|
Freewheeling and snubber diodes Requirements for freewheeling and snubber diodes Modern fast switching devices require the use of fast diodes as freewheeling diodes. During each switch on process, the freewheeling diode switches from on to off state. This process requires the diode to have soft recovery characteristics. However, for a long time, the role of fast diodes was ignored, which limited the increase in the operating frequency of switching devices. In the past few years, it has received great attention, especially by improving its reverse recovery characteristics. Reverse blocking voltage and forward on-state voltage From the definition of the reverse blocking voltage VR, it can be known that the leakage current of the diode at this voltage value shall not be greater than the critical value IR , as shown in FIG11. 
Figure 11 Definition of diode reverse and forward voltage The values in the parameter table provided by the manufacturer are the values when the temperature is equal to 25℃. When the temperature becomes lower, the reverse blocking ability decreases. For example, for a 1200V diode, its drop rate is 1.5V/K. If it is operated below room temperature, this should be paid special attention to when designing the circuit. When the temperature is higher than room temperature, the reverse blocking voltage increases accordingly, but the leakage current also increases at the same time. Therefore, the leakage current value at high temperature (125℃ or 150℃) is usually given in the parameter table. The forward voltage V F indicates that under a given current, the voltage drop of the diode in the on state should be less than a certain critical value. Generally speaking, this value is measured at room temperature, but one of the main factors that determine system losses is the forward voltage at high temperature. Therefore, its dependence on temperature is given in all parameter tables. Enable Features When the diode turns on, the voltage first rises to V FRM , the repetitive forward peak voltage, and then drops to the forward on-state voltage. Figure 12 shows the commonly used definitions of V FRM and turn-on time t fr . 
Figure 12 Turn-on characteristics of a power diode But for the freewheeling and snubber diodes used in IGBTs, this definition does not explain much, because 1) The rate of rise of the turn-on current, di / dt , will be so high that the V FRM of a 1700V diode will reach 200-300V, which is more than 100 times the V F. 2) In actual application, the diode goes from cutoff to conduction, and the resulting V FRM is much higher than that from zero voltage to conduction state. For the snubber diode, since the snubber circuit can only play a role after the diode is turned on, a lower VFRM is one of its most important indicators. Even for freewheeling diodes with a reverse blocking voltage greater than 1200 V, repeatable forward peak voltage plays an important role. When the IGBT is turned off, the parasitic inductance of the line will induce a voltage spike, which is superimposed on the VFRM of the freewheeling diode. The sum of the two may cause overvoltage. Shutdown Characteristics In the process of a diode changing from on to off, the electricity stored inside it must be released. This process causes the current of the diode to flow in the reverse direction. The waveform of this reverse current can be described by the reverse recovery characteristic. FIG13 shows a simplest measurement circuit, where S represents an ideal switch, IL is a current source, Vk is a voltage source for commutation, and Lk is the inductance in the commutation circuit. 
Figure 13 shows a simple measurement circuit When the switch S is turned on, the current and voltage curves of a soft recovery diode are shown in FIG14 . 
Figure 14 Current and voltage characteristics of the reverse recovery process of the soft recovery diode The commutation speed di / dt is determined by the voltage and inductance, that is,  (7) At t0 , the current reaches zero. At tw , the diode begins to bear the reverse voltage. At this moment, all the carriers in the pn junction of the diode are cleared. At tirm , the reverse current reaches the maximum value I RRM . After tirm , the current gradually decays to its leakage current value. Its trajectory is completely determined by the diode. If the decay process is very steep, it is called rigid recovery characteristics; if the decay process is very slow, it is called soft recovery characteristics. The reverse recovery time is defined as t rr , which starts from t 0 and ends when the current decays to 20% of I RRM . As shown in Figure 14, by subdividing t rr into t f and t s , a coefficient used to qualitatively describe the reverse recovery characteristics of the diode can be obtained, namely the softness coefficient s =  (8) Figure 15 shows a quasi-practical measurement circuit. 
Figure 15 Quasi-practical chopper circuit for measuring reverse recovery characteristics in a buck converter (double-pulse operation) The commutation speed di/dt can be adjusted by the gate resistance of the switching device. Vk is the DC bus voltage, and there is a parasitic inductance on the wire between the capacitor, IGBT and diode. Figure 16 shows the driving signal of the IGBT and the current waveforms of the IGBT and diode when double pulses are applied. When the IGBT is turned off, the load current is switched from the IGBT to the diode, thereby showing the recovery characteristics of the diode during this period. When the IGBT is turned on, the IGBT also continues the reverse recovery current of the freewheeling diode. Figure 17 shows this process with a higher time resolution. Figure 17 (a) shows the current and voltage waveforms of the IGBT and the losses during the turn-on process; Figure 17 (b) shows the current and voltage waveforms and losses of the diode. 
