0 Introduction Currently, power modules are developing towards integration, intelligence, and modularity. Power modules provide an ideal interface for connecting low-voltage and high-voltage circuits in mechatronic equipment. Under any operating condition, power modules need to be protected to avoid being subjected to unacceptable current stress, that is, to prevent the power module from operating outside its given safe operating area. Operating outside the safe operating area will damage the power module and shorten its lifespan. In severe cases, it can even lead to immediate failure of the power module. Therefore, it is crucial to first detect critical current states and faults, and then respond appropriately to them. This article mainly focuses on overcurrent protection for IGBTs, but the analogy can be applied to power MOSFETs. 1 Types of Fault Currents Fault current refers to the collector or drain current exceeding the safe operating area. It can be caused by faulty control or load. Fault currents can damage power semiconductors through the following mechanisms: 1) Thermal damage due to high power loss; 2) Dynamic avalanche breakdown; 3) Static or dynamic latch-up effect; 4) Overvoltage caused by overcurrent. Fault currents can be further divided into overcurrent, short-circuit current, and ground fault current. 1.1 Overcurrent Characteristics: 1) Low collector current di/dt (depending on load inductance and drive voltage); 2) Fault current forms a loop through the DC bus; 3) Power module has not left the saturation region. Causes: 1) Reduced load impedance; 2) Inverter control error. 1.2 Short-Circuit Current Characteristics: 1) Collector current rises sharply; 2) Fault current forms a loop through the DC bus; 3) Power module leaves the saturation region. Causes: 1) Bridge arm shoot-through short circuit (Case 1 in Figure 1) – caused by power module failure; – caused by incorrect drive signal. 2) Load short-circuit current (Case 2 in Figure 1) – caused by insulation failure; – caused by human error (e.g., incorrect wiring). 1.3 Ground Fault Current (Case 3 in Figure 1). Characteristics: 1) The rise rate of the collector current depends on the grounding inductance and the voltage acting on the loop; 2) The ground fault current does not form a closed loop through the DC bus; 3) Whether the power module leaves the saturation region depends on the magnitude of the fault current. Cause: Due to insulation failure or human error, a connection exists between the live conductor and the ground potential. 2 Characteristics of ICBTs and MOSFETs under Overload and Short Circuit 2.1 Overcurrent In principle, the switching and conduction characteristics of the device under overcurrent are no different from those under rated conditions. Since a large load current will cause higher losses within the power module, the overload range of the power module should be limited to avoid exceeding the maximum allowable junction temperature. Here, not only the absolute value of the junction temperature under overload, but also the temperature variation range under overload are limiting factors. Specific limits for several ICBTs and MOSFETs are given by the safe operating area of a typical power module shown in Figure 2. 2.2 Short Circuit In principle, ICBTs and MOSFETs are safe short-circuit devices. That is, they can withstand a short circuit under certain external conditions and then be turned off without damage to the device. When examining short circuits (using IGBTs as an example), it's crucial to distinguish between the following two scenarios: 1) Short Circuit I: Short Circuit I occurs when the power module is switched on within a short-circuited load loop. In other words, the DC bus voltage under normal conditions drops across the power module. The rate of increase of the short-circuit current is determined by the drive parameters (drive voltage, gate resistance). Due to the parasitic inductance in the short-circuit loop, this current change generates a voltage drop, manifested as a sharp drop in the collector-emitter voltage characteristic, as shown in Figure 3. The steady-state short-circuit current value is determined by the power module's output characteristics. For IGBTs, the typical value can reach 8 to 10 times the rated current. 2) Short Circuit II: In this scenario, the power module is already in a conducting state before the short circuit occurs. Compared to Short Circuit II, the impact on the power module is far greater. To explain this process, Figure 4 shows the equivalent circuit diagram of Short Circuit II and its qualitative characteristic curves. Once a short circuit occurs, the collector current rises rapidly, the rate of increase determined by the DC bus voltage VDC and the inductance in the short-circuit loop. During time period 1, the IGBT exits the saturation region. The rapid change in collector-emitter voltage generates a displacement current through the gate-collector capacitance, which in turn causes the gate-emitter voltage to rise, resulting in a dynamic short-circuit peak current IC/SCM. After the IGBT has completely exited the saturation region, the short-circuit current tends to its steady-state value (time period 2). During this period, the parasitic inductance of the loop will induce a voltage, which manifests as an overvoltage of the IGBT. After the short-circuit current stabilizes (time period 3), the short-circuit current is turned off. At this time, the inductance Lx in the commutation loop will induce another overvoltage on the IGBT (time period 4). The overvoltage induced by the IGBT during a short circuit may be several times that during normal operation, as shown in Figure 5. To ensure safe operation, the following critical conditions must be met: 1) Short circuits must be detected and shut off within 10 μs; 2) The time interval between two short circuits must be at least 1 s; 3) The number of short circuits during the total operating time of the IGBT must not exceed 1000. Both short circuit I and short circuit II will cause losses in the power module, resulting in a rise in junction temperature. Here, the positive temperature coefficient of the collector-emitter voltage has an advantage (also applicable to drain-source voltage), which reduces the collector current during steady-state short circuits, as shown in Figure 6. 