LTE small cell base station antenna matched for maximum efficiency
This application note demonstrates how Pulse Electronics (Pulse) was able to design, tune and optimize their antenna systems through a combination of AWR’s Microwave Office® circuit design software and Optenni’s Optenni Lab™ matching circuit generation and antenna analysis software. The end result of this cross-company design flow yielded a higher performing product, a more cost effective design, and faster time-to-market.
While the thought of matching circuit design sounds simple and quite appealing, there are a few guidelines that must be followed. First, it is important to optimize for efficiency and not for best possible impedance match. Second, realistic component models of inductors and capacitors should be used in the matching circuit design, as the differences between an ideal and a real component are often substantial. Third, the sensitivity of the matching circuit with respect to component tolerances should be well accounted for and verified.
Novel antenna design
The antenna design showcased in this article is based upon Pulse’s work on small cellular base station antennas using directional patch radiators with two feed ports that have vertical and horizontal polarizations. The operating frequency for the antenna system is LTE Band 8, 880-960 MHz. While patch antennas are commonly known and widely used throughout the antenna industry, Pulse’s use of modern design tools has introduced a new antenna design optimization process and work flow.
Figure 1: The initial prototype was built and tested by Pulse to collect baseline data prior to starting the simulation work.
Prototype antenna
One of the first design challenges Pulse faced was how to integrate the feed structure into the constraint space. To start, an aperture coupling structure was selected because of its traditionally good port-to-port isolation characteristics due to the orthogonal port excitation resonant modes. However, because of the low operating frequency of 880-960 MHz, the physical size of the feeding aperture was deemed too large (exceeded constraint space requirement) if made symmetrical. Therefore an asymmetric configuration resulted: Port 1 feed aperture was of optimal length and Port 2 feed aperture was short, yet tuned to frequency by widening the arms at the end of aperture (Figure 2).
Figure 2: Port 1 and Port 2 asymmetrical feeding aperture configuration.
Note that the traditional iterative process for design optimization of this feed structure would have been one in which multiple prototypes were constructed, measured, and tweaked through a trial-error-correction process. The way in which the Pulse final asymmetrical design was uncovered is novel in and of itself. The alternative approach relied upon virtual prototyping, i.e., the use of simulation (synthesis and analysis) software. Through simulation, Port 2 matching was accomplished simply by adding an LC-matching circuit and the original antenna and feed element designs were not impacted nor altered at all. Suffice it to say this saved significant design time.
Antenna simulation
The antenna itself was simulated with a target bandwidth of 880-960 MHz using AWR’s Analyst™ 3D electromagnetic (EM) simulator within Microwave Office. For this design, a full 3D EM simulator was necessary given that the feed lines were supported by a narrow printed circuit board (PCB) substrate with such finite dielectrics that edge couplings had to be accounted for.
The initial results (Figure 3) revealed that while Port 1 was inherently well matched, Port 2 required a matching circuit to tune the resonance. The isolation between the ports was very good, in the -40 dB range (Figure 4).
Figure 3. Port 1 and Port 2 return loss of the initial antenna design.
Figure 4: Isolation between Port 1 and Port 2 of the initial antenna design.
Matching circuit design
Next, Optenni Lab software was employed for the matching circuit design for Port 2. Optenni Lab provides an easy-to-use interface for direct optimization of antenna efficiency that accounts for optimization over a wide range of vendor libraries, tolerance analysis, and more.
The antenna impedance data was read from a Touchstone file, the operation frequency ranges were input, and the desired number of components and the desired component series were selected. Within a matter of seconds, Optenni Lab provided multiple optimized matching circuit topologies. The resultant matching circuit (Figure 5) was synthesized to maximum efficiency over the band. The remaining fine-tuning steps involved included the layout details for placement of the discrete components.
Figure 5: The optimized three-element matching circuit for Port 2 using the Murata GJM15-series capacitors and LQW18-series inductors.
The parallel-series layout near Port 2 (Figure 6a) was grounded by folding a strip around the edge of the PCB and soldering it to the ground plane. This, however, changed the matching because the shunt capacitor grounding involved inductance as well, and there was a delay of a couple of degrees between the first and last elements (Figure 6b). The final implemented matching circuit of the prototype shown in Figure 6b reflects the change in the matching due to these effects (Figure 7).