Figure 16 Driving signal and current waveforms of IGBT and freewheeling diode under double pulse operation conditions (see Figure 15 for the circuit) 
(a) IGBT turn-on process 
(b) Turn-off process of the freewheeling diode Figure 17 Current, voltage and power loss of the circuit shown in Figure 15 
Figure 18 Relationship between commutation peak voltage and forward current of diodes with different switching characteristics When the IGBT receives the reverse peak current of the freewheeling diode, its voltage is still at the level of the DC bus voltage (1200V in Figure 17 (a)). At this moment, the turn-on loss of the IGBT is at its maximum. The reverse recovery characteristics of the diode can be further divided into two parts. 1 ) The first part is the stage when the current rises to the peak value of the reverse recovery current and the subsequent decline process at the rate of di / dt . For a soft recovery diode, the values of dir / dt and di /dt are roughly the same, and the peak value of the reverse recovery current I RRM has the greatest impact on the switching device. 2) The second part is the tail current part, that is, the process in which the reverse recovery current slowly decays to zero. In this process, t rr no longer has obvious significance. Because there is voltage on the diode at this time, the main part of the loss in the diode is generated in the tail process. For a rigid diode without tail current, although its switching loss is very low, it cannot be used in practice. For IGBT, because its voltage has dropped to a very low level in the tail stage, the tail current has little effect on the loss of IGBT. In practice, the diode losses are much lower than the IGBT switching losses (Figure 17(b) shows the diode losses on the same scale as the IGBT losses in Figure 17(a)). Therefore, if the sum of the IGBT and diode losses is to be kept small, the peak value of the reverse recovery current should be minimized while retaining most of the stored charge until the tail phase before releasing it. The realization of this design concept is limited by the maximum switching losses that the diode can dissipate. Therefore, the most important parameter for a diode's contribution to overall losses is its peak reverse recovery current, I RRM , which should be as small as possible. Let's look at a typical power electronic circuit, for example, a DC chopper placed in a module. Its parasitic inductance L σges is about 40nH, which plays a role in reducing overvoltage. Because ideal switches do not exist, the voltage of the IGBT will drop during the reverse recovery of the diode. The actual measured voltage value is - V ( t ) = - V k - L σges + V CE ( t ) (9) Where: V CE ( t ) is the instantaneous value of the voltage applied to the IGBT. For a typical soft recovery diode, when the current rise rate is not too high (≤1500A/μs) and the parasitic inductance is minimal, the voltage v ( t ) is less than V k at any time and there is no voltage spike. Figure 18 shows an example of using this method to describe the recovery characteristics. Under the conditions shown in Figure 18, let's compare the overvoltage of two diodes. One of them is a diode in which the carrier lifetime is adjusted by a platinum diffusion process to obtain soft recovery characteristics by reducing the efficiency of the p-emitter; the other is a CAL diode. At the rated current (75A), the platinum-diffused diode has the same soft characteristics as the CAL diode. However, at lower currents, due to the excessively rigid switching characteristics of the former, an overvoltage is generated, the maximum value of which may be greater than 100V at 10% of the rated current. At even lower currents, the overvoltage is reduced again due to the slower switching of the applied IGBT. The CAL diode does not show a significant overvoltage in all these cases. Requirements for freewheeling diodes in rectification and inversion operation In a converter using IGBTs or MOSFETs, the requirements for the freewheeling diode depend on whether it is operating in rectification or inversion. Even when delivering the same power, the losses in the two operating states are different. The characteristic of inverter operation is that energy flows from the DC voltage bus terminal to the AC terminal. In other words, the AC terminal is connected to a user and supplies power to it (for example, a three-phase AC motor). In the rectifier operation state, energy flows from the AC end to the DC voltage bus end. In this case, the converter works as a chopper rectifier at the grid end or generator end. Under the condition of transmitting equal power, the different losses in the power semiconductor are mainly determined by the phase between the AC terminal voltage and the fundamental current during rectification and inversion operation. This can be further illustrated by the basic circuit shown in Figure 19. 
