Main Content : This paper analyzes the commutation process of large rectifier devices and the causes of commutation overvoltage, and proposes solutions to reduce commutation overvoltage and commutation losses. Keywords : Commutation, recovery charge, di/dt, RC protection Introduction Megawatt-level rectifier devices are generally used for excitation of large synchronous generators, DC drives, and DC power supplies. Their output voltage ranges from several hundred volts to over one thousand volts, and their output current ranges from over one thousand amperes to several thousand amperes. For outputs below two thousand amperes, single-bridge operation can be used; for outputs above two thousand amperes, multi-bridge parallel operation is generally adopted. Due to the high operating voltage and large output current of the rectifier bridge, the commutation overvoltage and commutation losses caused by the reverse recovery of the rectifier tubes during commutation are particularly prominent. Improper handling will lead to large rectifier tube overvoltages and high temperatures, sometimes necessitating the method of increasing the rectifier tube voltage and current ratings to solve the problem, resulting in increased device costs and wasted resources. Key Issues in the Commutation Process In most cases, the load does not have high requirements for the DC output voltage ripple factor, and a three-phase bridge rectifier is sufficient. Moreover, the load circuit often has a significant inductive component, which itself provides good filtering. Therefore, the output of large rectifiers can generally eliminate the need for bulky and expensive filtering devices. If the load has stringent requirements for voltage ripple, due to the large output current, a series inductor is used for filtering. Therefore, if no measures are taken to address the commutation problem, the equivalent circuit of the system is shown in Figure 1. Looking at the three-phase voltage waveforms, the commutation process of the load current transitioning from phase A to phase B is shown in Figure 4. Before time Ta, because UAC is at its maximum, the load current IL = Ia, as shown in Figure 1, flowing from phase A through D1, the load, and D2 into phase C. From time Ta, UBC is at its maximum, thus commutation begins. After commutation is complete at time Tb, the load current I_L = I_b, as shown in Figure 2. It flows from phase B through D3, the load, and D2 into phase C. The first issue we need to analyze is the current change in the rectifier diodes between Ta and Tb. The current ID₁ of rectifier diode D1 decreases from time Ta, reaching 0 at time Tb. However, the commutation process is not yet complete, Uba is positive, and D1 bears reverse voltage. Since the rectifier diodes have a certain reverse recovery charge, D1 has not yet recovered to cutoff. Therefore, there must be a reverse recovery time Tr and a reverse recovery current Ir, as shown in Figure 3. Ir flows from phase B through Lr, the positive direction of D3, the reverse direction of D1, and back to phase A. Tr and Ir are important data for analyzing the reverse recovery characteristics of the rectifier diodes. Therefore, the second issue we need to analyze is estimating the magnitudes of Tr and Ir. Under certain operating voltage and current conditions, the values ​​of line reactance Lr, reverse recovery charge Qr of the rectifier tube, reverse recovery time Tr, and reverse recovery current Ir determine the magnitude of commutation overvoltage and commutation loss. Taking reasonable measures can suppress the magnitude of commutation overvoltage and reduce commutation loss, which is the third problem we want to study. Equivalent calculation of commutation process Lr is the equivalent series reactance of the rectifier transformer and the line. Due to its existence, commutation requires a certain time ΔT, that is, ΔT = Tb - Ta. During the time ΔT, Ia gradually decreases while Ib gradually increases. Both are positive currents, and Ia + Ib = IL. After time Tb, Lr not only limits the increase of the reverse recovery current Ir, but also generates commutation overvoltage when D1 recovers and cuts off. Of course, the line resistance Rr also exists, but during commutation, di/dt is relatively high, and the effect of Lr is far greater than that of Rr; therefore, Rr is negligible here. The equivalent circuit of the rectifier circuit between Ta and Tb is shown in Figure 5. Since ΔT is only a few hundred microseconds, it can be assumed that dv/dt remains constant within Ta-Tb. Therefore, the following simplified equation can be listed: The initial value of i1 is IL, the initial value of i2 is 0, and Uxm is the peak value of the three-phase line voltage. Now, we need to find the value of t when i1 = 0. This value of t is ΔT. The above equations can be combined into: For example, a rectifier device has a single-bridge output IL = 1000A, an operating frequency of 100Hz, ω = 628, a peak three-phase line voltage Uxm = 1000V, and an equivalent line inductance Lr = 50uH. Substituting into equation ①, we can obtain t = 5.6 × 10⁻⁴ S, or 560 μS. That is, starting from Ta (i₂ = 0, I₁ = IL), after 560 μS, i₂ = IL and I₁ = 0, reaching Tb. At this point, D1 begins reverse recovery. The speed and peak current of the rectifier's reverse recovery are determined by the reverse recovery charge Qr and the reverse di/dt; Qr depends on the characteristics of the rectifier and the magnitude of the forward current, while the reverse di/dt is determined by Uba and Lr. At time Tb: Ub_a = ωU_xmt = 352V. The reverse recovery current change rate di/dt = Uba/(2Lr) = 3.52A/µs. Since the reverse recovery process lasts only tens of microseconds, Uba is assumed to remain constant during this process, therefore di/dt also remains constant. If the rectifier recovers a charge of 2000µC (microcoulombs) under this condition, referring to Figure 6, the reverse recovery time Tr1 can be estimated to be approximately 19.4µs, and the peak reverse recovery current Irr approximately 69A. When the reverse peak current flows through the rectifier diode, the diode quickly returns to cutoff. The magnitude of di/dt at the point of return to cutoff is determined by the characteristics of the rectifier diode and its operating conditions. The maximum di/dt occurs at the initial cutoff point. Therefore, due to the presence of line inductance, a commutation overvoltage occurs. If the maximum