Design Optimization by Including the Effects of PCB Layout and Transmission Lines in Optenni Lab
From this article, you will learn about:
- Design challenge: transmission line effects on antenna matching
- Design flow and single-port example
- Matching circuit synthesis with ideal components
- Matching circuit synthesis with PCB layout
- Tolerance analysis
- Multiport layout models and Schematic mode in Optenni Lab
Introduction
As is well known, any practical realization of an antenna matching circuit needs something where the matching elements like inductors and capacitors are placed. Usually a printed circuit board (“PCB”) is used for this purpose. As an example of an antenna prototype, see Figure 1.
Figure 1: Antenna prototype with a temporary coax soldered to the matching area for measuring the raw antenna impedance, without any matching.
In radio engineering, any physical object with a dimension has the potential to act as a transmission line. This means that the radio signal propagating in the transmission line will experience at least a phase shift, and typically also some attenuation and other types of distortion (e.g. crosstalk). The phase shift can alter the matching results tremendously. A 180 degree phase shift can turn an open circuit to a short circuit and vice versa, and smaller phase shifts generate less dramatic, but still very substantial sources for errors.
Most commonly, the transmission lines used in antenna matching are well known for the RF and microwave engineers – they are microstrip lines, coplanar waveguides or striplines. Optenni Lab can support designs with these transmission line types in various ways. Also more complex layouts, even in 3D, can be studied, if the S-parameters of the complex layouts can be derived, for example through EM simulations.
Example of transmission line effects
It is easy to verify the effects of a transmission line (or a PCB in general) to the matching synthesis results.
Let’s study the example antenna data (measured_antenna.s1p) which is available in our Tutorial compendium of the Optenni Lab installation (Optenni Lab’s Help > Documentation, Tutorial Chapter 1.1).
By performing a three-element discrete component matching at 1600 – 1800 MHz and 4000 – 4200 MHz to said .s1p file, we get the frequency response, and the best circuit comprising ideal components as shown in Figures 2 and 3.
Figure 2: Response for the two-band matching circuit with no physical dimensions
Figure 3: The best three-component topology with zero dimensions, for two bands
It is evident that the matching circuit is ideal in terms of the elements, but also in terms of the connectivity between the matching elements. There are no physical dimensions in the elements. Further, it is as if Port 1 and Antenna 1 are placed with zero distance from each other, in an infinitesimally small space which also contains the matching elements. Clearly, this is not the way things will happen in reality. So how do we add reality to all this?
Optenni Lab has basically two ways to inject layout information to the design:
1) By using the discrete transmission line elements (transmission line, microstrip line, stripline, coplanar waveguide etc.), or
2) By using S-parameter blocks representing the layout (the S-parameters derived e.g. from a 3D EM simulation)
- For Single Port Matching and for Multiport Antenna Matching modes, you have the special Layout Block element (Figure 10). This element assures that you have one input and one output per layout (to not generate branchings, as these two modes need one external and input node per matching network and thus are not compatible with branched signal paths
- For the Schematic mode, you have the full liberty of representing layouts with the generic S-parameter Block (Figure 12)
So, let’s see how things turn out by adding some microstrip lines to the design to make it a bit more realistic, as presented in Figure 4.
E.g. by having short 1 mm microstrips with Z0 = 50 Ohms between each of the elements and the input and output of the device, we have the following topology:
Figure 4: A matching circuit with non-zero dimensions
The comparison of results is in Figure 5. As can be seen, even at the lower band, there is a maximum of 1.6 dB difference in the results, and in the higher band, the results are completely off, with over 9 dB difference. This kind of behavior where lower frequencies (and thus, longer wavelengths) are less affected, stacks up well with the RF theory, as the device gets larger and larger in terms of wavelengths when the frequency goes up, and thus the phase error is also higher at higher frequencies.
As can be seen also from Figure 5, 3 GHz is usually the boundary where transmission lines start having a really serious impact on the results. This is the classical boundary between RF and microwaves.
Figure 5: Comparison of total efficiency, zero dimension (green line) vs. four 1 mm microstrip lines (purple line). The higher the frequency gets, the more dramatic the difference is
Are transmission lines evil?
Now you might ask:
Are transmission
lines evil if they
distort the results
this badly?
Of course they are not! They are an inevitable aspect of radio engineering, and you simply have to take transmission lines into account when designing. In fact, often it is so that the PCB (including the transmission lines) becomes fixed in the design iterations, and then it is up to the designer to “live with the board”, and modify the connected inductors and capacitors accordingly.
