Electronics 101

 

3. Active Components

There are many types of active components but only two are the basic building blocks of power circuits: diodes and transistors. Thyristors, once the component of choice in power electronics, are still the dominant player in a specialized sector: very high power applications, mainly in utilities substations.
 

3.1 Diodes

The distinguishing feature of a diode is that it conducts current in one direction only, the “forward” direction, when the voltage at its anode is positive. It blocks voltage in the “reverse” direction (Figure 3).

Figure 3. A diode conducts when its anode is positive with respect to the cathode (reference terminal). Notice from the curve, that it will block voltages only up to a certain voltage, after which it goes into breakdown. Diodes rated for avalanche will take a specified amount of energy in avalanche. Other would likely fail.

 

An ideal diode would be a perfect conductor in the forward direction and a perfect insulator in the reverse direction. The limitations of a real diode are described in its data sheet and amount to the following:

  • When it conducts it has a voltage drop in the order of 1V across its terminals;
  • When it blocks, its blocking capability is limited to its voltage rating. The small reverse current that flows in the reverse direction can be safely ignored, except in the case of Schottky diodes.
  • When the voltage across its terminal reverses it goes through a “recovery” transient. It cannot switch instantly from forward conduction to blocking and vice-versa. Reverse recovery behavior is characterized in the data sheet of diodes targeted for high frequency applications. Forward recovery is seldom characterized.

As implied in the previous paragraph, diodes come in two broad categories: rectifier diodes and high frequency diodes.

Rectifier diodes are used to convert line voltage to a DC   voltage and operate at line frequency or a small multiple thereof. Forward voltage drop and blocking capability are both of primary importance. Recovery characteristics are of lesser importance, except for their influence on input EMI and, consequently, input filter.

High frequency diodes are used in most switching and resonant applications. In these applications forward and recovery characteristics are more important than forward voltage drop.

 

3.2 Power Transistors

The distinguishing feature of a transistor is a control terminal that changes the state of the device from blocking to conducting and vice-versa (Figure 4). While a diode always conducts when its anode goes positive, a transistor conducts only if its control terminal is biased for conduction.

Figure 4. The most common power transistors. The reference terminal of power transistors is the emitter or the source. Voltages are defined with respect to the reference terminal.

 

Conduction characteristics. Like a diode, a transistor has a small voltage drop across its terminals when it’s in conduction. This drop causes conduction losses during the period of time the device is in conduction. This energy can be calculated by a simple formula:

Econ =  Iload * Von  * ton

If the transistor is operated at a fixed frequency and duty cycle the conduction power losses are:

Pcon = Iload * Von  *DC * f

In all semiconductors Von changes with temperature: it goes up in MOSFETs and goes down in diodes. The proper value of Von must be used in the calculation. The same method of calculation applies to diode conduction losses.
 
Blocking characteristics. Like a diode, a transistor has a polarity, but this is where the similitude ends.  A transistor can block or conduct, depending on the control terminal, but in one direction only. The reverse blocking capability of the bipolar transistor is around 5V. The MOSFET has no reverse blocking capability. The IGBT has a reverse blocking capability in the range of 20 to 50V, depending on the technology. Data sheets for IGBTs introduced in the last few years do not specify it any longer. For this reason transistors are mainly used in topologies that operate from a DC bus, while most diodes and thyristors are used in AC circuits.
 

Switching characteristics. The transition from the conducting to the blocking state is not instantaneous. At turn-on the voltage across the transistor swings from the bus voltage to its voltage drop, while the current goes from zero to whatever current is required by the load (plus the reverse recovery of the diode). During the turn-off transient the voltage swings back to the supply voltage and the current back to zero. As a first approximation and for a resistive load, we can say that during a switching transient the average voltage across the device is one half of the supply voltage and the current one half of the load current. We multiply this for the switching time and the frequency to obtain the switching power losses:

Eon = 0.5 x Iload * Vsupply  * ton

Eoff = 0.5 x Iload * Vsupply  * toff

Psw = Iload * Vsupply  * (ton + toff) * f

In practice loads are seldom resistive and the designer needs a much more accurate number for switching losses than what this expression can provide, as the thermal design is largely predicated on this number.
To complicate matters further, the switching losses are a strong function of temperature.
This topic will be revisited in Chapter 5.
 
Control. A voltage (MOSFETs and IGBTs) or a current (bipolar transistors) applied to the control terminal causes the device to transition from the blocking to conduction state or vice-versa.
 
The vast majority of power circuits operate in switchmode. This means that the transistor acts as a switch, either fully on or fully off. In a limited number of applications the transistor is operated in linear mode, i.e. the voltage across its terminals is modulated by the control terminal.
As we have seen, the power dissipated in the transistor is the product of the voltage across its terminals times the current through it. If the voltage across its terminals is being modulated, the power dissipated inside the transistor is significant. This power needs to be removed from the device and dissipated with a suitable heatsink or the transistor will fail. For this reason, if at all possible, power circuits operate in switchmode. Thermal design is covered briefly in Chapter 5.

 

3.3 Majority carrier devices v. minority carrier devices
In the previous two sections we have briefly looked at two types of active components, diodes and transistors, on the basis of their functionality in the application. The same two types of devices could also be classified according to their physical structure, as we are about to see.
Most readers are interested in the device performance and much less interested in its physical structure. However, a few words on the physical structure will give a powerful insight into the switching behavior of a power semiconductor.
 
