Why does my antenna seem to lose performance once my system is enclosed inside a plastic casing
I have never been an antenna expert and do not pretend to be one. After reading this article, I hope you understand that ready-made antenna layouts are in most cases suboptimal in spite of what you hear and how useful a real antenna expert would be to your #IoT project.
If you are designing a wireless product, the time will come when you will need to make decisions about the antenna:
- Can you afford an external antenna at the expense of killing the design of your good-looking gateway ?
- Can you make it a wire neatly folded inside your sensor plastic casing ?
- Should you consider a PCB-printed antenna instead ?
- How about these more expensive chip antennas ?
Antennas do not like metal around
Since you have some radio-frequency background, you are already aware that antennas should be kept away from metal for optimal radiation. Metal such as metal casing, screws and bolts, batteries, railing, wiring, “captures” the electromagnetic wave like a parasitic receiving antenna and takes away some precious dB from the useful transmitted power converted into Eddy currents slightly heating your device, which is not what you want. In effect, it is recommended to keep metal outside the near-field region at a minimum distance of 1 wave length away from your antenna in order to avoid such losses, which is, of course, in many cases completely unrealistic:
- At 868MHz-915MHz, 1 wave length in the air = 34cm = 13in
- At 2.45GHz, 1 wave length in the air = 12cm = 4.8in
… which is by the way the reason why antennas are always specified with radiation patterns in free space and never in a real practical environment where you are only left with empiricism and despair.
For those of you lucky enough to have access to a VNA (Vector Network Analyzer), IF AND ONLY IF the antenna metallic environment never changes, it is possible to retune the antenna impedance and compensate for part of the losses. I remember doing so 4 years ago when an industrial customer asked my friends at #TD-Next and I to embed a #SIGFOX tracker inside the foot of a steel container:
Stan Marly and I clearly had to retune the antenna impedance having moved from 50 ohms down to 5 ohms because of the surrounding metal in the near-field region. Believe it or not, after proper matching the signal was heard by 3 different base-stations distant of 12km each, not bad for a 14dBm system transmitting from inside an 8mm-thick steel bar!
Antennas should not care about plastic
It is true that non-conductive materials do not exhibit the same effect than metal in the vicinity of antennas.
However, many non-metallic materials we use for casing such as polymers, glass, rubbers, wood are not completely passive when it comes to interacting with electromagnetic waves.
Indeed, these insulating materials exhibit the property of opposing any electric field applied to them. This property is useful when designing a capacitor or a super capacitor since it allows for higher charge densities, hence for smaller capacitors with ever more capacity.
But the corollary is that electromagnetic waves propagate slower than c the speed of light inside these materials. Since Maxwell, we know that electromagnetic waves propagate at the speed of c = 1 / sqrt( mu_0 . epsilon_0) = 3e8 m/s in vacuum.
Inside an insulating material always exhibiting a permittivity epsilon higher than vacuum, electromagnetic waves propagate at the lower velocity v = c / sqrt(epsilon_r) with epsilon_r called the relative permittivity of the material (to vacuum).
In turn, the wave length of an electromagnetic wave is shortened by the same factor sqrt(epsilon_r) in such material.
By the way, this is how tiny chip antennas work: by molding their antenna core into a highly dielectric ceramic with epsilon_r over 100, the geometry of these antennas can be reduced by a factor of 10 or more compared with their “air” equivalent. Their near-field region is also “mostly” self-contained inside their packaging, making them in theory less susceptible to losses and detuning.
It means that an antenna with a physical form factor designed for operation in vacuum (or air) will be detuned to a lower frequency when encased into a dielectric insulator or brought close to human tissues for instance (the well-known hand-effect detuning).
For instance, you are designing a #LoRaWAN sensor for the 868MHz band and have chosen a quarter wave antenna in the form of a wire normal to your ground plane. You have calculated the length of your wire to be l = lambda / 4 = c / ( 4.f ) = 8.6 cm and your prototype works fine.
Now fit your product with its antenna into an ABS plastic casing and guess what ? The radio range looks like it has degraded.
The reason is simple: the ABS plastic casing exhibits an epsilon_r of 3. If your antenna were entirely molded into a 30cm-cube of ABS its geometry should be reduced by a factor sqrt(3) in order to stay tuned to your 868MHz frequency:
Molded in ABS, a 8.6cm quarter-wave antenna is tuned to 868MHz / sqrt(3) = 500MHz, not at all optimized for 868MHz operation.
Because you are not molding your PCB and antenna into ABS but only enclosing them, it is highly probable that your antenna is detuned to a frequency closer to 800MHz, but still this detuning will yield losses of a few dB, harmful to your application.
This can be solved either by accurately tuning the antenna from inside the casing with a VNA, or more empirically by prototyping with slightly shorter antenna geometries and keeping the best performing design.
Bottom line: Hire an antenna expert at the very start of your project so that he can work with the mechanical engineer optimizing both the casing and the antenna performance.