacitance behavior of the diode). Hence the turn-on stress of Q1 is not significantly increased. When Q1 is in its saturated "on" state, a current will flow from the auxiliary supply and capacitor C1 to reestablish the forward-bias condition of the snubber diode D5, part of this energy being the previous recovered charge. As soon as the snubber diode is conducting, it is conditioned for a further turn-off cycle. 11 "IDEAL " DRIVE CIRCUITS FOR HIGH-VOLTAGE BIPOLAR TRANSISTORS FIG. 4 shows a combination of the "snubber diode" and "Baker clamp" circuits, with a push-pull base drive to Q1. This arrangement is particularly suitable for high-voltage flyback converters where the collector voltage may be of the order of 800 V or more during the flyback period. It operates as follows. When the drive voltage goes high, Q2 is turned on and Q3 off. Current flows via R3, Q2, C2, and D7 to the base of the power transistor Q1. The overdrive provided by the low impedance R3, C2 network turns Q1 on rapidly. As Q1 turns on, the collector voltage falls. When this reaches 12 V (the auxiliary supply voltage), the snubber diode D5 will be forward-biased and current will flow via R2, D5 into the collector of Q1. Q1 continues to turn on, taking the collector voltage toward zero, until the Baker clamp voltage is reached. At this value D3 becomes forward-biased, diverting part of the base drive current into D3, D5, and the collector of Q1. At this point C2 will have charged to a voltage such that the drive current will be diverted via D1, D2, and L1 into the base of Q1. The voltage on the collector of Q2 will now be defined by the sum of the voltage drops across Q1 (Vbe), D1, D2, and Q2 (Vsat )-say, 2.5 V. The collector clamp voltage will be this value less the voltage drop across D3, D5-say, 1 V. This voltage can be increased by introducing more diodes in series with D1 and D2. (The voltage across L1 is negligible.) Hence, during the remainder of the "on" period, the main drive current path is via R3, Q2, D1, D2, L1, into the base emitter of Q1. Baker clamp action is provided by D3, D5. At the end of the "on" period, the drive voltage goes low, turning Q2 off and Q3 on and clamping the cathode of D4 to the -5-V bias line. Diodes D7, D1, and D2 will be reverse biased, and the turn-off current path is via D4 and L1. However, L1 was conducting current in the forward direction before turn-off and will continue to maintain this forward (but now decaying) current for the first part of the turn off action. Hence the turn-off current in L1 will decay to zero and then reverse via D4, providing the ideal turn-off current ramp specified for high-voltage transistors. Resistor R4 discharges C2 during the "off" period. When all carriers have been removed from the base-emitter junction of Q1, the junction will block, and the flyback action of L1 will force the base-emitter into reverse breakdown. The breakdown voltage (approximately -7.5 V) is less than the -5-V bias, and this breakdown action stops when the energy in L1 is dissipated. Note: Many high-voltage transistors are designed for this breakdown mode of operation during turn-off. At the same time, as Q1 turns off, the collector voltage will be rising toward the fly back voltage (800 V). However, when the collector voltage reaches 12 V (the auxiliary voltage), the snubber diode D5 will be reversed-biased, and the collector current will be diverted into D5, D6, and the auxiliary line. The reverse recovery time of D5 is longer than the turn-off time of Q1, and Q1 turns off under low-stress conditions with only 12 V on the collector. When Q1 has turned off and D5 blocks, the collector voltage will rise to the flyback value. The recovered charge of D5 is stored on C1 for the next forward drive pulse. Although this circuit does not provide proportional drive current in the conventional way, the Baker clamp adjusts the current into the base of the power device to suit the gain and collector current. Hence the action is similar to that of the proportional drive circuit except that the drive power needs are greater. In conclusion, this drive circuit combines most of the advantages of the proportional drive circuit, the snubber diode, and the Baker clamp. It also provides a correctly profiled drive current to give low stress and fast and efficient switching action in high-voltage, high power bipolar switching applications. 12 QUIZ 1. Explain what is meant by the term "snubber network." 2. Explain the two major functions of a typical snubber network. 3. Discuss the criteria for selecting snubber components, for a bipolar transistor with an inductive load, if secondary breakdown is to be avoided. 4. Why is a large snubber capacitor undesirable? 5. Describe a low-loss snubber technique that may be used in place of the conventional RC snubber network. 6. Using the snubber network shown in FIG. 1, calculate the minimum snubber capacitance required to prevent the collector voltage on Q1 exceeding 70% of Vceo during Q1 turn-off. (Assume that the fall time of Q1 is 0.5 µs, the collector current IP is 2 A, and the Vceo rating is 475 V.) Also see: Our other Switching Power Supply Guide |
PREV. | Next | Guide Index | HOME |