Mastering antenna impedance measurements – Volume I
From this article, you will learn about:
- Design challenge: sources of error in vector network analyser (VNA) impedance measurements affecting the matching circuit synthesis and verification
- VNA measurement methods recommended for the frequency ranges up to 2-3 GHz
- Port extension method
- On-board calibration method
- Impedance measurement example and verification
Introduction
Your RF and antenna design flow may start from a 3D EM simulation model of an antenna, selecting a suitable catalogue antenna, or directly from an antenna prototype. Whatever is the end-to-end design flow, prototype measurements and verification steps will be eventually needed to match the design with the target specification. Optenni Lab can greatly accelerate your design flow with the matching circuit synthesis, and it can also interface with VNAs in real time. A shortcut to guaranteed success in circuit synthesis and prototype verification is the correctly measured impedance data.
Measuring antenna impedance is hard. You might protest and say: “how come, just take a VNA, connect it to the antenna and there you go!”
The keyword
here is connect –
how do you
actually connect
to the antenna?
Perhaps your protest in fact describes measuring a chain of antenna, transmission line and a connector? I agree – this is easy. But it is surprisingly tricky to get around the influence of the connector and the transmission line between the connector and the antenna.
Often there are no connectors anyway. Let’s just think of an SMD antenna module, or antenna formed by shaping the metallization on a PCB. The antenna should be matched such that the RF chip RF node is loaded by a specified impedance, typically 50 Ohms. Both the matching circuit design and the result verification require measurement of the impedance on a specific location on the PCB (see Figure 1).
Figure 1: Sketch of on-board measurement setup. Location A. for antenna impedance measurement, and location B. for antenna + matching circuit impedance measurement.
In Figure 1, the location A. refers to the point where antenna impedance is measured (before matching circuit design) and that measurement gives the input data for Optenni Lab circuit synthesis. Impedance measured at location B. refers to the point where the impedance of antenna and matching circuit combination is measured, and that measurement gives the input data for the design verification. Measuring the impedance correctly at these two points is critical for the matching circuit design and verification, as the errors in antenna impedance measurement means that incorrect input data is used in matching circuit synthesis leading to potentially huge errors in realized impedance, and errors in design verification impedance data hampers making correct conclusions about the success of the matching.
In this article we focus on the antenna impedance measurements (Location A. in Figure 1), but the same methods are valid also for the design verification measurements (Location B. in Figure 1).
Reference plane in impedance measurements
The reference plane of the VNA measurement should be established at the specific point on the PCB where the impedance needs to be determined. Establishing a reference plane is most commonly done by attaching calibration standards to the ends of the VNA measurement cables (usually with SMA connectors) during calibration, which establishes the end of those cables as your calibration reference plane. This is a standard procedure in the vast majority of the VNA impedance measurement cases. However, as presented in Figure 1, the SMA connector at the end of the VNA cable is not the targeted measurement point on the PCB due to the pigtail cable, transmission lines and component pads on the PCB. And this difference may be a great source of measurement error especially on high frequencies.
We describe here
two methods which
usually are
sufficiently accurate
for impedance
measurements at
2-3 GHz frequencies.
As we often deal with non-connectorized designs, we describe the use of so-called pigtails, which are thin coaxial cables with SMA connector in one end, and open-ended inner conductor extension in the other (Figure 2).
Figure 2: Pigtail overview and tip close-up.
The SMA end of the pigtail mates with the VNA measurement cable, and as described above, the routine calibration of the VNA sets the reference plane at the SMA connector of the VNA cable. The inner conductor extension of the pigtail is suitable for positioning and soldering on the PCB. The only problem now is that the calibration is in the wrong end of the pigtail which affects the S11 response (Figure 3). How to move the measurement reference plane to the pigtail tip?
Figure 3: S11 response of a pigtail after routine calibration, fmax = 8.5 GHz.
Port extension method
The first method is readily available in most VNA models, and it is called port extension. It is based on measurement of the group delay of the open-ended pigtail, and computationally compensating – or de-embedding – the delay. The benefit of the method is that it is instantaneous to apply. The drawback is that its accuracy is questionable, because it only works perfectly if the group delay of the pigtail is constant over frequency, and this is never the case in reality. The inconsistency is revealed by looking at the S11-response of the open-ended pigtail: it is not a dot at the open circuit point in Smith chart but it spreads over a larger or smaller area (see Figure 4).
Figure 4: S11 response of a pigtail after port extension, fmax = 2.5 GHz.
Nonetheless, port extension is OK for low frequency measurements, the upper frequency limit depends on the pigtail type and accuracy criterion, but typically 2-3 GHz is reasonably safe.
On-board calibration method
The second method is more laborious, but more robust and independent on the pigtail type. We call it “on-board calibration”: here the calibration standards are directly built on the PCB at the desired reference plane. In other words, an open circuit, short circuit and 50 ohm impedance load are created in turns by having a gap in the RF line (footprint of a serial component in the RF line), by soldering a short circuit to ground (GND) and by soldering resistors as a 50 ohm load to the PCB at the position of the reference plane (Figure 5). Further benefit of the method is that the pigtail can be soldered on the PCB some distance away from the reference plane which can be helpful if the pigtail shell is suspected to couple with the antenna or the reference plane is tricky to reach with pigtail e.g. due to tight enclosure.
Figure 5: Building ad hoc on-board calibration standards. Calibration standards are created by utilizing the pads of a serial component footprint in the RF line. For short and load, the gap in the RF line disconnects the antenna and calibration loads.
The VNA measurement is calibrated using these calibration standards and ideal termination models. The limitation of the method stems from these models: the physical calibration standards are not ideal. If we have the possibility to do EM simulation, we could try to simulate the terminations and use EM-simulated responses in place of ideal models. Unfortunately, RF resistor S-parameter models are seldom available, and it may thus be difficult to reach accurate calibration standard models even if EM simulation is used. Nonetheless, also with ideal termination models one should easily reach 3 GHz and perhaps even 4 GHz or higher.
Note also that the on-board calibration method may require that one includes several suitable measurement points in the PCB layout design, or necessary hardware modifications of the measurement prototype (refer to the Location A. and Location B. in Figure 1).
Impedance measurement example and verification:
We used these two methods to match the same antenna for Band 3 (1710 MHz – 1880 MHz). The reference plane was located at the series component gap shown by the dashed line in Figure 5. The unmatched S11-response in Smith chart due to both methods is shown in Figure 6, the markers are set at 1 GHz steps to help the comparison.
Figure 6: Unmatched calibrated S11-response of test antenna using port extension and on-board calibration methods.
We can see that the methods agree quite well up to 4 GHz in this case, beyond which the difference is big. The antenna is almost self-tuned, and the matching is rather easy. Looking at the on-board calibration case, after adjusting the component interconnect models, the agreement between the simulated prediction and measurement is phenomenal (Figure 7).
Figure 7: Simulated vs measured S11-response of matched test antenna using the on-board calibration method.
Conclusions
In this blog article, we highlighted the importance of correct calibration and measurement reference plane determination in antenna impedance measurements. The quality of antenna measurement data is a key aspect for optimal matching circuit synthesis and prototype verification. Two methods for controlling the measurement reference plane were introduced: port extension and on-board calibration. They are recommended for the frequency ranges up to 2-3 GHz.
How about higher frequencies, for example UWB or 5 GHz WLAN? Controlling the measurement and simulation model accuracy gets harder as new parasitic phenomena enter the scene, requiring a refined approach. Port extension or on-board calibration, using ideal models, no longer offer a solution. Dealing with higher frequencies will be the topic of a future blog article.
Account Manager
jaakko.juntunen (at) optenni.com
