Buck Converters

48 V_IN – 1.2 V_OUT Buck Converters

Now that there is a FET that can switch reliably and with low loss in less than 10 ns [3], many of the limitations and problems created by the minimum on-time of today’s silicon MOSFETs vanish. Small and efficient high-ratio single-stage step down converters can now be built. Converting 48 V to load voltages such as 1.2 V, or converting 12 V to 0.7 V in one stage is possible. We will first look at this extreme case of a 48 V – 1.2 V single stage converter, followed by other, less extreme examples.

The silicon transistors that were selected are representative of the best generally available in the market today. Another criterion was to match as closely as possible the on-resistance and current rating of the silicon and eGaN devices. Table 5.1 gives a summary of the key characteristics of the transistors used in this buck converter comparison.

Table 5.1:  MOSFETs and eGaN FETs used in the 48 V – 1.2 V buck converter

There are a few key points to understand in this comparison. First, the gate charge of the silicon devices is many times that of the eGaN FETs. The silicon control FET requires 18 nC to switch; the eGaN FET only requires 2.7 nC. This 6:1 ratio means the eGaN FETs will have a substantially lower switching loss. The synchronous rectifier FETs also have a large difference in total gate charge, but the losses tend to be dominated by device R_DS(ON).

Secondly, due to the much lower gate capacitance, the standard figure-of-merit (FOM) of the eGaN FETs is three to six times better than that of the MOSFETs.

Lastly, note the PCB area required by the devices. The silicon devices require 61.5 mm² of board space. The eGaN FETs only require 8.5 mm². The MOSFETs require seven times the PCB area of the eGaN FETs. The savings in board space by the eGaN FETs will be a significant savings in a system with many voltage rails.

Figure 5.3: 48 V – 1.2 V efficiency vs. output current comparing 100 V EPC2001 eGaN FETs  against state-of-the-art 60 V silicon MOSFETs.

Figure 5.3 is a comparison between the eGaN FETs and the MOSFETs operating at 500 KHz. Also plotted in this figure is the efficiency for the MOSFET-based buck converter operating at 300 kHz. The eGaN FET-based converter is superior at all frequencies. The data in figure 5.4 demonstrates that the silicon devices are dissipating about 1 watt more than the eGaN devices at an output current of 8 A.

Figure 5.4:  48 V – 1.2 V losses vs. output current comparing 100 V EPC2001 eGaN FETs
against state-of-the-art 60 V silicon MOSFETs.

Figure 5.5 shows the measured switch node voltage and output current (inverted) of the eGaN FET-based converter switching at 500 kHz with an on-time of about 100 ns. Even at this very short on-time the waveform has nearly vertical edges and flat tops. A controller could easily reduce the on-time by a factor of five in order to maintain control and good transient response during a sudden load reduction.

Figure 5.5: 48 V - 1.2 V eGaN FET-based converter switch node voltage with 500 kHz switching and optimized dead-time.

If the goal is fastest possible switching, figure 5.6 shows a hard-switched transition where the rise time is about 2.5 ns. This kind of switching speed is unheard of with silicon power MOSFETs.

Figure 5.6: 48 V - 1.2 V eGaN FET hard switching transition time.

If the goal is to minimize the losses, figure  5.7 shows that soft-switching transitions of less than 5 ns are possible. The caveat is that the soft-switching transition time is load dependent.

 Figure 5.7: 48 V - 1.2 V eGaN soft switching transition.

While these fast transitions enable operation at very high frequencies with good efficiency and small size, they require care in the physical design. As discussed in chapters 3 and 4, with voltages and currents switching in just a few nanoseconds, parasitic capacitances and inductances that are negligible today will not be negligible in an eGaN FET-based system.