The following sections show the cross sectional views and describes the operation of six different devices. More detailed information is available in each device's specific Training Module. The goal of this section is to compare and contrast the various devices: pn diode, Schottky diode, SCR, MOSFET, IGBT, and Control IC.

pn Diode

As shown in Figure 13, the top metal is the anode, while the bottom is the cathode. The action occurs at the interface, called the junction, between the implanted p-type and n-type materials. When a positive voltage is applied between the anode and cathode, current will flow through the diode, provided the voltage is greater than "a diode drop" which, for standard pn diodes, is usually around 0.7V. As the forward current (IF) increases, the voltage drop (VF) will also increase. However, most of the voltage drop is the initial 0.7V drop which occurs when any amount of current flows through the diode.

Figure 13. PN Diode

If a negative voltage is applied across the pn junction (anode to cathode), the device exhibits very high resistance to current flow, and the small amount of current that does conduct is called the leakage current (IRM).

When a diode is conducting in the forward direction and is asked to block in the reverse direction, it "forgets" it is a diode for a period of time and allows current to conduct. After this short period of time, called the reverse recovery time, or trr, the diode "remembers" it is a diode and begins blocking current. However, during this recovery time, a large current conducts through the diode, called reverse recovery current, or Irr. The shape of the waveforms during this period are critical to the operation of the rest of the circuit, which is why IR has developed the HEXFRED® diode, an ultra-fast, but ultra-soft diode unlike snappier diodes from our competitors that usually cause excessive voltage ringing in the circuit. As temperature increases, the forward voltage decreases, while the reverse recovery current and charge increase.

Schottky Diode

As shown in Figure 14, the Schottky diode is very similar to a standard pn diode, but instead of having an implanted p-layer, the action occurs at the interface between the barrier metal and the silicon. The guard rings are used to make the device's reverse breakdown characteristics more rugged. Since both metal and the silicon are n-type materials, the conduction occurs through majority carriers only, with no minority carrier injection, storage, or recombination . This explains the Schottky diode's lack of reverse recovery, making it ideal for high frequency applications.

Figure 14. Schottky Diode

The barrier metal also is responsible for the Schottky diode's low forward voltage drop, making it ideal for use in low voltage systems. Of course, the tradeoff is the reverse leakage current, which is many times that seen in pn-junction diodes. In some applications, and especially during burn-in, this leakage current may cause the device to exceed its rated junction temperature. It needs to be included in any junction temperature calculations. As temperature increases, the forward drop decreases, while the reverse leakage current greatly increases.

SCR

The SCR (silicon controlled rectifier), or thyristor, is one of the original high power semiconductor switching technologies. As shown in Figure 15, the SCR is a four layer device, npnp from top to bottom (or cathode to anode). It is a latching device; once it is turned on, or "fired," it remains on until the current is removed. For this reason, its primary application is phase-control of ac signals. Figure 16 shows that by controlling where on the cycle the SCR is turned on, the output power level is controlled. SCRs designed for these line frequency (50-60 Hz) applications are called phase control SCRs.

Figure 15. SCR

Figure 16. Phase Control of ac Waveform

The second family of SCRs is the inverter type. These are used in pulsed power applications involving higher frequencies. The main difference between the two families is the turn-off time (tq). A device's tq is measured as the time required for the device to be in the "off" state before voltage is reapplied. Inverter SCRs typically have a tq of less than 30 microseconds (ms). Similar phase control SCRs have tq ratings of several hundred ms.

Like with the pn diode, as the temperature increases, the voltage drop decreases. Perhaps more importantly, as temperature increases, the current required to fire the SCR decreases. At low temperatures, the gate triggering circuitry must supply enough current to ensure the device fires, while at high temperatures, the SCR is susceptible to spurious firing due to noise. The device's gate triggering circuitry must ensure this does not happen.