di/dt at the point of return to cutoff under this condition is 12A/µs, then the overvoltage is Lr × di/dt = 1200V. Therefore, without taking measures, the highest voltage on the rectifier diode will be nearly 2200V. On the other hand, the reverse recovery peak current also causes additional commutation losses. During the process of the rectifier diode returning to cutoff, the energy stored in the line inductance by the reverse peak current will be consumed on the rectifier diode. This energy totals 0.5 × 2 × Lr × Irr² × 6f = 143W. Measures to Improve Commutation From the formation mechanism of commutation overvoltage and commutation loss, it is known that if a sufficiently large energy storage element clamps the overvoltage within a certain value during the rectifier diode's recovery from cutoff, it can reduce the commutation overvoltage. Simultaneously, if excess energy in the energy storage element (energy stored in the line inductance by the reverse peak current) is released to the load or the dissipating resistor, the commutation loss of the rectifier bridge can be reduced. Overvoltage protection methods suitable for large rectifier circuits include single-tube RC protection, full-bridge RC protection, and a combination of both, as shown in Figures 7 and 8. The two protection circuits operate on the same principle. Starting from the zero-crossing of the rectifier diode's forward current, the reverse recovery current I_r increases with a certain di/dt, reaching its peak value I_rr after passing through T_r1. This current also flows through the line inductance. When the rectifier diode recovers and is cut off, the reverse recovery current flows through the RC circuit. As long as the RC protection circuit is properly valued, the overvoltage can be controlled to a maximum of approximately I_rr × R. For example, if R is 5 ohms, when I_rr = 69A, the maximum overvoltage is 345V. In the single-diode protection method, although the capacity of each rectifier diode's RC circuit can be smaller than that of the full-bridge protection method, it obviously cannot be as small as one-sixth. In practical applications, it can be half the capacity of the full-bridge protection method. Therefore, the total capacity of the RC circuit in the single-diode protection method is three times larger. However, a drawback of the full-bridge protection method is that the distributed inductance energy in the internal connection section of the rectifier bridge is still consumed by the rectifier diodes and cannot be absorbed through the RC circuit. In some fast rectification applications (above 400Hz), if fast recovery rectifier diodes are used, the rectifier diodes have a fast recovery speed and a high number of commutation cycles, making the full-bridge protection method unsuitable. If it is not fast rectification, the number of commutation cycles is less, and the energy is relatively smaller, so the full-bridge protection method should be used to simplify the circuit, improve reliability, and reduce costs. When designing the RC protection circuit, the value of capacitor C should be determined first, followed by the value of resistor R. The basis for determining capacitor C is as follows: under maximum load, the reverse recovery peak current Irr is also the largest. Irr charges capacitor C through the line inductance, and the maximum voltage rise of capacitor C should be much less than the allowable overvoltage value. The value of resistor R is determined based on the allowable overvoltage, the reverse peak recovery current Irr of the rectifier diode, and the prevention of input power supply oscillation. A larger R value results in a higher commutation overvoltage, but also greater damping of input power supply oscillation, making it less prone to oscillation. Conversely, a smaller R value results in a lower commutation overvoltage, but less damping of input power supply oscillation, making it more prone to oscillation. With RC protection, the maximum allowable overvoltage should be lower than the overvoltage value when no measures are taken. In this case, because some of the energy stored in the line inductance of Irr is released to the load, some is charged to the capacitor, and some is consumed in the resistor, the energy consumed in the rectifier diode is very small, thus achieving the purpose of reducing the temperature rise of the rectifier diode and improving reliability. Referring to the previous example, under maximum load, the rectifier bridge has I[sub]rr[/sub] = 69A. To control the commutation overvoltage to within 300V, and to allow the capacitor C voltage to rise by 50V after one commutation, the minimum C value can be estimated using the energy method: 0.5C(10502-10002) = 0.5Lr692. Therefore, the minimum C should be 4.64uF. Taking C = 4.7uF, its withstand voltage should be above 1600V. At this time, the LC time constant TLC = 21.68uS, the critical value of the resistor R is L/TLC = 4.61Ω, and the maximum commutation overvoltage is 318V, which does not meet the requirements. Taking C = 6.8uF, then TLC = 26.1uS, the critical value of resistor R is 3.83Ω, and the maximum commutation overvoltage is 264V, which meets the requirements. In practical applications, C = 6.8uF/1600V and R = 3.9Ω/100W are used. The above are the calculated values ​​of RC protection using full-bridge RC protection. When using single-tube protection, the capacitor and resistor can be taken as half the value of full-bridge protection, i.e., C = 3.3uF/1600VDC and R = 2Ω/50W, which can achieve satisfactory results. It is worth noting that a non-inductive resistor and non-inductive capacitor should be used to form the RC protection circuit, and the distributed inductance of the circuit should be minimized during application; otherwise, the actual commutation overvoltage and commutation loss may be higher than the calculated values. Conclusion: 1. Large rectifier circuits will experience commutation overvoltage and commutation loss during commutation. 2. The magnitude of commutation overvoltage and commutation loss is related to the rectifier circuit's operating voltage, operating current, operating frequency, line inductance, and the reverse recovery characteristics of the rectifier diodes, and can be quantitatively calculated within a certain range. 3. Adopting single-diode RC protection or full-bridge RC protection can effectively reduce commutation overvoltage and commutation loss.