This article describes how isolation resistor temperatures are affected by phase imbalances in a real-world power combiner, followed by the results of power testing a modified back-to-back (B2B) combiner that stresses the resistors without causing a failure. A photo of a Nuvotronics 80 W, 6 to 18 GHz, four-way air-coax power combiner is shown in Figure 1a. This network serves as the foundation for the examples presented in this article. As shown in the simplified schematic of Figure 1b, the combiner is a cascaded Wilkinson design featuring five 100 Ohm isolation resistors that provide a minimum of 13 dB of isolation between inputs across the entire band. It is rated for temperatures up to 85°C and exhibits a loss of just 0.34 dB at the center frequency. This combiner uses the PolyStrata® manufacturing process to create its air-coaxial transmission lines.
Figure 1 (a) Nuvotronics combiner. (b) Simplified combiner schematic.
Figure 2 Safe operating area for diamond thin film resistors.
In a Wilkinson power combiner, removing heat from isolation resistors can be a challenge. Typically, improving the thermal path adds capacitance to the structure, which can detune the RF performance. This is especially true in air-coax structures. In the 6 to 18 GHz combiner, the thermal resistance of the isolation resistors was evaluated to be 28°C/W using ANSYS software. Future designs have identified a path toward 3°C/W and these designs may become available in 2025.
The isolation resistors that are used in the PolyStrata combiners are CVD diamond thin film products from Smiths.1 These resistors are typically configured in a 0402 package and rated to 20 W dissipation when attached to a suitable heat sink. According to the manufacturer, the resistors are rated to full power at a film temperature of up to 125°C, but they must be derated linearly to 0 percent power at 150°C. This derating curve is shown in Figure 2. The temperature and power capability of the isolation resistors limit the allowable out-of-balance conditions when a combiner is used in a solid-state power amplifier (SSPA).
HFSS/MWO “HYBRID” COMBINER MODEL
The 3D HFSS model for the five-port, four-way power combiner was reconfigured to remove the resistors. This was done to model the power combiner as closely as possible to the actual configuration while allowing access to resistors to evaluate their power dissipation under different configurations. RF ports were added to the resistor sites, resulting in a 15-port network. The HFSS model is illustrated in Figure 3a, where Ports 6 through 15 are not depicted in the overall model. Figure 3b is an example resistor, showing where Ports 6 and 7 have replaced the resistor film.
Figure 3 (a) 3D circuit model showing Ports 1 to 5. (b) Port definition at R1 resistor interface.
Figure 4 (a) HFSS model with five external resistors. (b) Power combiner block diagram.
Figure 5 75 W power stage model.
Using Microwave Office (MWO), resistors were reintroduced into the network in a “hybrid” HFSS/MWO model. MWO enables the convenience of harmonic balance for quickly assessing power performance. An MWO schematic of the rebuilt combiner is shown in Figure 4a. The block diagram in Figure 4b maps the positions of the five resistors within the network. For example, resistor R1 is closest to the common port or the output of the combiner.
The 6 to 18 GHz power combiner can use available 20 W GaN power amplifiers to achieve a 75 W SSPA power stage. A very simple linear model of such a power stage is shown in Figure 5. In this model, the A1, A2, A3 and A4 amplifiers are fed from a lossless four-way divider, providing 20 dB of linear gain to feed power into the hybrid combiner model. With 29 dBm (800 mW) of input power to the network, the output power is predicted to be approximately 74 W at 10 GHz. This is slightly less than the available 80 W, as the combiner network has an insertion loss of 0.34 dB. Lossless phase shifter elements, Φ1, Φ2, Φ3 and Φ4, are in the model to vary transmission phases between amplifiers, as they will not be ideally matched in real-world conditions. In this study, amplitude imbalances are neglected because they create less mismatch.
PHASE MISMATCH EXAMPLE 1
In this example, amplifiers A1 and A2, as well as amplifiers A3 and A4, are assumed to be phase-matched pairs. When there is a phase difference between these pairs, all the mismatched power dissipates in a single resistor, R1, as shown in the schematic of Figure 4b. For this example, the phase difference between the pairs was set to 0 degrees for the in-phase case and then 20 degrees for the out-of-phase case.
The output power drop resulting from these phase conditions is shown in Figure 6. When the phases are aligned and the phase difference between the amplifiers is 0 degrees, the output power is 74.2 W at the band center of 12 GHz. When the amplifier pairs are out of phase by 20 degrees, the output power decreases to 72 W.
Figure 6 Example 1 power output response resulting from phase mismatches.
Figure 7 illustrates the power dissipated in the R1 resistor during CW operation. For the in-phase condition, no power is dissipated in R1. However, in the case where the amplifiers are 20 degrees out of phase, R1 would dissipate 2.2 W at the worst-case value of 10.1 GHz over the 4 to 20 GHz frequency band.
Figure 7 R1 power dissipation for the Example 1 phase mismatch conditions.
Using the power dissipation response and the 28°C/W resistor thermal resistance of the resistive film calculated earlier, the temperature increase in the isolation resistor in the model can be calculated. Assuming an 85°C baseplate temperature, the 20-degree phase difference creates the thermal profile shown in Figure 8. The maximum temperature increase is 146°C at 10.1 GHz.
Figure 8 R1 film temperature profile caused by a 20-degree phase mismatch.
