When selecting the appropriate power switches for motor control applications designers are tasked with balancing performance requirements and total system costs. When system costs are the highest priority, IGBTs often are chosen for the inverter. In general terms, the lower cost of the IGBT can be attributed to higher current density resulting in a smaller die size, compared to a MOSFET with similar current density.
Suitable IGBTs are tuned by device manufacturers for specific applications to balance between conduction losses and switching losses. This device tuning introduces a variety of effects, which impact the system in different ways depending on the type of motor used in the end product. Consequently, there are multiple ways for designers to achieve an optimal design.
The switching and conduction performance of IGBTs is a function of device structure. Fundamental and early device structures include symmetric blocking IGBTs and asymmetric blocking IGBTs. Symmetric structures, also called “reverse blocking,” have inherent forward and reverse blocking capabilities, which make them well suited for AC applications such as matrix (AC-to-AC) converters or three-level inverters. Asymmetric structures maintain only forward blocking capability and offer a lower on-state voltage drop than symmetric IGBTs. This makes them ideal for DC applications like variable speed motor control, where an anti-parallel diode is used across the device allowing operation in only the first quadrant of i-v characteristics.
Asymmetric structure IGBTs have been the focus for most manufacturers looking to optimize on-state and switching speed in medium voltage motor control applications. The key attribute of this structure is the field-stop layer created by an n-buffer region that is added beneath the n-drift region and above the lower p-doped layer. This buffer region serves to support the electric field and allows for a thinner n-drift region, which thereby decreases conduction losses. To further these gains, manufacturers have developed structures that incorporate trench-style gates instead of the traditional planar approach and backside implanted emitter concepts. The backside-implanted emitter, also called a transparent emitter structure, refers to a very thin and lightly doped p-layer. This emits few carriers into the n-base region, which reduces the stored charges and facilitates lower switching losses (Baliga).
Optimization for switching versus conduction from the perspective of the IGBT physical architecture is primarily related to the doping levels of the n-doped base region and the p-doped emitter and its available carriers. When voltage is applied to the gate, the collector current increases based on the modulation of conductivity in the n-doped base region. When the gate bias is removed, the stored charge in the n-doped base region results in a current tail that causes switching losses. These losses can be mitigated by reducing the doping levels and consequently the carrier lifetimes. Similarly, the p-doped emitter where the fourth layer with more heavily doped emitters creates more carriers, reduces conduction losses at the expense of higher switching losses.
Choosing a device that offers high switching frequencies through lower switching losses will result in higher conduction losses. This in turn requires a larger heatsink, which adds system cost and creates space issues. Alternatively, a device with lower conduction losses is most efficiently operated at lower frequencies, which may introduce audible noise due to coil vibration.
From a system perspective, the designer must also consider the motor’s size and performance along with the thermal requirements of the electronic controls. With increasing efficiency requirements, BLDCs are increasingly coming into favor in home appliances. These compact permanent magnet motors exhibit lower inductance than traditional AC induction motors. When controlling the BLDC through the PWM process, the IGBTs attempt to create a digitized sinusoidal wave. The outcome typically is the intended sinusoid with an imposed waveform that needs to be filtered. The slower the digitization, the higher the ripple and the less efficient the process becomes. In traditional AC induction motors, the motor’s inherent inductance acts as a circuit filter and reduces unwanted current ripple. With the BLDC’s lower inductance, the current ripple is large and thus reduces the motor’s efficiency.
Faster IGBTs that achieve a higher PWM frequency will reduce ripple current and the required filters can be made smaller because the output waveform is closer to the desired waveform. For other control methods, such as sensor-less FOC (Field Oriented Control), faster switching leads to faster sampling rates and better reconstruction of the Back-EMF used to calculate motor position.
As noted, both IGBT device architecture and tuning are applied to achieve both system efficiency and lower cost. In the latest technology iteration, cost is addressed by integrating a reverse recovery diode monolithically into the IGBT device itself.
Two families of Reverse-Conducting Drive (RC-D) IGBTs developed by Infineon Technologies illustrate this approach. The devices incorporate an asymmetric structure with a field-stop layer, trench-style gate and backside-implanted emitter along with a cathode in the bottom p-layer. The cathode allows for reverse current flow, while reducing the components and system cost but limits switching speed to 5 kHz. While this is acceptable for home appliances, designs that require faster speeds need another option. By varying doping techniques, higher operating frequency can be achieved with the same physical structure. The resulting RC-Drive Fast (RC-DF) IGBT family supports switching speeds up to 30 kHz using the same low-cost device architecture.
Comparing the two device families to determine suitability for different applications is based on several key parameters: conduction losses (VCESAT(for IGBT), VF (for diode)) and switching losses (ETS (for IGBT), QRR(for diode). In general, the RC-DF devices have at least 50 percent lower switching loss (ETS (mJ)) values when compared to the lower frequency devices. The low switching losses serve the needs of systems utilizing higher frequency BLDC motors, which optimize design space, performance and reliability. The RC-D devices stand out in terms of lower VCESAT values and VF values which are the thermal or conduction losses, at the expense of the higher switching losses.
Taking a closer look at the conduction losses (Fig. 1) we can also see that RC-D devices at 25°C and 175°C, represented by the thin lines, show superior VCESAT values when compared to the faster IGBT.
A closer look at the waveforms associated with the turn-on and turn-off of the IGBTs lets us compare switching losses. In the turn-on comparison (Fig. 2) the higher frequency device, shown as the red IC DUT, exhibits a lower peak reverse recovery current and quicker settling time as a result of the lower recovery charge of the integrated diode when compared to the RC-D or IC REF.
During switch turn-off (Fig. 3) the comparison shows that the high-frequency RC-DF (IC DUT) provides shorter commutation than the RC-D (IC REF).
Comparing the two devices with essentially the same device structure illustrates how doping techniques have a large impact on end performance. The overall compromise between switching losses (EOFF) and conduction losses (VCESAT) is shown in Fig. 4 (Chiola). This illustrates how understanding system requirements is key to choosing the device to meet the motor system requirements.
- Baliga, B. J. Fundamentals of Power Semiconductor Devices. New York, New York: Springer Science Business Media, LLC, 2008.
- Chiola, Davide. Reverse Conducting IGBT for Drives. Application Note. Villach Austria: Infineon Technologies, 2009.