ТаЫе 4,3 Characteristics of several commertial IGBTs

the expense of a somewhat increased on-resistancc. The current gain of the effective PNP tran.sistor con also be minimized, causing to be greater than Ij. Nonetheless, the trun-off switching time of the IGBT is significantly longer than that of the MOSFET, with typical turn-off times in the range 0.5 to 5 s. Switching lo.ss induced by IGBT current tailing is di.scussed in Section 4.3.1. The switching frequencies of PWM converters containing IGBTs arc typically in the range 1 to 30 kHz.

The added diode junction nf the IGBT is not normally designed to block significant voltage. Hence, the IGBT has negligible reverse voltage-blocking capnbility.

Since the IGBT is a four-layer device, there is the possibility of SCR-type latchup, in which the IGBT cannot be turned off by gate voltage control. Recent devices are not susceptible to this problem. These devices arc quite robu.st, hot-.spot and current crowding problems arc nonexistent, and the need for external snubbcr circuits is minimal.

The on-state forward voltage drop of the IGBT can be modeled by a forward-biased diode junction, in series with an effective on-resistancc. The temperature coefficient of the IGBT forward voltage drop is complicated by the fact that the diode junction voltage has a negative temperature coefficient, while the on-resistance has a positive temperature coefficient. Fortunately, near rated current the on-resistancc dominates, leading to an overall positive temperature coefficient. In consequence, IGBTs can be easily connected in parallel, with a modest current derating. Large modules are commercially available, containing multiple parallel-connected chips.

Characteristics of several commercially available single-chip IGBTs and multiple-chip IGBT modules are listed in Table 4.3.

4.2.5 Thyristors (SCR, СТО, MCT)

Of all conventional .semiconductor power devices, the silicon-controlled rectifier (SCR) is the oldest, has the lowest cost per rated kVA, and is capable of controlling the greatest amount of power. Devices having voltage ratings of 5(KK) to 7(KK) V and current ratings of several thousand amperes are available. In utility dc transmission line applications, series-connected light-triggered SCRs arc employed in inverters and rectifiers that interface the ac utility system to dc transmission lines which carry roughly 1 kA and 1 MV. A single large SCR fills a silicon wafer that is several inches in diameter, and is mounted in a hockey-puck-style case.

Anode (A)

Fig. 4.40 The SCR: (a) scUematit symbol, (b) equivalent circuit.

i Gate CG)

Anode

Cathode {K)

Gate

i Cathode

The schematic symbol of the SCR is illustrated in Fig. 4.40(a), and an equivalent circuit containing NPN and PNP BJT devices is illustrated in Fig. 4.4(Xb). A cross-section of the silicon chip is illustrated in Fig. 4.41. Effective transistor Q, is composed of the n, p, and n~ regions, while effective transistor Q2 is composed of the p.rf, and /J regions as illustrated.

The device is capable of blocking both positive and negative anode-to-cathode voltages. Depending on the polarity of the applied voltage, one of the p-(rjiinctions is reverse-biased. In either case, the depletion region extends into the lightly doped n~ region. As with other devices, the desired voltage breakdown rating is obtained by proper design of the n region thickness and doping concentration.

The SCR can enter the on state when the applied anode-to-cathode voltage f- is positive. Positive gate current i. then causes effective transistor Q, to turn on; this in turn supplies base current to effective transistor Q2, and causes it to turn on as well. The effective connections of the base and collector regions of transistors Q, and 62 constitute a positive feedback loop. Provided that the product of the current gains of the two transistors is greater than one, then the currents of the transistors will increase regeneratively. In the on state, the anode current is limited by the external circuit, and both effective transistors operate fully saturated. Minority carriers are injected into all four regions, and the resulting conductivity modulation leads to very low forward voltage drop. In the on state, the SCR can be modeled as a forward-biased diodejunction in series with a low-value on-resistance. Regardless of the gate current, the SCR is latched in the on state: it cannot be turned off except by application of negative anode current or negative anode-to-cathode voltage. In phase controlled converters, the SCR turns off at the zero crossing of the converter ac input or output waveform. In forced commutation converters, external commuta-

Flg. 4.41 Physical locations of the effective NFN and PNP components of ihe SCR.

Fig. 4,42 Statiti-ij.thamcterisititsoflheSCR.

Reverse blocking

Reverie breakdoyvn

Forward conducting

increasing i

Forward / AK blocking

tion circuits force tite controlled tiirn-off of the SCR, by reversing either the anode current or the anode-to-oathode voltage.

Static i-vf, characteristics of the conventional SCR are illustrated in Fig. 4.42. It can be seen that the SCR is a voltage-bidirectional two-quadrant switch. The turn-on transition is controlled actively via the gate current. The turn-off transition is passive.

During the turn-off transition, the rate at which forward anode-to-cathode voltage is reapplied must be limited, to avoid retriggering the SCR. The turn-off time t is the time required for minority Stored charge to be actively removed via negative anode current, and for recombination of any remaining minority charge. During the turn-off transition, negative anode current actively removes stored minority charge, with waveforms similar to diode turn-off transition waveforms of Fig. 4.25. Thus, after the first zero crossing of the anode current, it is necessary to wait for time / before reapplying positive anode-to-cathode voitage. It is then necessary to limit the rate at which the anode-to-cathode voltage increases, to avoid retriggering the device. Inverter-grade SCRs are optimized for faster switching times, and exhibit smaller values of l.

Conventional SCR wafers have large feature size, with coarse or nonexistent interdigitation of the gate and cathode contacts. The parasitic elements arising from this large feature size lead to several limitations. During the turn-on transition, the rate of increase of the anode current must be limited to a safe value. Otherwise, cathode current focusing can occur, which leads to formation of hot spots and device failure.

The coarse feature size of the gate and cathode structure is also what prevents the conventional SCR from being turned off by active gate control, One might apply a negative gate current, in an attempt to actively remove all of the minority stored charge and to reverse-bias the p~ii gate-cathode j unction. The reason that this attempt fails is illustrated in Fig. 4.43. Ihe large negative gate current flows laterally through the adjoining the p region, inducing a voltage drop as shown. This causes the gate-cathode junction voltage to be smaller near the gate contact, and relatively larger away from the gate contact. The ncganvc gate current is able to reverse-bias only the portion of the gate-cathode juncdon in the vicinity of the gate contact; the remainder of the gate-cathode junction continues to be forward-biased, and cathode current continues to flow. In effect, the gate contact is able to influence only the nearby portions of the cathode.

The gate turn off thyristor, or ОТО, is a modern power device having small feature size. The gate and cathode contacts highly interdigitated, such that the entire gate-cathode p--u junction can be reverse-biased via negative gate current during the turn-off transition. Like the SCR, a single large СТО can fill an entire silicon wafer. Maximum voltage and current ratings of commercial GTOs are lower than those of SCRs.