Episode 3 & 4 of my Youtube series about LPWAN is about antennas for LPWAN and impact on object design. The episode 4 go deeper on the way to improve radio performance by tuning and adapting antennas with matching circuit.
As usual, this series of post are summarizing in English the content of my French video on youtube.
Antennas are a critical element of a LPWan design, it can make radio working correctly and preserve energy consumption of the batteries. The first important things is to have an antenna matching the radio frequency wave length. For 868MHz the right size in 34,5cm compared to 2,4G (wifi, BLE…) where we have 12,5cm. Even if the size if this one for the wave length, we generally use 1/2 wave length or 1/4 wave length to get good results. It is really import to have the antenna exactly matching the 1/4 wave length 8,6cm.
Antennas can have many different form factors (see video) in any case we will have somewhere a net with the right size, it can be straight or rolled, eventually draw on a PCB net or included in a ceramic component. Basically we will have a different radio result depends on the antennas. The difference can be related to the antenna gain (or loss) and the directivity of the antenna.
The ground plan is part of the object antennas, it’s size will have a direct impact on performance. For this reason and the wavelength in 868MHz it is really complicated to make small but efficient radio emitter. Not impossible but it requires engineering.
Due to all of this, antenna generally have a gain (gain can be negative). As a consequence the energy provided by the antenna over the air is not equal to the energy provided by the emitter to the antenna. We usually have loss. We usually use dB as a scale to express this gain/loss. Basically, dB is a logarithmic unit to express difference. A difference of -3dB means half of the power has been killed by the antenna circuit. The dB scale double every 3dB. It means if you emit +25mW on a -3dB antenna circuit, you will effectively push +12,5dB over the air. Regulation is based on what you push over the air, not what your emitter send. So basically you can emit with more power to compensate the antenna circuit loss. The problem related to this solution is the power consumption, twice mode power means an autonomy divided by two on your batteries.
The right way to solve this issue is to make an antenna matching circuit to solve this issue.
The antenna circuit loss is related to you PCB design but not only, the box where the circuit is installed also have a large impact. For this reason the network operators wants to have you device certified. This is a way to ensure the object radio is correct enough to work on most of the area covered by the network.
Sigfox classify the object into 4 categories :
When U0 it means your antenna will deliver more than half of the nominal 25mW power.
-7dB means your object only deliver less than 6mW over the air…
Basically having an object U0 will give you the best network coverage but U1, U2, U3 may also concern working object. 0dB does not mean 0mW but about 1mW. Eventually, depends on the antenna position and distance to your object it can communicate correctly. For this reason when I discuss with startup who have limited budget on antenna design and tuning I usually recommend to target U0 for U0 but to try to get the best radio efficiency for the budget they have.
Being a U1 with 12-10dBm is already a nice result and will work in many case, mainly when you firstly target an urban market where your object will be listen by many antennas. A later tuning once the market will be proven will be more easy with more money.
To have a clear view of what does it looks when you check a not matched antennas with a VNA :
We see at 868 MHz a power loss of -2,5dB so it means we are about to loose 50% of the energy. This is visible on the following graph in red.
We see the same curve in the opposite direction. This graph is energy kept when the previous one was energy loss.
The associated software is able to propose a matching circuit. It is a set of resistor, inductance and capacitor you can add between your antenna and your emitter to make the matching. For this it is critical in a radio design to add component footprint for adaptation circuit.
Once adapted the theoretical performance is totally different as seen on this second graph:
The expected radio performance is now 92% so << -1dB so it is really good.
We can verify this, once the matching circuit added with the VNA and get the new response curve.
The frequency response is a little bit different from the theoretical one but are at -13dB at 868Mhz – about 6% loss.
If in the laboratory we have really good result we should expect a great difference and we should be able to measure this difference on the network antenna RSSI indicator. This is where the reality is fighting against the theory. We can take a look on the two following network report : the first one before matching, the second one after matching :
As you can see OAE1 station rssi is better but 3D14, ODC0… are worst after matching. In fact they are not really worst because of matching, but I assume that the weather, the air temperature, moisture have changed and I assume these change are a factor 1 in the RSSI at this level.
The real impact of the matching is really difficult to measure when we are talking about 2-3dB of difference and the real impact not really visible.
In the case of an antenna with an initial poor design like a PCB antenna the result can be totally different : with the same power emitted at the same location, here are the RSSI results :
You can notice we had only 4 base stations able to see the object communication compared to the 8 of the previous object and on the 4 one 2 had only 1 repetition of the message (the 2 other has been loss). When the first object had a RSSI -116dB on 0DC0 the second one have -123dB it is a difference of 7dB, means 1/4 of the power only. On 3D14 it is -19dB…
Here is why antenna design and matching matters.
Thank you to exotic-systems for helping me on antenna matching. Don’t hesitate to contact them if you have objects to tune they are experts.