6.5kV IGBT Module Delivers Reliable Medium-Voltage Performance Part 1: Electrical

Compared with GTOs and IGCTs, IGBT modules offer the advantage of easier driving due to the MOS-gate and easier cooling due to a fully isolated package.

Traction, industrial drive, and pulse power applications employ high voltage, high-current power switches. Many of these high-power applications use GTOs or IGCTs, but this will change with the introduction of 6.5kV IGBT power modules shown in Fig. 1. To understand the impact of these new modules, we'll investigate power switching in these three applications.

Traction applications operate at 3kVdc, and medium-voltage industrial drive applications are at ac line voltages of 2.3kV and above. Compared with GTOs and IGCTs, IGBT modules offer the advantage of easier driving due to the MOS-gate and easier cooling due to a fully isolated package. One approach is to use a series connection of two 3.3kV IGBT modules for the above mentioned line voltages. However, a single 6.5kV IGBT can handle these tasks, which simplifies driving, control, and insulation.

Modules applied for traction purposes are exposed to a nominal 3kVdc link voltage. These line voltages show tolerances of +20%. However, from a practical point of view, voltage fluctuations and regenerative braking can push the worst-case dc level to even higher values. Thanks to their high-blocking voltage capability, the 6.5kV modules ensure a large safety margin for inductive overvoltage spikes — even allowing short-circuit turn-off at a 4500Vdc. Under these conditions, the 6.5kV IGBT module can handle safe turn-off for a di/dt up to 10kA/ms with stray inductance up to 200 nH.

Medium-voltage industrial drives are also a candidate for using the higher voltage IGBT modules. The definition of medium voltage (MV) varies by industry, country, and application. For ac motor drives, the range reaches from above 600V up to 15kV. The MV threshold in Europe is 1kV. More practically, existing MV drive products have a range of 2.3kV to 7.2kV, standard North American medium voltages are 2.3kV and 4.16kV, while predominating line voltages in Europe and the rest of the world are 3.3kV and 6.6kV. Typical medium-voltage drive applications are pumps, fans, compressors, extruders, and conveyor systems found in mining, water, wastewater, petrochemical, steel, cement, and paper production industries, as well as machine and ship building.

You can easily project converters for a 2.3kVac line, 3.25kV dc-link voltage with the switches directly exposed to the line voltage. Applications at 4.16kVac lines (5.9kVdc-link) are possible with a three-level neutral point clamped (NPC) circuit design. By keeping the standard module shape, you can adapt the new voltage class to existing inverter designs that operate at lower line voltages.

Pulse power applications using IGBTs increasingly gain in importance worldwide. New design approaches target the replacement of thyratron tubes with solid-state modulators, requiring rise and fall times in a few nanoseconds. The devices must be able to switch peak power ratings of above 10 MW, supplying klystrons as well as magnetic cores with pulses in the microsecond time frame.

6.5kV Chip Technology

The newly developed 6.5kV module employs two innovative component concepts for power electronics: the Field Stop (FS) IGBT and the EMCON diode ^[1]. The FS IGBT overcomes drawbacks of PT (punch-through) and NPT (non-punch-through) IGBT technology. The PT IGBT needs a high carrier concentration at the back, resulting in undesired high turn-off tail current and losses. Or, it needs extremely high lifetime doping, leading to a high on-state voltage. In contrast, the NPT IGBT has a more favorable low carrier concentration at the back, but the n- layer has to be thick to produce a triangular electrical field distribution to block high voltage. If the NPT structure were thinner, it would result in higher static and dynamic losses.

For better high-voltage performance, it was necessary to convert the NPT structure to a device with a trapezoidal field distribution under blocking conditions (typical for the PT IGBT). However, it was important to do this without giving up the inherent advantages of the NPT concept of its low efficiency emitter and high carrier lifetime. This is possible by implementing the so-called Field Stop layer (Fig. 2, on page 15).

The field stop behavior results in a different dependency of the turn-off losses vs. collector-emitter voltage. For an NPT IGBT the increase of losses is linear, for the FS IGBT the increase is slightly more. For an inductive turn-off at low collector-emitter voltage it's well-known behavior of the NPT IGBT to have a very low, long tail current. The tail current endurance reduces when the collector-emitter voltage changes to values where the electrical field reaches the field stop layer. At high collector-emitter voltages there's virtually no tail current left. Figs. 3 and 4, on page 16, show the turn-off and turn-on behavior, respectively, of the 6.5kV module.

Diode recovery influences IGBT turn-on losses. Controlling the di/dt-capability of the freewheeling diode optimizes the IGBT's di/dt to ensure switching within the diode's safe operating area (SOA). As done with 3.3kV devices, the IGBT has an additional gate-emitter capacitor, C_ge, which controls the turn-on di/dt and dv/dt independently. Raising the IGBT's gate turn-on resistor (R_g) would decrease the di/dt, but also decrease dv/dt, so the turn-on losses would increase. You can bypass this problem by using R_g and C_ge || C_gc (C_gc=Miller capacitance) for the turn-on dv/dt only. Because C_gc >> C_ge, in a first step, the dv/dt can be fixed to an appropriate value just by R_g (and not C_ge). In a second step the di/dt can be fixed with the additional component C_ge by the time constant given by R_g and C_g || C_ge (C_g=gate capacitance).

The IGBT-related RBSOA diagram gives the maximum allowed peak voltage vs. turn-off current for the IGBT, roughly saying V_ce should never exceed the nominal blocking voltage and the turn-off current not exceed two times the nominal current. In opposition, the diode SOA defines the maximum peak power on the diode. Fig. 5(a), on page 17, shows the diode reverse voltage vs. the recovery current. By tracing the curves of v_R(t) and i_R(t), in the case of the 600A module, their product should not exceed the peak power of 1.8MW. Fig. 5(b), on page 17, shows two examples for the locus of v_R×i_R: the recovery for 600A against 3.6kV and 1200A against 4kV.

By using the recommended C_ge, not undershooting the minimum R_g(on), and not exceeding a commutation inductance of 280 nH provides reliable operation without verification.

Besides the influences on the switching behavior there are, due to the reduced wafer thickness, significant reductions possible for on-state and forward voltage of the IGBT and diode. You can achieve values as low as shown in Figs. 6(a) and 6(b) only by the introduction of the field stop technology. Similar to the IGBT, the EMCON diode makes use of the field stop structure to obtain a vertically optimized device with soft recovery behavior. Due to a further reduction of the already weak carrier lifetime killing of the EMCON diode, a reduction of the forward voltage and smaller temperature coefficients of the electrical parameters are possible.

The 6.5kV chip combines the advantages of the field stop with the well-known merits of the NPT technology: its short circuit ruggedness and high current turn-off capability. The module's design handles are designed for a maximum dc-link voltage of up to 4500V. Even at this voltage they are able to turn-off double the rated current and survive a short circuit (Fig. 7) for 10 μsec.

Another advantage of the NPT technology is the positive temperature coefficient of V_CEsat. The diode shows only a slightly negative coefficient below and a positive coefficient for nominal current and above, which is a benefit for paralleling a large number of chips and modules.