So let’s try to alter the design by placing generic reactances to the topology at the location of the inductors and capacitor, and resynthesize the result. We get the following:
Figure 6: Reoptimized topology
Figure 7: Results compared with elements reoptimized
Looking at Figure 7, it is evident that the simple resynthesis with generic reactances (which eventually yielded the same component types, two inductors and one capacitor), a clear improvement of over 8 dB is obtained esp. at the higher band of 4 – 4.2 GHz. In general,
with Optenni Lab,
tackling the PCB
issues is easy.
Effect of transmission line tolerances
Studying the tolerance effects of the PCB is very easy when the PCB is modeled with transmission lines. For example, by zeroing the tolerance of the inductors and capacitors of circuit in Figure 7, by setting the tolerance of the microstrip width to 50 micrometers, and by running 200 Tolerance analysis samples, we get the result of Figure 8.
Figure 8: Transmission line effects can also be studied in terms of tolerance easily in Optenni Lab
Thus, there is an approx. -2.4 dB – -2.13 dB = 0.27 dB variance in the results in the lower edge of the higher band between the nominal and worst results. This can be something that needs consideration in the later design stages of the actual device.
Of course, this tolerance analysis is applicable to many other aspects of Optenni Lab, e.g. component models. Especially with library components for inductors and capacitors, this is a very powerful tool. But this is a topic of some other blog posting.
What about more complicated transmission line effects
Optenni Lab supports various transmission line models in addition to microstrip lines, for example striplines and coplanar waveguides. However, sometimes the structure is so complex that the modeling needs measurements or 3D EM simulations. For example, consider the 3D EM simulated four port structure of Figure 9.
Figure 9: A bit more complicated PCB structure modeled in 3D EM
Port 1 is the input port, and port 2 is the output port. Port 3 is a shunt port to grounding, and port 4 is a series port for a matching element in series. The grounding pins (looking like thick cylinders) are clearly visible under the lower node of port 3. Such wide metallizations are likely not well modeled with simple via models. Thus, 3D EM modeling is needed.
There are various ways to incorporate this four-port structure into Optenni Lab. In the past, until Optenni Lab ver 5.2, the only way to incorporate multiport S parameter data for the layout was the already-discussed Layout Block. The Layout Block allows any complexity for the interconnects, as long as there were two external ports, one for input, and one for output. The Layout block is still available in the current Optenni Lab version of 6.1, as it is the go-to-method with Single Port Matching and Multiport Antenna Matching. For example, for the structure in Figure 9, the Layout block could be connected as in Figure 10:
Figure 10: The Layout Block
Since Optenni Lab version 6.0, the interconnections between have been completely free in the Schematic mode. This is a remarkable improvement, as now it is possible to model arbitrary layout and interconnectivity aspects. For example, consider Figure 11 where we have a snippet of a layout, focusing on the interconnection area of a three-throw switch with ports 1 – 3, and ports 4 – 6 reserved for matching components.
Figure 11: Layout area of a switch. Both layout and switch are part of a complex mobile terminal model, with an antenna modeled separately.
This six-port layout can now be inserted into the matching task of Optenni Lab, for example, as denoted as S1 in Figure 12. Please note that the Antenna is now a two-port structure, and separately simulated (with impedance and radiation characteristics), and the switch (Switch 1) is interposed between the layout model S1 and the aperture tuner port 2 of the antenna. In addition, there are the synthesized elements (0-4 component Synthesis Block on the left) and three generic reactances for Optenni Lab to come up with proper component types (inductors or capacitors on the right) and their values.
Figure 12: Synthesis task where layout aspects are incorporated through the S-parameter block S1
It is a great advantage to be able to separate the antenna (with signal port 1 and aperture port 2) from the layout model (comprising a six-port) as now both of these portions of the complete structure (here, a mobile terminal) can be modeled quickly, with a 3D resolution that best matches each of these two simulation domains, antenna and switch layout.
Conclusions
Optenni Lab caters for incorporating the PCB and layout designs into the matching network synthesis in various ways. It is vital to consider the non-zero lengths of the transmission lines of the layout at virtually all relevant frequencies. Luckily, Optenni Lab offers many ways to do this, from simple transmission line models to a full integration of complex 3D EM models into the matching and synthesis project.
Finally, for those willing to try out Optenni Lab with PCBs and other layout aspects, the related tutorials are available in Optenni Lab’s Help > Documentation, Tutorial Chapters 2 and 5.2.
Ready for a free trial? Start here >
Contact Us for a product demo and further information on the latest Optenni Lab features.
Olli Pekonen
Sales Director
Email: olli.pekonen (at) optenni.com