(a) Basic circuit 
(b) Related waveform Figure 19 Basic circuit of one phase of inverter using IGBT and freewheeling diode We can see that: 1) If V out is positive and i L > 0, current flows through S 1 ; 2) If V out is negative and i L > 0, current flows through D 2 ; 3) If V out is positive and i L < 0, current flows through D 1 ; 4) If V out is negative and i L < 0 current flows through S 2 . Therefore, for a given effective value of the current, the conduction losses occurring in the IGBT and the freewheeling diode are determined by the power factor between the voltage and current fundamentals and the modulation m of the converter (which determines the duty cycle). In inverter operation, there is a relationship of 0(<=) m cosφ(<=)1. If m cosφ=1, the power semiconductor loss reaches its limit. Under this condition, the conduction loss and the total loss of the IGBT reach the maximum value, and the diode loss reaches the minimum value. In the rectifier operation, there is a relationship of 0 (>=) m cosφ (>=) -1. When m cosφ = -1, the loss of the power semiconductor reaches its limit. Under this condition, the conduction loss and the total loss of the IGBT reach the minimum value, and the loss of the diode reaches the maximum value. Applying this theory to Figure 19, the situation occurs when the chopper rectifier absorbs only pure active power from the grid (in terms of the current fundamental). At this time, the star midpoint of the grid should be connected to the midpoint of the DC bus voltage. Figure 20 plots the above relationship. 
(a) Conduction loss 
(b) Switching loss Figure 20 Switching and conduction losses of IGBT and freewheeling diode in voltage source inverter When the DC bus voltage and AC current effective value are given, the switching loss of the device is only related to the switching frequency, and there is a linear relationship between the two. A large number of IGBT and MOSFET modules with freewheeling diodes on the market are designed for inverter operation (e.g. cosφ=0.6～1) in terms of the losses that can be dissipated at rated current. Since the conduction loss and total loss of the diode in this operating state are much lower than those of the IGBT, the design value of the diode loss is also much lower than that of the IGBT [IGBT/diode loss design ratio is about (2～3):1]. Therefore, when designing a chopper rectifier, if its power is equal to that of the corresponding chopper inverter, it is recommended to use a power module with a higher current rating. For example, the power flow of a transmission system is grid (400V/50Hz) → chopper rectifier ( fs = 10~12kHz ) → DC bus → chopper → inverter ( fs = 10~12kHz) → three-phase AC motor (400V/50Hz/22kW), then 1) The chopper rectifier uses a standard IGBT half-bridge module of 1200V/100A ( Tc = 80℃) ; 2 ) The chopper inverter uses a standard IGBT half-bridge module of 1200V/75A ( Tc = 80℃). If the power module itself has enhanced diodes, this distinction is unnecessary. Construction of fast power diodes We need to distinguish between two main forms of diodes, the Schottky diode and the pin diode. In a Schottky diode, the contact surface between the metal and the semiconductor forms a blocking pn junction. Unlike a pin diode, the pn junction has no barrier formed by diffusion. Therefore, if the n-region is very thin, its on-state voltage drop is smaller than that of any pin diode. In the transition process from the on state to the off state, in theory, only the space charge region needs to be charged. Therefore, this type of diode is suitable for very high frequencies (>100kHz). However, this advantage is limited to when the voltage is less than about 100V (currently the maximum can reach 250V). Therefore, Schottky diodes are suitable for use as freewheeling diodes for MOSFETs. On the other hand, when the designed withstand voltage is higher, then 1) The on-state voltage increases rapidly due to the increase in base width WB and the presence of only one type of carrier (unipolar type); 2) The cut-off leakage current increases rapidly, which may cause an imbalance in temperature rise. Therefore, when the voltage is greater than 100V, the pin diode begins to show its superiority. For the diodes currently produced, its middle part is no longer i (intrinsic), but an n-type semiconductor with a much lower concentration than the edge area. In the pin diode using epitaxial growth technology [Figure 21 (b)], an n - region is first shunted off on a high-concentration n+ substrate (epitaxial growth), and then the p region is diffused. With this method, the base width W B can be adjusted to a very low level, up to several μm ; at the same time, the silicon wafer has sufficient thickness to make the yield rate in production very high. By introducing the method of recombination center (mostly using gold diffusion process), a very fast diode can be realized, and at the same time, because its W B is very small, the on-state voltage can still be very low. Of course, the on-state voltage is always greater than the diffusion barrier of the pn junction (0.6~0.8V). The main application range of epitaxial growth diodes is between 100 and 600V. Some manufacturers have also realized epitaxial growth diodes with a withstand voltage of 1200V. From 600V upwards, the n - region is already wide enough to allow the use of diffusion technology to produce pin diodes (Figure 21 (c)). The p and n + regions are diffused into an n - substrate separately . Similarly, in order to adjust the dynamic characteristics of the freewheeling diode, it is necessary to introduce recombination centers. 
