Advancements in power converters are constantly striving for greater efficiency, density, reliability, and cost-effectiveness, despite the escalating demands of complex technology. However, the progress of traditional silicon (Si) power devices has reached its peak as they near their maturity stage, necessitating the search for a transformative solution. Gallium Nitride (GaN) devices are a promising alternative to silicon that is reshaping power conversion applications and opening up new possibilities.

With its high electron mobility, band gap, and velocity, GaN devices, such as the commercially available eGaN FETs, outperform traditional MOSFETs and exceed theoretical limits in terms of on-resistance versus breakdown voltage. These devices not only shrink the power device's active size but also offer significant in-circuit performance gains. From high-frequency point-of-load converters to power factor correction designs, GaN devices excel in hard-switching applications and even showcase improved performance in resonant and soft-switching scenarios.

Ⅰ. An Overview of the Performance of GaN and Si

Over the years, Figures of Merits (FOMs) have been extensively utilized as a means to assess and compare the performance of different device technologies. The previous FOMs that have been applied to evaluate hard-switching applications specifically focused on those suited for high-current low-voltage scenarios (typically below 200 V).

1. Hard switching devices

During the switching process in traditional hard-switching applications, the losses incurred are primarily influenced by two important device parameters: QGD, also known as the Miller charge and QGS2. These parameters play a crucial role in controlling the voltage and current transition times during switching events. QGD, or the Miller charge, specifically affects the rising and falling transition times of the voltage (tVR). It determines how quickly the voltage across the device rises or falls during the switching process. This parameter directly impacts the switching losses in a hard-switching transition.

On the other hand, QGS2 represents a portion of the gate-to-source charge that extends from the device's threshold voltage to the gate plateau voltage. It is responsible for controlling the current's rising and falling transition times (tCF). During low voltage-high current applications, the QGS2 term becomes significant in terms of loss contribution

Fig 1: Best suited hard switching waveforms at device turn-off transition

The initial enhancement mode gallium nitride (GaN) devices that became commercially available feature a lateral structure and operate within a voltage range of 40-200 V. These GaN devices function in a manner similar to traditional silicon (Si) MOSFETs but offer enhanced switching and packaging performance.

Fig 2: 40V device figure of merit comparison

Figure 2 presents a comparison of the switching Figure of Merit (FOM) specifically for 40 V devices. In this comparison, it is observed that the eGaN FET (enhancement mode GaN Field-Effect Transistor) exhibits a 45% lower FOM than the most advanced Si device currently available.

2. Resonant or soft-switching devices

In resonant and soft-switching applications, the aim is to minimize switching-related losses through the implementation of techniques such as zero-voltage switching (ZVS) and zero-current switching (ZCS). By employing these techniques, the losses associated with switching events are greatly reduced.

In traditional hard-switching applications, the device parameters QGD and QGS2 played a critical role in determining circuit performance, as they primarily influenced the losses. However, in resonant and soft-switching scenarios, where switching losses are minimized, these parameters are no longer the key factors determining device performance.

Resonant and soft-switching applications encompass a wide range of techniques, making it impractical to distil a single Figure of Merit (FOM) that can effectively capture the performance of different circuit topologies. Each technique has its own specific characteristics and requirements, and hence a simple metric for evaluating performance is not feasible.

i. Output Charge (QOSS)

The performance of resonant and soft-switching converters is significantly influenced by the output charge. In applications where zero-voltage switching (ZVS) is employed, the output charge does not contribute to switching losses directly. However, it plays a crucial role in determining the circulating energy required to achieve ZVS.

Various factors determine the time needed to achieve ZVS and can be calculated using a specific formula.

Eqn 1

The time needed to discharge the effective output charge, referred to as tZVS, and the current required to achieve ZVS, denoted as IZVS, are important factors in resonant and soft-switching converters. It is assumed that the current IZVS remains constant during the process.

The effective output charge, represented by QOSS, needs to be discharged before ZVS can be achieved. The time tZVS measures the duration it takes to complete this discharge process.

ii. Gate Charge (QG)

The frequency capability of resonant and soft-switching topologies is influenced by the gate charge, denoted as QG. The gate charge here is the quantity of charge necessary to completely turn on or off the transistor in the circuit. The dissipation of the gate charge in every switching cycle results in a gate drive loss. This loss is quantified by a specific equation, which takes into account the gate charge and other relevant parameters.

Eqn

In the above equation, VDR is the gate drive voltage and fs is the switching frequency.

Ⅱ. GaN vs SiC Devices at 48V IBCs

To experimentally validate the advantages of replacing Silicon (Si) MOSFETs with enhancement-mode Gallium Nitride (GaN) transistors in high-frequency resonant converters, a comparison is conducted. The evaluation involves 48 V to 12 V unregulated isolated bus converters operating at a switching frequency of 1.2 MHz and supporting an output power of up to 400W. Two different sets of power devices, one using Si and the other utilizing GaN, are compared in this setup.

The converter topology in Figure 3, incorporates a soft-switching technique to achieve Zero Voltage Switching (ZVS) for the primary devices. A resonant approach is employed to achieve Zero Current Switching (ZCS) for the secondary devices. This resonant technique also serves the purpose of limiting the turn-off current in the primary devices to the magnetizing current.

Fig 3: High frequency resonant intermediate bus converter - Schematic and its waveforms

The transition from Zero Voltage Switching (ZVS) occurs at the end of the power delivery period. From t1 to t2, the magnetizing current of the transformer charges and discharges the output capacitances of devices Q2 and Q4, facilitating a ZVS turn-on transition. If the ZVS transition period is prolonged, the body diodes of Q2 and Q4 may turn on and conduct current, as observed in the t2-t3 period. At time t3, this process is repeated for the other switching leg, with current flowing through switches Q2, Q4, and S2, as well as the leakage inductance LK2. This enables power delivery to the load while maintaining flux balancing in the transformer.