Figure 17. HEXFET power MOSFET

MOSFET

As shown in Figure 17 above, the HEXFET® power MOSFET is named for the hexagonal shape of its individual cells. Current flows from the source metallization down through the device, and out through the drain contact. Vertical current flow is the reason the HEXFET is also called a vertical MOSFET. Nearly all power MOSFETs on the market employ this vertical structure.

The MOSFET (Metal Oxide Semiconductor Field Effect Transistor) is used primarily in medium-power circuits where switching speed is critical. This power device is extremely easy to drive as it requires voltage, not current on the gate. And due to its wide acceptance, the MOSFET market is growing at a rapid pace.

It is important to select the proper voltage MOSFET as the RDS(on) increases exponentially with increasing breakdown voltage. Also, as the device heats up due to power dissipation, its RDS(on) increases. Thus, in most applications, the 25 degC RDS(on) rating is not accurate. The actual RDS(on) rating could be twice as high. This is not always a negative attribute. It's what allows power MOSFETs to easily be used in parallel.

IGBT

The IGBT (Insulated Gate Bipolar Transistor) represents the union of a power MOSFET and a power bipolar (BJT) transistor, incorporating the best features of each. But while the MOSFET can be used in applications exceeding 1 MHz, the fastest IGBTs are limited to only a fraction of that. Therefore, the only real drawback of the IGBT is its switching speed. Yet the conduction characteristics of the IGBT really outshine those of the power MOSFET, especially at voltages greater than about 200V. If you have a midfrequency, high voltage design, look at IGBTs.

Increasing the operating temperature of an IGBT causes its switching losses to increase substantially, thus decreasing the maximum operating frequency. The conduction voltage of IGBTs is not strongly affected by temperature, and can even decrease with temperature at certain current densities. This causes concern for designers with regards to the IGBT's parallelibility. Still, IR believes that by following a few simple guidelines, IGBTs can be successfully paralleled. (See Design Tip 94-6 for more information on parallel operation of IGBTs.)

Control IC

In 1987, IR introduced the IR2110, a half-bridge, high voltage, MOS Gate Driver, or Control IC . Since that time, many variations on the IR2110 have been introduced. These devices are unique in that they simplify the drive circuitry required to drive a high-side, n-channel MOSFET at voltages up to 600V. These devices allow designers to reduce the parts count and design time for the drive circuitry. Figure 18 shows a cross-section of the die. One of the most successful devices in this family is the IR2151, a self-oscillating, half-bridge driver designed primarily for electronic ballast applications.

Future 18. Control IC

Modes of Operation

The switching devices (MOSFET, IGBT, and SCR) can be operated in one of several modes. The SCR is unique in this group in that it is a self-commutating device, which means the user is required to turn on, or "fire" the SCR. But at the zero crossing, the SCR switches off. This makes the SCR extremely useful in ac applications, or where the current decays to zero at the point when the SCR should be turned off. One limitation of the SCR is that it cannot easily be used to switch dc loads.

By far the largest number of IR devices are used in hard switching as depicted in Figure 19. The switching waveforms on data sheets are for hard switching. The conditions are very difficult on the switching device, i.e., high current must be switched off to high voltage. Considerable power is dissipated in the switching interval due to these conditions.

Figure 19. Hard Switching Waveforms

Pulse-Width Modulation (PWM) is a special case of hard switching. In many applications, it is desirable to replicate a sine wave as shown in Figure 20. One way to accomplish this is to approximate the sine wave with narrow square pulses of varying duty cycle. After this waveform is smoothed, typically by an inductor, it appears very similar to the desired sine wave. The frequency of the desired sine wave is called the carrier frequency, while the frequency at which the switch operates is called the modulation frequency. To make the replicated waveform closely match the desired waveform, the modulation frequency is usually at least ten times the carrier frequency.

Some applications employ resonant mode switching. In these applications, the current and/or voltage is a sine wave as opposed to the square waves common in PWM techniques. Operation in the resonant mode has lower switching losses, and is used with devices that have high switching losses, or to push operating frequency higher.

Figure 20. Sine Wave Generation through PWM.