Figure 6a: Layout detail of the matching components placement.
Figure 6b: Construction of realized matching circuit in measured prototype.
Figure 7: Optimized Port 2 return loss with ideal versus real connectivity of the matching components.
While the ideal versus real connectivity difference appeared to be rather small, the power delivered to the antenna (Figure 8) dropped by 0.2 dB over the band. Further fine-tuning of the design identified a more suitable and appropriate choice of components, which resulted in a reduction of the efficiency loss to 0.1 dB. The matching component values after this fine-tuning were determined to be 5.6 nH series, 2.2 pF parallel, and 2.7 pF series from the same Murata component series as before.
Figure 8: The layout arrangement of the matching components reduces the efficiency by 0.2 dB (dashed line). Re-optimization of the component values corrects the situation by 0.1 dB (green line). Click to enlarge.
Figures 9a and 9b depict the measured prototype antenna efficiencies with and without the matching circuit at Port 2.
Figure 9a: Measured prototype efficiency with matching circuit.
Figure 9b: Measured prototype efficiency without matching circuit.
Measurements
Finally, the antenna prototype was manufactured and measured at Pulse. Figure 10 shows the simulated and measured port impedances on a Smith Chart without the matching circuit. Here, Port 1 is neatly matched over most of the ideal bandwidth.
Figure 10: Simulated (dashed line) and measured (solid line) port impedances without the matching circuit.
Figures 11a and 11b show the measured prototype return loss and isolation with and without the matching circuit. Figure 12 shows the corresponding results with the matching circuit, and Figure 13 shows the isolation worsens as the resonance for Port 2 enhances. The agreement between the simulations and measurements overall was good.
Figure 11a: Measured Return loss and isolation with matching circuit.
Figure 11b: Measured Return loss and isolation without matching circuit.
Figure 12: Simulated (dashed line) and measured (solid line) port impedances with the matching circuit.
Figure 13: Isolation of the ports in the final design. Dashed line = simulation, solid line = measurement.
A closer look at the frequency shift between the simulated and measured data, as shown in Figure 14, led to a further investigation, largely for educational sake. Statistical analysis of the discrete component tolerances showed a relatively stable performance. Yet, a change of 1.25 mm – or two degrees – in the feed line length could sufficiently explain the difference. This then revealed that care should be taken to account for the dimensions of the structure and how this feed line length serves as a straightforward means to tune the matched antenna frequency, often by several tens of MHz. In the end, the measurements confirmed that the designed matching circuit improved the efficiency of Port 2 radiation by more than 20 percent and the antenna gain by about 2 dB.
Figure 14. Port 2 return loss with matching components. Solid line is the measurement. Dashed magenta line is simulated result showing the statistical analysis due to component tolerances. Dashed blue line is simulated result with 1.25 mm longer feed line. Click to enlarge.
Conclusions
The virtual software design methodology described in this application note provides a “first-time-right” matching circuit design flow that is more efficient and cost effective than traditional methods, and equips antenna designers with quantitative guidelines for antenna frequency tuning that ensures a higher quality product. It was particularly insightful for the design of Pulse’s novel dual-feed single radiator aperture-coupled patch antenna for LTE small cell base stations.
Total radiation pattern of port 1.
About the authors
Kimmo Honkanen, RF Engineer, Pulse Electronics
Kimmo Honkanen graduated in 2006 from the Kajaani Polytechnic Communications and Transport Degree Programme with a degree in Information Technology. He has been an RF Engineer for 5 years at Pulse Electronics Kempele, Finland.
Jussi Rahola, Managing Director, Optenni Ltd
Jussi Rahola obtained his D.Sc. (Tech.) degree in numerical mathematics from Aalto University, Finland in 1996. He has previously worked for the CSC-IT Center for Science, Finland and for the Nokia Research Center, Finland. Since 2009 he is the Managing Director of Optenni Ltd, specializing in the development of the Optenni Lab matching circuit optimization software.
Dr. Jaakko Juntunen, Head of EM Applications, AWR Europe
Jaakko Juntunen obtained his MSc in mathematics and applied physics at Helsinki University of Technology in 1995 and he completed his doctoral degree in 2001, with a thesis focusing on the finite-difference time-domain (FDTD) method.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
