9 Synchronous GaN Bootstrap Method

The active switch method from the previous section uses a diode to block the high voltage of the HS node and a low-voltage switch in series to activate or deactivate charging. Using a single high-voltage FET to fulfill the roles of both the switch and the diode is also possible. However, one issue is that MOSFETs have body diodes, which means that when the device is turned off, the body diode can still conduct and allow overcharging during dead time.

The GaN FET lacks a body diode, conveniently presenting an opportunity to address the body diode conduction issue. The GaN FET series switch prevents bootstrap overcharge without an extra diode, as shown in Figure 9-1. As discussed, the lack of a body diode does not mean that GaN FET cannot conduct during the dead-time. However, adjusting the voltage drop of the bootstrap GaN FET to match the voltage drop of the low-side GaN FET is possible. Matching the voltage drop of each FET cancels out the negative voltage, and prevents overcharging.

M1 is the synchronous bootstrap GaN FET. M1 needs a blocking voltage to handle the full HS voltage, preferably with a low Coss and Cgs (for fast switching). The M1 GaN FET source is connected to VDD and driven with LO, like the LMG1210 in the previous section. In addition, a level shifter consisting of D1 and C1 steps up the LO signal above VDD. The level shifter is required because the source of the GaN FET is tied to VDD instead of 0 V, so LO needs to be higher than VDD to have a positive Vgs.

There are a few advantages to using the synchronous GaN FET bootstrap. The forward voltage drop during the normal charging period is smaller than the VF of a diode, meaning the bootstrap voltage is closer to VDD. This method also prevents bootstrap overcharge with higher efficiency due to fewer series elements. The GaN FET has no reverse recovery charge or time, which makes the design effective at high switching frequencies.

In Figure 9-2, different VF diodes were used in the level-shift circuit. A diode with a low VF of 0.3 V allows more charging in the dead time, as shown by the current spike around 10 ns. The amplitude of Iboot is related to the VF of the level-shift diode. In the second dead time around 60 ns, the 1.2 V VF diode does not allow conduction.

Equation 1 demonstrates that VGoff partially determines the VSD. The diode in the level-shift circuit has a forward voltage drop. This VF results in the effective Vgs of the GaN FET bootstrap shifting down by the diode drop. Instead of being 0 V to 5 V, the Vgs of the GaN FET is –0.7 V to 4.3 V. Therefore, the VGoff is equal to the VF of the diode. The goal is to match the VSD of the bootstrap GaN FET to the VSD of the low-side GaN FET, so selecting a diode with a different VF is an excellent tool to achieve that.

Schottky diodes and GaN FETs do not have reverse recovery. However, both have effective capacitance that requires charging and discharging every switching cycle. Charging and discharging this capacitance results in losses proportional to switching frequency. The Coss of a GaN FET is smaller than the capacitance of an equivalent Schottky diode. Therefore, a GaN FET bootstrap has less recovery losses than a Schottky diode and is more efficient at high switching frequencies.