(a) Schottky diode 
(b) Epitaxial growth diode 
(c) Diffused diode Figure 21 Schematic diagram of the structure and concentration profile of the diode Series and parallel connection of fast power diodes Series The series diode circuit is shown in Figure 22. When connected in series, attention should be paid to the symmetrical distribution of the static cut-off voltage and the dynamic cut-off voltage. 
Figure 22 RC circuit for fast diode series connection In static state, due to the manufacturing deviation of the cut-off leakage current of each diode in series, the diode with the smallest leakage current bears the highest voltage and even reaches the holding state. However, as long as the diode has sufficient holding stability, it is not necessary to use parallel voltage-equalizing resistors. Only when diodes with a cut-off voltage > 1200V are connected in series, it is necessary to add parallel voltage-equalizing resistors. Assuming that the cut-off leakage current does not change with voltage and ignoring the resistance error, for a series circuit of n diodes with a given cut-off voltage Vr , we can obtain the simplified formula (10) for calculating the parallel resistance. R <  (10) Where: Vm is the maximum voltage in the series circuit ; ΔIr is the maximum deviation in diode leakage current when operating at maximum temperature. Make a sufficiently safe assumption that Δ I r =0.85 I rm（11） Where: Irm is given by the manufacturer . Using the above estimate, the current in the resistor is approximately 6 times the diode leakage current. Experience has shown that a resistor value is sufficient when the current through it is approximately three times the diode leakage current at the maximum cut-off voltage. Even under this condition, however, considerable losses still occur in the resistor. The dynamic voltage distribution is different from the static voltage distribution. If the carriers of one diode pn junction disappear faster than the other, it will be subjected to the voltage earlier. If the capacitance deviation is ignored, when n diodes with a given cutoff voltage Vr are connected in series, we can use the simplified formula (12) to calculate the parallel capacitance. C >  (12) Where: Δ Q RR is the maximum deviation of the diode storage capacity. Make a sufficiently safe assumption that Δ Q RR =0.3 Q RR（13） The condition is that all diodes are from the same manufacturing batch. Δ Q RR is given by the semiconductor manufacturer. In addition to the stored energy when the freewheeling diode is turned off, the energy stored in the capacitor also needs to be continued by the IGBT being turned on. According to the above design formula, we find that the total stored energy value may reach twice the storage energy of a single diode. Generally speaking, the series circuit of the freewheeling diode is not common because of the following additional loss sources: 1) n-diffusion voltage of pn junction; 2) Losses in parallel resistors; 3) Additional storage power needs to be connected by IGBT; 4) Increase in components due to RC circuits. Therefore, when a diode with a high cut-off voltage can be used, the series connection scheme is generally not used. The only exception is when the application circuit requires very short switching time and very low storage power, these two points happen to be possessed by low withstand voltage diodes. Of course, at this time the conduction loss of the system will also be greatly increased. in parallel No additional RC snubber circuit is required for parallel connection. What is important is that the deviation of the on-state voltage should be as small as possible when paralleling. An important parameter for determining whether a diode is suitable for parallel connection is the dependence of its on-state voltage on temperature. If the on-state voltage decreases with increasing temperature, it has a negative temperature coefficient, which is an advantage for losses. If the on-state voltage increases with increasing temperature, the temperature coefficient is positive, which is an advantage in typical parallel applications because the hotter diode will withstand a lower current, thereby maintaining system stability. Because diodes always have certain manufacturing deviations, a large negative temperature coefficient (>2mV/K) may cause a risk of temperature rise imbalance when diodes are connected in parallel. The parallel diodes will produce thermal coupling, the path is: 1) Through the substrate in a module with multiple chips in parallel; 2) Through the heat sink when multiple modules are connected in parallel to a heat sink. The temperature dependence of the on-state voltage of different types of diodes is shown in FIG23 . 
(a) When the temperature coefficient is negative 
(b) When the temperature coefficient is positive above the rated current Figure 23 Dependence of on-state voltage of different types of diodes on temperature Generally, for weaker negative temperature coefficients, this type of thermal coupling is sufficient to prevent the diode with the lowest on-state voltage from going into temperature rise imbalance. However, for diodes with negative temperature coefficients greater than 2mV/K, we recommend using them at a reduced rating, that is, the total rated current should be less than the sum of the rated currents of each diode.
| 
提升卡
变色卡
千斤顶
京公网安备 11010802033920号
