Design and evaluation of a 5-W X-band PA using a low-cost plastic-packaged GaN transistor
Gallium nitride (GaN) discrete transistors suitable for operation at microwave frequencies are now commercially available from a number of vendors. The higher breakdown voltage and maximum junction temperature of GaN transistors make them well suited to the realisation of high power amplifiers.
Most commercially available GaN transistors intended for X-band operation are offered as bare die or in ceramic packages. The use of low cost over-moulded SMT plastic packages for such discrete devices has the advantage of greatly simplifying handling and assembly as well as providing a significant reduction in product cost. There are obviously challenges in developing SMT packaged microwave power transistors: an assembly approach that avoids excessive RF performance degradation must be found; thermal performance must be adequate; the assembly must be very repeatable to provide consistent part to part performance; and accurate device models are required to account for the performance of the packaged part. Here it is demonstrated that excellent performance can be realized at X-band from such an SMT packaged part, in this case the TGF2977-SM from Qorvo, shown in Figure 1.
The TGF2977-SM is a 0.25µm GaN device of 1.26mm gate width, housed in an over-moulded plastic package measuring just 3- x 3-mm. The package incorporates a solid copper base to ensure good thermal and electrical contact to the PCB ground.
Design of PA
The performance of the transistor, mounted on the chosen PCB material, was measured; drain bias was 32V with a quiescent bias current of just 25mA. The chosen PCB material was Rogers 4003, 0.008-inch, with 1oz (25g) final metallization, and copper-filled through-PCB vias were used for transistor grounding to provide improved thermal performance.
The measured maximum available gain of the transistor showed breakpoints occurring at 4.8 GHz and 12.5 GHz, indicating that the band of interest lies well within a region of unconditional stability and that a maximum gain of around 14 dB is available. The final amplifier gain will obviously be less than this, due both to real losses associated with adding the matching and bias networks and to the fact that the amplifier output will be power-matched rather than matched in a conjugate fashion.
Although the device is unconditionally stable across the whole of X-band, at frequencies above 12.5 GHz and below 4.8 GHz it is potentially unstable, so steps must be taken to ensure stability in these regions. There is also a significant amount of low frequency gain and it is good design practice to suppress this to avoid potential low frequency stability issues in the final amplifier.
Large-signal load-pull data for the mounted device demonstrated that at 9.4 GHz the device is capable of delivering around 37.5 dBm output power at 3 dB compression. The target load impedance to present at the output reference plane was found to be 12.98Ω – j9.39Ω, and the target source impedance to present at the input reference plane was found to be 11.64Ω – j55.75Ω. The reference planes are set at the edges of the SMT pads required for attachment of the transistor to the PCB.
The amplifier schematic is shown in Figure 2 – all passive components are 0603. The bias networks for both the drain and gate make use of radial stubs which provide a short circuit at mid-band. This was the preferred means of implementing an RF short circuit as it offers better performance, tolerance and bandwidth compared to a shunt SMT capacitor. It also provides a convenient point at which other bias decoupling components can be added without affecting in-band performance. These components are used to provide lower frequency supply decoupling and stabilization, and significantly reduce the low frequency gain.
The short circuit provided by the radial stub is subsequently transformed to an open circuit by a narrow high-impedance quarter-wavelength line, such that the drain and gate bias networks have almost no effect in-band. This offers some flexibility in positioning the bias networks: convenient points in the distributed matching networks were chosen that were beneficial for both the layout and out of band performance.
A damping circuit was employed at the input to the amplifier, which was a further measure for ensuring low frequency stability and gain reduction. The circuit consists of two parallel short-circuited stubs in front of a parallel arrangement of a capacitor and a lossy inductor, formed by a resistor and two narrow lines. The position and topology of the circuit was carefully chosen to provide maximum low-frequency rejection with minimum impact on in-band performance. An important feature of the topology is that it avoids presenting any undesirable impedance to the device which could lead to oscillation.