Silicon is a “semiconductor” and, as such, it’s a bad conductor and of little use in any electrical circuit. Charges need to be implanted in the crystal structure to lower its resistivity. These charges can be positive (holes) or negative (electrons). After the introduction of these impurities the silicon becomes N-type or P-type. In the N-type silicon the carriers are electrons, in the P-type they are holes.
 

Visualize for a moment a fish tank with water coming in and flowing out. By injecting the water into the tank we are not changing the content of the tank itself: it was water and remains water. If we shut-off the water, the tank is immediately at rest.

Air is also injected in a fish tank. Now the tank contains a mixture of water and air. If we shut off the air, the tank remains in a mixed state until all the air bubbles have migrated to the top.

 
The conduction of current in a majority-carrier device is by means of one type of charge, normally electrons. It is like the injecting water into the fish tank. When the control terminal stops the injection the device is instantly at rest.
Minority-carrier devices conduct current by injecting minority carriers [holes, into N-type silicon, where the majority carriers are electrons or electrons into P-type silicon, where the majority carriers are holes]. This is like injecting air in the fish tank. When the control terminal stops this injection the device remains in a transitional state until all the minority carriers are disposed of. For this reason minority carrier devices have a “storage time” or a “recovery time” or a “tail” or a “recombination time”. Majority carrier devices do not have these phenomena.
 
Schottky diodes and power MOSFETs are majority carrier devices (Figures  5, 6 and 7). Bipolar transistors and IGBTs are minority carrier devices (Figure 8). When looking at a data sheet of a Schottky diode or a MOSFET the critical parameter most engineers look for is the voltage drop (or on-resistance). In these devices high frequency performance is taken for granted.
When looking at the data sheet of an IGBT the critical parameter is switching energy. Unfortunately, most engineers still look at the voltage drop and attach more relevance to it than warranted.

Figure 5. A Schottky diode is made of a metal, like platinum or tungsten, deposited over a heavily doped N+ silicon layer. There is no P material, hence there is no junction. This metal-on-silicon structure is called a “barrier”.

 

 

Figure 6. A lateral MOSFET is made of two N+ diffusions into a P substrate. A layer of gate oxide (an insulator) bridges the two N+ diffusions. When no voltage is applied to the gate this structure is like two diodes against each other and cannot conduct. When voltage is applied to the gate charges collect under the gate oxide so that current can flow from the drain to the source. This figure is just to explain its operation. The cross-section of a Power MOSFET is shown in Figure 7.

 

 

Figure 7. Cross-section of a typical planar power MOSFET.  Transistor current consists of electrons injected into the N- region (majority carriers). Diode current consists of holes injected into the N- region. As such, diode current is a minority carrier current and the holes have to recombine or be swept away before the diode recovers its blocking capability (Figure 9).

 

 

Figure 8. Typical structures of planar IGBTs. Notice that the IGBT is a four-layer structure (P-N-P-N). Holes are injected into the N- region. When the gate stops the injection the N- region is still flooded with charges. Some recombine with electrons, others are swept away by the electric field between collector and emitter due to the voltage increase at turn-off.

 

It is important to remember that MOSFETs have good high-frequency performance but conduction losses that go with the square of the current (R I2). IGBTs, on the other hand, have higher switching losses, but conduction losses almost linear with current. The only way to reduce the conduction losses in a MOSFET is to increase its die size. Beyond a certain level of current this proposition becomes uneconomical.
 

Figure 9 shows a typical diode reverse recovery. An “ideal” diode cannot conduct in reverse. This particular diode conducts a reverse current that is over twice the forward current. Notice, also, that the diode does not block instantly when the voltage is reversed. A significant amount of power is dissipated in the device during the recovery transient but an even larger amount of power is dissipated into the transistor that has to conduct such a large amount of current.

Figure 9. Reverse recovery in a 1200V soft-recovery diode. Soft recovery is obtained with a specific device design that includes a precise He implant in the N-region.  Forward current: 25A, reverse voltage 600V, 150°C.

 

The shape of the reverse recovery waveform is particularly important for its EMI (ElectroMagnetic Interference) implications. This is because the di/dt during the tb portion of the waveform is frequently the highest di/dt in the circuit and generates a significant amount of radiated noise. Two terms are frequently used to characterize the recovery behavior: “soft” and “snappy”: the former has a lower di/dt during the tb portion of the reverse recovery, hence a lower EMI signature. The waveform in Figure 9 shows a very soft recovery.

As mentioned above, conduction in the MOSFET integral rectifier is through minority carriers. As such, its recovery characteristics are the same as those of a standard minority carrier diode. This may be a serious limitation to the frequency of operation in those topologies where the MOSFET diode is a critical component.

 
Figure 10 shows the characteristic IGBT “tail”. The area underneath the tail gives a good indication of the stored charge in the N- region. The pink curve shows the energy that is being dissipated inside the device during this transient. The frequency of operation of a minority carrier device is limited by the power dissipated during the turn-on and turn-off transients.

Figure 10. The IGBT turn-off is a two-step process: first it decays from 100A to 20A in appx. 500ns, then it tapers off to zero in approximately 1µs. Notice that the IGBT is blocking the full voltage (600V) while collector current is still flowing. During this period of time the minority charges are being cleared out of the N- layer. Waveforms taken at 150°C.

 

Semiconductors have their own parasitic capacitances across junctions and oxide layers. Packages have a small amount of parasitic inductance in the wire bonds. These parasitics may be negligible at low frequency (relatively speaking) but they become the limiting factor in the frequency performance of majority carrier devices.

Package inductance can be mitigated by eliminating wire bonds and with parallel runs. Device capacitances can only be addressed with a more complex silicon structure.