Shown here are the waveforms of a device being turned on into a shorted load. Because it is being driven into a shorted load the part enters power limitation immediately. If the power limitation was not there, the junction temperature would rise to thermal shutdown almost immediately within a few tens of microseconds. There is a 30ms slew of junction temperature rise essentially evening out the temperature across the die over time. Slewing the junction temperature in 30ms versus 10µs eliminates the huge temperature gradients across the die, which causes extreme stress. The Coffin-Manson Thermal Model states that thermal gradients greater than 60°C cause inelastic stress on the surface of the die. This looks like mud cracks across the metallization. Shown in the lower right corner of this slide is a zoom of a simulation showing what happens to the temperature of the die during a short. It starts at an ambient of 40°C then the part is turned on into a shorted load, the junction temperature in the hottest part of the die would rise to about 100°C and shut off. This is because the coolest part of the die is still at 40°C and there is a delta between the two of 60°C. Once it turns off it will cool down to about 60°C, turn back on, and continue to cycle until the junction temperature reaches the thermal shutdown, typically about 165°C. At that point the part will go into thermal shutdown and reduce the current limit value. The current is much lower so the stress is much less. The device will continue current limiting and shutting off at the lower current level until commanded off by the input. The Vcs trace just under the Vin trace at the top represents the current sense feedback pin voltage. The Vcs pin normally reflects the current in the output and when faulted is pulled high. In this instance the trace shows that the diagnostic during the fault is stable. The current sense circuit has a natural limitation. It requires a voltage on the measured output to be able to have a feedback voltage on the current sense pin. In a shorted load the output voltage is zero. That would normally mean that the current sense pin would be zero volts as well because of this limitation. When the device enters into power limitation or thermal shutdown the current sense pin is overridden and driven high, to something like 8V, by the fault circuitry. This is typically well above the microprocessor I/O voltage range, so this is another reason to put that protection resistor between the device and the microprocessor. The device stays in power limitation until the junction temperature reaches the thermal shutdown level, around 165°C. At that point the current goes from some high level down to about half or about a third and will then cycle on and off based on the thermals of the application. In other words, how much thermal capacitance there is in the circuit board, what the ambient temperature is, will determine what this duty cycle is and its frequency. When the fault is removed the part will cool down to a lower temperature threshold and the diagnostic flag will then drop back down. Power limitation is occurring during the in-rush phase. This is very handy when driving a lamp. Driving a very heavy duty motor, the power limitation may be hit at very high voltages, very cold temperatures, or things of that nature. In an automobile high voltages are achieved in cold temperatures. Motors have a higher stall current to cold temperature because the thermal coefficient to copper causes the motor resistance to go down with the temperature. There is a lower current limitation after thermal shutdown which reduces the stress in a permanent short circuit. It is advised to customers to mitigate shorted loads. If there is a fault, turn it off, and do not turn it back on until it gets fixed. That keeps the part from wearing out – because even after doing all of this, the part will still wear out if users continue to drive it into a short.