The DC blocking capacitor employed at both the input and output was a high quality 0603 multi-layer component. 2-port s-parameter simulations were carried out using a representative model which included the effects of the mounting pads on the chosen substrate. The value selected was the one which gave the lowest in-band insertion loss.
Input and output matching networks were implemented as distributed 2-section low pass structures. Moving backwards from the input reference plane, the input matching network is essentially: shunt C, series L, shunt C, series L. In a similar way, moving forwards from the output reference plane, the output matching network is essentially series L, shunt C, series L, shunt C. It is worth noting that an unbonded pin was conveniently used as an open-circuit stub to form part of the first input shunt C. This in Figure 3, which shows the layout of the final RF metal work.
A photograph of the final assembled amplifier is shown in Figure 4. The 0.008-inch thick Rogers 4003 PCB is mounted onto an aluminium alloy T-carrier, which in turn is mounted onto a heat-sink. A fast drain switching circuit is provided on the same PCB. This area is coated in green solder resist which is omitted from the RF section.
An end plate mounts onto the rear face of the T-carrier through which bias and control is applied. The blue wire is for the gate bias, the red wire for the drain bias, the yellow wire enables the drain switching and the black wire is 0V. The RF input and output have been designed to mate with edge-mount South West Microwave connectors. Note the hole in the front face of the T-carrier – this is to allow a thermocouple to be placed underneath the packaged device. Thermal analysis determined that the thermal resistance between the thermocouple and the package (through the PCB) was 8°C/W, which allows the package temperature to be conveniently calculated for a given power dissipation.
Measured Performance
Small signal s-parameter measurements were carried out on the fully assembled amplifier at package base temperatures of -33°C, 25°C and 85°C. These measurements were carried out under CW conditions, and the results are shown in Figure 5. The s-parameter data demonstrates that the final amplifier has a small signal gain of around 11 dB across the band, which varies by around ±1.5 dB over temperature. The input return loss is hardly affected by temperature and is better than 15 dB across the band. The output return loss is nominally around 9.2 dB and varies by around ±1 dB over temperature.
Large signal measurements were also carried out over temperature. These measurements were performed under pulsed conditions and utilized the on-board drain switching circuit, which provided a turn-on time of just 20ns. The duty cycle was 10% and the pulse width was 500µs. The RF envelope during the 500µs pulse is shown in Figure 6 for the amplifier running at 3 dB compression at mid-band. Clean, fast edges were evident, and there is very little power drop across the pulse period.
During the large signal measurements, the input power was swept from around 10 dB back-off, up to and slightly beyond, 3 dB gain compression. Over the power sweep, the dissipated power in the package increases so the package temperature increases. The gain under linear conditions reduced with temperature from 12.5 dB at low temperature (-33°C) down to 10.1 dB at high temperature ( +85°C), with a 25°C nominal room temperature value of around 11.1 dB. The results were taken at mid-band. 3 dB gain compression occurred at an input power of 28 dBm at low temperature, 29 dBm at nominal temperature and 29.5 dBm at high temperature. A Pout versus Pin plot indicated that the mid-band output power at 3 dB compression is nominally 37.1 dBm, varying by just ±0.2 dB over temperature.
Drain efficiency was also measured, showing that at 3 dB compression, the nominal drain efficiency is 57%, while at low temperature it is around 55% and at high temperature around 54%. Power added efficiency (PAE) at 3 dB compression was 48.5% at nominal temperature, around 48.5% at low temperature also and around 44% at high temperature.
The key performance metrics at 3 dB compression, at nominal temperature across the band of interest, are summarized in the plot in Figure 7.
Summary
This article has demonstrated that excellent performance at X-band can be realized from a GaN transistor housed in an over-moulded SMT plastic package such as the TGF2977-SM from Qorvo. A single-stage power amplifier based on this device was designed for optimum performance in the range 9.3 – 9.5 GHz. At mid-band under nominal conditions, the amplifier has over 11 dB small signal gain, and provides over +37 dBm output power at 3 dB compression with a corresponding drain efficiency of 57%.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
