RF PA circuit design based on gallium nitride (GaN) devices

zqy1111

RF PA circuit design based on gallium nitride (GaN) devices [Copy link]

We use radio frequency (RF) communications in our daily activities, whether calling a relative, texting a friend, or even reading this blog post on a mobile device. There are many signals traveling quickly through the air, but where do most of these signals originate? Most RF communications originate from cellular towers or wireless base stations, such as the one shown in the figure below.

Mobile phone base station

There are so many components inside these base stations that a full summary would take the form of a PhD thesis! But rather than covering every detail, I will focus on one component that is critical to any base station: the power amplifier (PA).

As you might have guessed, the purpose of the PA is to amplify a low-power RF signal into a high-power RF signal that is driven into the base station transmitter.

Before any RF signal is injected, a direct current (DC) voltage (Vgate) is applied to the gate of the PA and adjusted until the desired drain current flows through the PA. This current is often referred to as the quiescent current - the current that flows when no RF input is present. The value of this quiescent current can be selected to suit the end application, including the modulation system and the device operating level. The figure below is a simplified schematic of a typical PA setup.

Simplified PA schematic

To better understand how gate voltage and quiescent current affect RF (AC) performance, you can replace the PA with a metal oxide semiconductor field effect transistor (MOSFET) model, yielding the following expression:

(MOSFET in saturation) (1)

(2)

Figure 3 is a graphical representation of the two equations. Driving a small RF input signal superimposed on the DC gate voltage generates an AC drain current ΔIds. This AC current oscillates around the quiescent current value Idsq (see Figure 3a). You can use the MOSFET transistor IV curve and load line analysis to find the corresponding AC drain voltage ΔVds (see Figure 3b). When determining the relationship between the AC drain voltage and the AC drain current, Equation 2 can be further simplified to Equation 3.

(3)

In Figure 3a, using the slope of the transconductance to calculate the resultant gm, you can further generalize the expression to:

(4)

The voltage gain of the amplifier, ΔVds/ΔVgate, is therefore interpreted as –Rs gm, which translates to:

–Rs 2 Idsq/(Vgate – Vth)

The above expression mainly illustrates that the gain of this configuration is directly proportional to the quiescent current Idsq. In addition, Idsq can be selected to ensure that the output voltage swing is not limited by saturation; this is why you should choose an Idsq value before RF operation, which requires a specified DC gate voltage.

Figure 3: (a) MOSFET Vgate vs. Ids curve

(b) MOSFET common source load line analysis

Traditionally, most PA bias systems have been discrete solutions, some as simple as a potentiometer (variable resistor divider) on the PA gate. Newer approaches leverage the accuracy and digital interface of precision digital-to-analog converters (DACs) and/or digital potentiometers. Different PA technologies such as laterally diffused MOSFET (LDMOS), gallium arsenide (GaAs), and gallium nitride (GaN) require different levels of gate voltage for device operation. For example, GaN and GaAs require a negative bias system, while LDMOS requires a positive voltage for device operation. For this reason, many PA bias solutions now also include a DAC with a bipolar range as part of the package.

PAs exhibit nonlinearities during device operation that are primarily temperature dependent. These nonlinearities can significantly impact performance as they lead to unpredictable device behavior.