Power management subsystems control the distribution of power in an electronic system. These subsystems consist of integrated circuits and power semiconductors that handle power levels that can range from microwatts to megawatts.
Power semiconductors employed in power management systems include power switches, and rectifiers (diodes). Power switches include MOSFETs, IGBTs, and BJTs (bipolar junction transistors). MOSFETs, IGBTs and BJTs are found in two different forms:
- Discrete Power Semiconductors- These devices are only a single type housed in a single package.
- Integrated Power Semiconductors- Integrated with other circuits in a single package, they may be housed in a multi-chip module, or MCM package, that is, interconnected with other devices in the same package.
Detailed descriptions of the key categories of power semicconductors follow.
Power MOSFETs (Metal-Oxide Semiconductor Field Effect Transistors) are three-terminal silicon devices that function by applying a signal to the gate that controls current conduction between source and drain. Their current conduction capabilities are up to several tens of amperes, with breakdown voltage ratings (BVDSS) of 10V to over 1000V. The power MOSFET in Fig. 1 is housed in a TSOP-6 package.
The insulated gate bipolar transistor (IGBT) is a three-terminal power semiconductor noted for high efficiency and modest switching speeds. It switches electric power in many modern appliances: electric cars, variable speed refrigerators, air-conditioners. IGBTs are usually only discrete devices, or may have an integrated diode. Fig. 2 shows IGBTs intended for motor drives and consumer products.
SiC (silicon carbide) power semiconductors can theoretically reduce on-resistance to two orders of magnitude compared with existing Si devices. The use of SiC device is expected to reduce power loss extensively, when applied to power conversion systems. SiC devices as well as power MOSFETs or IGBTs may be used with rectifier devices such as Schottky barrier diode (SBD). SiC-SBD have been introduced, but SiC power MOSFETs has been difficult to manufacture because usable SiC material has been difficult to produce. However, SiC power transistors are now available. The SiC transistors on the wafer in Fig. 3 are rated at 1200V and 1700V.
Gallium nitride (GaN) is grown on top of a silicon substrate. The end result is a fundamentally simple, cost effective solution for power switching. This device behaves similarly to Silicon MOSFETs with some exceptions. Fig. 4 is a fully functional eighth brick converter demo board that uses the 100 V EPC2001 eGaN FETs and the compatible LM5113 100V half-bridge gate driver.
GaN transistors behave in a similar manner to silicon Power MOSFETs. A positive bias on the gate relative to the source causes the device to turn on. When the bias is removed from the gate, the electrons under it are dispersed into the GaN, recreating the depletion region, and once again, giving it the capability to block voltage. Among GaN’s features are:
- GaN offers superior performance compared with both silicon and silicon carbide.
- Device-grade gallium nitride can be grown on top of silicon wafers.
- GaN-on-silicon offers the advantage of self-isolation and therefore efficient monolithic power integrated circuits can be fabricated economically
- Enhancement-mode (normally off) and depletion mode (normally on) GaN devices are available.
Ideal Power Semiconductor Switch
The individual power semiconductor switch (Fig. 5) applies power to a load when a control signal tells it to do so. The control signal also tells it to turn off. Ideally, the power semiconductor switch should turn on and off in zero time. It should have an infinite impedance when turned off so zero current flows to the load. It should also have zero impedance when turned on so that the on-state voltage drop is zero. Another idealistic characteristic would be that the switch input consumes zero power when the control signal is applied. However, these idealistic characteristics are unachievable with the present state-of-the-art.
Real World Power Semiconductors
In the real world, actual power semiconductors do not meet the ideal switching characteristics. For example, Fig. 6(a) shows a control signal applied to an ideal power semiconductor switch whose output exhibits zero transition time when turning on and off as shown in Fig. 6(b). When the transistor is off (not conducting current) power dissipation is very low because current is very low. When the transistor is on (conducting maximum current), power dissipation is low because the conducting resistance is low. In contrast, an actual power switch exhibits some delay when turning on and off, as shown in Fig. 6(c). Therefore, some power dissipation occurs when the switch goes through the linear region between on and off. This means that the most power dissipation depends on the time spent going from the off to on and vice versa, that is, going through the linear region. The faster the device goes through the linear region, the lower the power dissipation and losses.