3 Fault Detection and Protection Fault currents in the inverter can be detected at different nodes, and the responses to detected fault currents may vary. Fast protection will be discussed here, assuming the fault current is detected inside the power module and the power module is directly shut down by the driver. The total response time of the power module may be only tens of ns. If the fault current detection is located outside the power module, the fault current signal is first sent to the inverter's control board, from where it triggers the fault response procedure; this process is called slow protection. This process can even be handled by the inverter's control and regulation system (e.g., the system's response to overload). 3.1 Fault Current Detection Figure 7 shows a voltage-source inverter circuit. The test points where fault current can be detected are marked. Fault current detection can be divided as follows: 1) Overcurrent can be detected at points ① to ⑦; 2) Bridge arm shoot-through short circuit can be detected at points ① to ④ and ⑥ to ⑦; 3) Load short circuit can be detected at points ① to ⑦; 4) Ground short circuit can be detected at points ①, ③, ⑤, and ⑥, or obtained by calculating the difference between the currents at points ① and ②. In principle, controlling short-circuit current requires rapid protection measures to achieve direct control at the output of the drive circuit, because the power module must shut down within 10 μs after a short circuit occurs. Therefore, fault current can be detected at points ③, ④, ⑥, and ⑦. Measurements at points ① to ⑤ can be achieved by measuring the shunt or inductive current converter. 3.1.1 Shunt for Measurement 1) Simple measurement method; 2) Requires a power shunt with low resistance (10~100mΩ) and low inductance; 3) The measurement signal is highly sensitive to interference; 4) The measurement signal does not have potential isolation. 3.1.2 Current Transformer for Measurement 1) Much more complex than a shunt; 2) Compared with a shunt, the measurement signal is less susceptible to interference; 3) The measured value has been isolated. At test points ⑥ and ⑦, fault current detection can be performed directly at the terminals of the IGBT or MOSIEET. Here, the protection method can be vCEsat or vDS(os) detection (indirect measurement), or mirror current gun measurement. The latter uses a sensor to detect a small portion of the IGBT unit to reflect the main current (direct measurement). Figure 8 shows the schematic diagram. 3.1.3 Detecting Current with a Mirror ICBT In a mirror IGBT, a small portion of the ICBT unit is combined with an emitter resistor for detection and connected in parallel to the current arm of the main IGBT. Once the conducting collector current passes through the measuring resistor, its information can be obtained. When Rsense=0, the current ratio between the two emitters is equal to the ideal value, which is the ratio of the number of mirror IGBT cells to the total number of cells. If Rsense increases, the current conducting in the measuring circuit will decrease due to the feedback of the measuring signal. Therefore, the resistor Rsense should be controlled within the range of 1 to 5Ω to obtain sufficiently accurate collector current measurement results. If the current threshold used for turn-off is only slightly greater than the rated current of the power module, then during IGBT turn-on, current sensing must be turned off due to the effect of the reverse recovery current peak of the reverse freewheeling diode (in hard-switching circuits). If the sensing resistor tends to infinity (Rsense→∞), its measured voltage is equal to the collector-emitter saturation voltage. Therefore, mirror current sensing is converted into vCEsat sensing. 3.2 Reduction of Fault Current By reducing or limiting high fault currents, especially in the case of short circuits and low-impedance short circuits to ground, the power module can be better protected. As shown in Figure 1, under short-circuit II, a high dvCE/dt causes a rise in the gate-emitter voltage, resulting in a dynamic short-circuit overcurrent. The amplitude of the short-circuit current can be reduced by clamping the gate-emitter voltage. Besides limiting the dynamic short-circuit overcurrent, the steady-state short-circuit current can also be reduced by decreasing the gate-emitter voltage. This method reduces the power module's losses during short circuits, and the overvoltage is also reduced due to the lower short-circuit current requiring shutdown. The principle is shown in Figure 9. This protection technique can limit the steady-state short-circuit current of the surge-resistant power module to approximately three times the rated current. 4 Conclusion With the development of power electronics technology, the application of power modules such as IGBTs and MOS-FETs is becoming increasingly widespread. For their safe and efficient operation, overcurrent protection measures must be considered for power modules. First, overcurrent faults should be detected in the shortest possible time, and then appropriate measures should be taken to protect the power module. Sometimes, immediately shutting down the power module when an overcurrent occurs is not the optimal approach. A very simple protection method for dynamic gate control is to slow down the turn-off process by reducing the gate-emitter voltage in the event of overcurrent or short circuit in the IGBT and MOSFET. This is the "soft" turn-off process of the power module.