Rochester Institute of Technology
EE717 Tech Memo
To: From: Date: Re:
Dr. Venkataraman Alexander Ulik (Electrical Engineering) Due 2/09/12 Project #4 Microstrip Components
Abstract
The objective of this project was to investigate the properties of several different types of microstrip power networks. The layout and impedance properties of each of these components give the output ports special behaviors that are useful in modern microstrip configurations. The Quadrature Coupler, Circular Hybrid Coupler, and the Wilkinson power divider were investigated, with simulations done in Ansoft Designer. The simulation results were consistent with our expected behaviors for each network.
Theory
Each of the types of power networks being studied contains at least one plane of symmetry. Due to this, the networks can be analyzed using Even/Odd mode analysis. The Quadrature Hybrid contains two planes of symmetry, so it can be analyzed using Even/Odd mode analysis, or a superposition of 4 different modes. Both the Ring Hybrid and Wilkinson couplers are analyzed using Even/Odd mode analysis. Their respective S-matrix results can be seen in Equation 1:
Equation 1: S-matrix results from Even/Odd mode analysis
Using the provided schematics for the given construction of each coupler, Each circuit is constructed in Ansoft Designer and simulated using a frequency sweep around the center frequency of 10GHz. Using the microstrip length calculator, the segments are designed around the physical properties of duroid ( = 2.2; h = 1/32) and Z0 = 50. The final lengths of each segment were tuned accordingly once the design was completed. The completed circuits of each coupler are included below.
Results
Figure 1 - Quadrature Coupler Loss
Figure 2 - Quadrature Coupler Phase Difference
Figure 3 - 180 Ring Hybrid Loss
Figure 4 - 180 Ring Hybrid Phase Offset
Figure 5 - Wilkinson Power Divider Loss
Figure 6 - Wilkinson Power Divider Phase Offset
Sds
Figure 7 - Ring Hybrid Circuit
Figure 8 - Wilkinson Divider Circuit
Figure 9 - Quadrature Circuit
Discussion & Conclusion
When looking at the resulting S-matrices for each coupler design, we can see different properties emerge depending on which port(s) are fed with inputs. The Ansoft design simulations are meant to verify these mathematical conclusions, as well as for tweaking element values during analysis for better performance. In Designer, tuning variables were used for the widths of each microstrip circuit element (after they were initialized with Designers microstrip length calculator). This resulted in the final element values to be slightly different than the predicted; however this allows for better performance and closer matches to the center frequency. For the Quadrature coupler, the resulting S-matrix tells us that the device is lossless with matched ports. The matrix also tells us that with only power on one port, there will be another which is completely isolated. The power entered here will also be evenly distributed over the other two non-isolated ports. However, the phase offset will be j (90 degrees) between the two output ports. Figure 2 shows this phase relationship holds in simulation, as well as the power characteristics in Figure 1. Looking at the Ring Hybrid coupler, we notice that it is in-fact not lossless, and is a 3dB coupler. Again, the outputs will be out-of-phase; However, the phase difference will depend on the exact ports used for input. One input will deliver power to 2/3 of the outputs directly in phase, while the other port will produce power exactly opposite phase of each other. Another interesting property of the Ring Hybrid is the ability to use two outputs as summing inputs, which will produced the resulting signal in port 1. Figure 4 from the simulation shows the phase properties around the center frequency. Finally, the Wilkinson power divider is built and simulated. The Wilkinson circuit is designed to be a nearly ideal 3dB power divider. The resistor at the output allows the input to be matched at each output, as well as isolating the inputs from each other. The simulations in Figure 5 also show how the power is divided equally between the two ports, as well as showing the outputs are 90 degrees out of phase from the input. The simulations also demonstrate how ports 2 and 3 are (relatively) isolated due to their high loss between each other. Overall, the simulations for the first two couplers were extremely accurate, giving results as ideal as possible. The Wilkinson power divider coupler, however, produces very small values of attenuation in comparison to what its S-matrix would suggest, especially between ports 2 and 3 where the attenuation should be much higher. The simulation offers some good behavioral insight, however the model will need to be looked into more, possibly redesigned with different simulation components, in order for it to be as accurate as the first two.