The small antenna itself has an $S_{11}$ of -3 dB around MICS band (verified with simulation and network analyzer). In application, the antenna is mounted on a transceiver PCB of 30*15*0.8 mm which has 2 metal layers. From simulation, including the PCB into the antenna model (by modeling it as a thin cubic with 2 metal faces) made the $S_{11}$ reach to -15 dB, with an increase of 50 MHz in resonant frequency and the $S_{11}$ curve gets a lot sharper. The small antenna is quite poor as a radiator because there is a dimension restrict for the device. I guess adding the PCB just improves impedance matching with no contribution to radiation. My question is how should I conclude the influence from the PCB. Is it a grounding effect, or coupling, or extra impedance?
-
$\begingroup$ What kind of PCB material? Glass Epoxy FR4, or GC10? These have dielectric constants of perhaps 4.5. At 400 MHz, it is a bit lossy, so would be a dissipative model (Resistance) and an increased capacitance between those metal layers. $\endgroup$– glen_geekCommented Feb 28, 2018 at 19:12
-
$\begingroup$ The PCB dielectric is FR4. $\endgroup$– luwCommented Mar 1, 2018 at 10:52
-
$\begingroup$ In simulation tools, I have tried with 1-layer and 2-layer-metal for the PCB. The resulted S11 curves didn't show much difference. $\endgroup$– luwCommented Mar 1, 2018 at 10:54
1 Answer
The small antenna is quite poor as a radiator because there is a dimension restrict for the device.
It is important to separate cause and effect when designing antenna systems. Recall that the directivity of an infinitesimally small dipole has a directivity of 1.5 compared to a full size 1/2 wavelength dipole with a directivity of 1.64. That is less than a 0.5 dB difference in gain if the antenna systems have the same efficiency.
The key issue to focus on with small antenna systems is efficiency (Rr and Rl). This puts the antenna, its matching circuits, transmission lines, and the PA design under scrutiny to wring out every bit of possible efficiency improvement. For example, I often see students and experienced engineers struggling to match a 50 ohm PA output to a low impedance, compromise antenna. What they fail to consider is that the naturally low output impedance of the PA is undergoing an unnecessary and lossy double impedance transformation. So look at the problem from a systems approach and not just as an antenna design exercise.
In the end, you are trying to optimize a compromise based on cost, performance, reliability, and time to market.
My question is how should I conclude the influence from the PCB. Is it a grounding effect, or coupling, or extra impedance?
You haven't given adequate geometric details to fully consider this question. But in general, the PCB material will act as a dielectric that is in the near field of the antenna. As such it will exhibit resistive losses (think efficiency!), it can alter the distributed inductance or capacitance of the antenna thereby shifting the optimal operating frequency (from a matching perspective), and it can introduce phase delay in the near fields.
...made the $S_{11}$ reach to -15 dB, with an increase of 50 MHz in resonant frequency and the $S_{11}$ curve gets a lot sharper.
The $S_{11}$ parameter is only an issue when considering the matching circuit. As a secondary indicator, however, when $S_{11}$ remains low over a broader frequency range, the Q of the antenna has been reduced. If all other factors are equal, this is a strong indicator of increased losses in the antenna (system). Recall that the "best" and broadest $S_{11}$ parameter is realized with a fully resistive, non-radiating load.