Power Semiconductor Reliability
Excessive operating voltage can cause power semiconductor failures because the devices may have small spacing between their internal elements. An even worse condition for a power semiconductor is to have high voltage and high current present simultaneously. A few nanoseconds at an excessive voltage or excessive current can cause a failure. Most power semiconductor data sheets specify the maximum voltage that can be applied under all conditions. The military has shown very clearly that operating semiconductors at 20% below their voltage rating provides a substantial improvement in their reliability.
Power Semiconductor Failures
Another common killer of power semiconductors is heat. Not only does high temperature destroy devices, but even operation at elevated, non-destructive temperatures can degrade useful life. Data sheets specify a maximum junction temperature, which is typically between 100°C and 200oC for silicon. Most power transistors have a maximum junction rating of 125°C to 150°C, the safe operating temperature is much lower.
Power semiconductors can be destroyed by very short pulses of energy. A major source of destructive transients is caused by turning on or off an inductive load. Protection against these problems involves a careful combination of operating voltage and current margins and protective devices.
Power dv/dt And di/dt
The terms dv/dt and di/dt reflect a time rate of change of voltage (dv/dt) or current (di/dt) describe their reaction to turning on or off a reactive load. These problems can occur in power semiconductor switches, because all sections of the device do not behave in an identical manner when subjected to very high rates of change. It is not only important to look at the dv/dt and di/dt values generated within a circuit, but also turn-on and turn-off times as well.
Switching power on and off at a rapid rate can cause electromagnetic interference (EMI) that can affect nearby electronic systems. Domestic and international standards define the amount of EMI that can be emitted.
Unclamped Inductive Switching (UIS)
Whenever current through an inductance is turned off quickly, the resulting magnetic field induces a counter electromagnetic force (CEMF) that can build up surprisingly high potentials across the switch. With transistor switches, the full buildup of this induced potential may far exceed the rated voltage breakdown of the transistor, resulting in catastrophic failure.
There are two failure modes when subjecting a MOSFET to UIS. These failure mechanisms are considered as either active or passive. The active mode results when the avalanche current forces the MOSFET’s parasitic bipolar transistor into conduction. In the passive mode the instantaneous device temperature reaches a critical value. At this elevated temperature, the MOSFET’s parasitic bipolar transistor causes catastrophic thermal runaway. In both cases the MOSFET is destroyed.
As a semiconductor chip gets larger its cost grows exponentially. And, there is the cost of the package that houses the integrated power device and the cost of interconnections. In deciding whether to integrate a power semiconductor into an integrated circuit or use two separate devices, look at the die size of each. If an integrated power semiconductor and a discrete power semiconductor have large die, the die cost dominates the overall cost, it would be cheaper to use two parts.
Integrated power semiconductors make sense when the die sizes are moderate, or there are multiple outputs. This is so because the package and handling costs offset the increased silicon cost. A major impact on cost is the number of good devices that can be obtained from silicon wafer, usually called yield. Not only does a larger die size mean a disproportionately larger cost, but key parameters may not be the same for all functions of each device on the die.
Power semiconductor on-resistance is important because it determines the power loss and heating of the power semiconductor. The lower the on-resistance, the lower the device power loss and the cooler it will operate. Low on-resistance drastically reduces heat-sinking needs in many applications, which lowers parts count and assembly costs.
Maximum junction temperature, TJ(max), is a function of the electrical characteristics of the device itself, as well as the package employed. Package thermal properties determine its ability to extract heat from the die. The junction-to-ambient and junction-to-case thermal resistance is a measure of the power semiconductor’s ability to extract heat. Data sheets rate thermal resistance in terms of either °C/W or K/W. The lower the thermal resistance, the more efficient the package is in eliminating heat. A heat sink may be required to maintain the device junction temperature below its maximum rating.
Because power semiconductors are primarily used as power switches, they have conduction and switching losses. Conduction losses are determined by the product of operating current and on-resistance of the device. Switching losses depends on how fast the device can switch from on to off and vice versa. The faster the switching speed, the more efficient the device.
Junction-to-case (RθJC) thermal resistance of a power semiconductor can range from 30-50 °C/W for a typical surface mount package, to 2°C/W or less for a TO-220 package. Data sheets may also specify RθJA, the power device’s junction-to-ambient thermal resistance, which is indicative of the package’s ability to remove excess heat.