by Franco Musiari [ DESIGN IN ]

Born around the year 2000 they already are at the second or the third generation and still represent the highest efficiency.

The miniaturization of the systems is a trend followed in all the electronics systems and particularly in the portables ones. In many of them the dimension and the weight of the power supplies are a dominant parameter of the whole system and more and more switching power supplies manufacturers (SMPS – switching mode power supply) define the road-map to increase the density power of their devices.

The most common solutions used are focused on two aspects:

• The reduction of the passives components dimension (increasing the switching frequency) and
• Reduction of losses (and the corresponding effort needed for the cooling through heatsinks and / or ventilators).

To achieve these aims the main components must reduce significantly the losses of the switching and that’s why the unipolar semiconductors such as MOSFET and Schottky barrier diodes (SBD Shottky Barrier Diode) replaced, in the most cases, the bipolar devices. The beauty of the unipolar semiconductors is the absence of accumulation of the minority charge carrier and, therefore, in a theoretical instantaneous switching only limited by the small parasitic capacities. These days it is possible to find silicon power MOSFET with blocking voltages up to 1.000/1.200 and 1500 Volt but this is not true for the SBD because they are capable of blocking voltages up to 250/300 volt for the intrinsic characteristics of the base materials such as silicon (Si) or Gallium arsenide (GaAS).

The SiC.

There are about 170 crystal structures of this chemical compound whose hardness is almost comparable with the diamond hardness, but only two structures are available, classified as 4H-SiC and 6H-SiC, the former being the more commonly used in electronics for the characteristics of greater uniformity in three dimensions (3D).

In table 1 are compared the fundamental electronics characteristics of : Silicon, Gallium arsenide and SiC.

Table
1: Basic electronics characteristics of Si, GaAs, and 4H-SiC

The high value of the electric breakdown field allows the utilization of the SiC in applications that can function from 600 to 2.000 volt. The advantages offered by the intrinsic properties of SiC can be summarized in three points:

• A 10 times higher electric breakdown field reduces drastically the resistance of conduction (on resistance) – see the graphic of Fig. 1: at 600 V the SiC diodes have a Ron of 1,4 m-cm2 considerably smaller of 6,5 m-cm2 of GaAs and of 73 m-cm2 for SBD in Si.

• The high band energy brings to a higher height of the schottky metal-semiconductor barrier meaning that the leakage currents are extremely smaller even at high junction temperatures

• The high thermal conductibility reduces the thermal resistance facilitating heat dispersion.

Schottky Diodes - SiC Vs Si
Historically the power systems operating from 600 to 1200 volt use silicon pin diodes that, when direct biased, tend to store big quantities of minority carriers. These charges must be removed, for recombination with the majority carriers before the diode is in “non-conduction”. This process takes time (hundreds of nanoseconds) and causes a particularly significant reverse current (with values comparable to the direct current) which not only must be taken in account when sizing the complementary devices (as the power MOSFET), but also heavily influences on the switching losses of the system.

Power Factor Correction (PFC)

This phenomenon is drastically reduced in the case of SiC SBD : the switch off time is less than 10 ns and reverse current is about zero. The graphic from Fig. 2 compares the behavior of a SiC diode of first generation with an equivalent silicon pin diode. Also SiC diodes have a small switch off current (<2A) due to the junction charge, but it is independent of temperature, of the level of the direct current and of the derivative di/dt. One of the applications that most likely will be involving SiC Schottky diode is the correction systems of the power factor (PFC - Power Factor Correction). In the traditional power supplies the Ac input has a heavily inductive load (transformer) and the following switching power supply stage doesn’t reduce the influence of the inductive part of the load which means a power factor far from 1. The laws, whose goal is to reduce the unnecessary consumptions, and the utility companes that provide electricity impose this factor to be one. It’s necessary to insert a PFC circuit which allows correcting this parameter. A typical scheme performing this function is shown in Figure 3 where a MOSFET when conducting forces the charging of the inductance, whose stored energy, once the MOSFET is switched off , is released to the output circuit. When the MOSFET is ON it is necessary to prevent that the charging flows from the output filter capacitor to the MOSFET and that's why diode D is used. When the MOSFET is ON the diode is OFF and vice-versa but during the switching transient - when the MOSFET passes in conduction - the reverse current of the diode adds up to the current charge of the inductance causing an overload of power transistor. This forces to an excessive MOSFET dimensioning, with higher costs, and also causes an increase of the switching losses that limit the operating frequency of the circuit by preventing the use of smaller components (inductance first of all).

In a test PFC circuit operating at a frequency of 90 kHz with a constant input tension of 120 V RMS and output tension of 370 V DS with MOSFET from 14 A, 500 V of International Rectifier (IRFP450) were tested a super-fast PiN diode of 6 A, 600 V from IR (HFA08TB60) and a SBD SIC of Cree from 4 A, 600 V (CSD04060) under full load conditions and halved load calculating related losses. The result of the measures is shown in the graph of Fig. 4 where reduction of the losses of 24 and 27% in two different load conditions is remarkable.

When flow of current is stopped in an inductive load (and an induction motor is certainly an inductive load) this tries to reverse the polarity and to increase the tension. This reverse voltage can also reach significant and sufficient levels to damage the switching device (IGBT, thyristor, etc). The freewheeling diode is normally placed in parallel to the switching element to allow the flow of this reverse current and disperse the effects.

If in the MOSFET is possible to realize the freewheeling diode within the transistor structure, this is not possible on the most diffused IGBT used in the inverter to drive motors when the power is over a certain level. The freewheeling diode, normally a fast silicon diode assembled in parallel onthe IGBT within the same package (copacked). If in the last two decades the solution for freewheeling diode was, for issues of availability and cost, only silicon, with recent improvements in technology for the realization of 4H-SiC wafers of high quality and without defects use of SiC diodes for the realization of IGBT with freewheeling diode of the SiC become reasonable. Also in this case Cree realized some test on a driving system for motors from 2,3kW (3HP), 240V which operates at a frequency of 16kHz PWM, and realized with standard silicon IGBT in TO-220 package. The same tests were then made with the equivalent Cree from 15A, 600V (CID150660) where SiC diodes are employed. The graph of Fig. 6 compares the losses in the two cases where it highlights a reduction of 32% of the dissipated power in the power section of the inverter.