Some applications operate in the linear mode. As applied to switching devices, linear mode means that discrete changes in the control signal result in proportional discrete changes in the output. When a circuit is operated in the linear mode, the switching element limits the current in the circuit, while normally the circuit itself, rather than the switching element, limits the current. This limitation leads to high power dissipation in the switching device.

Parallel operation of semiconductors requires extra effort on the part of the design engineer. When operating semiconductors in parallel, the critical parameter is the temperature coefficient. The temperature coefficient reflects how the semiconductor's voltage drop responds to changes in temperature. In general, semiconductors have a negative temperature coefficient, with the one notable exception of the power MOSFET in that its temperature coefficient is positive. A negative temperature coefficient means that the voltage drop across the semiconductor decreases as the temperature increases. This causes problems such as current hogging, thermal runaway, and hotshots. As the device heats up due to normal power dissipation, the voltage drop decreases. This allows more current to flow, generates more heat, and further reduces the voltage drop, creating a regenerative effect. When paralleling any IR device (other than the MOSFET) special design considerations are necessary to prevent potential current sharing problems.

The following documents provide more information on this subject: IR Design Tip 94-6A, "Parallel Operation of IGBTs"; IR Application Note AN-990, "Application Characterization of IGBTs"; and "Paralleling of Power MOSFETs for Higher Power Output," by James B. Forsythe (PowerCon '81).

Packaging

Many IR devices are available either in plastic or hermetically sealed metal packages. In the past, the deciding which package to use was divided between commercial versus military. Today the division is not as clear. Military does not automatically mean hermetic, nor does hermetic automatically mean more reliable. However, most space applications still require hermetic packages. In applications where the device is exposed to high temperature, and high humidity, a hermetically packaged device will improve its reliability. The TO-3, once the power transistor package, is no longer competitive with the new package styles. The TO-3 is difficult to heat sink, and must be isolated externally. New packages, such as the TO-254 (M-Pak) are tab mounted, and isolated, making assembly much easier.

Most heat sinkable plastic packages have a metal tab connected to the heat sink by the user. The die is mounted directly on this tab, usually the positive terminal of the device. In some applications, it is desireable to have the heat sink grounded, while in other applications it is easier to insulate the heat sink from the rest of the system. In cases with a grounded heat sink, it is necessary to isolate the device from the heat sink. For this purpose, IR manufactures special versions of the TO-220 and TO-247 packages called Full-Paks that have a very thin plastic coating on the exposed metal of the device. The plastic provides up to 2500Vrms isolation voltage, while being thin enough to only moderately increase the thermal resistance of the mounting system. As an alternative to buying these Full-Pak devices from IR, the user can isolate the back of the package from the heat sink using a thermally conductive, electrically isolating material as shown in Figure 22. Some common types are mica, pressed ceramic wafer, polyimide, and elastomeric insulators, with the latter gaining more and more popularity. The screw hole in the TO-247ac package is already isolated, so only the back of the package needs to be isolated. The screw hole of the TO-220AB is not isolated, so the user must isolate it. One way is by using a nylon shoulder washer with a standard steel screw. The steel screw is typically 4-40 whereas a 6-32 screw is normally used when isolation is of no concern.

The extremely broad R product line is packaged in anything from a tiny,3-lead surface mount package (barely visible to the naked eye) to a huge "hockey puk" package (greater than 4" in diameter!). IR application note AN-995 discusses IR's various surface mount packages, and how to mount them. Most high power products (SCRs, and diodes in both discrete and module packages) are simple to mount. These devices are typically large, and connnected using large bolts, or are stud mounted. Remember, however, not to exceed the torque or force specified on the data sheet. For the hockey puk packages, a suitable mounting clamp must be used as shown in Figure 22 below. In fact, the puk will appear open-circuited if pressure is not applied since this is a compression bonded device.

Figure 22. Isolation System for Standard TO-220AB and TO-247AC

Figure 23. Mounting Hockey Puk SCRs and Diodes