Designing great LPWAN gateways comes with the need to consider many elements that make up a well functioning gateway – not least RF design.
You may have head that we are soon launching a Helium Hotspot Miner.
There are many LoRa gateway solutions from different manufacturers. Despite the fact they are all generally based on Semtech reference designs and share similar architecture, there are many elements that make each gateway design unique. The combination of the elements is what contributes to whether a gateway operates well or not.
Here are some Radio Frequency (RF) and other considerations for great design of Low Power WAN (LPWAN) gateways.
What’s a Gateway?
LoRa gateways create a bridge connection between low-bandwidth data from multiple LoRa nodes and a high-bandwidth network server or cloud on the Internet via WiFi, Ethernet, cellular connections and more.
Gateways are the central point of communication for multiple concurrent devices.
It is important to consider all these elements and form a broader view of the entire design (as a whole) when comparing the performance of different LoRa gateway solutions.
A gateway structure generally contains the following elements:
- Matching Circuit
- Front-end/ RF Switch and LNA
- Transceivers (I and Q modulator/demodulator) and peripherals
- Concentrator chip
The physical sensitivity of a gateway depends more on items 1-5 than on the concentrator itself. Of course, there is a sensitivity related to the coding scheme which is processed in the concentrator chip, but this does not have a relevant impact on the overall system sensitivity when compared to the rest of the elements mentioned above.
The schematic and Printed Circuit Board (PCB) design, even the selection of passive components (including the availability of parts when designing the gateway), have a considerably higher effect on the sensitivity when compared to the concentrator. Also, environmental conditions, installation, and coexistence with other radio systems can affect sensitivity and range.
The antenna is a crucial element and has a huge effect on the overall performance of any RF system.
Antennas are often selected based on a parameter that is commonly misunderstood: maximum gain. Antennas (passive antennas) don’t create radio energy, they merely concentrate the energy in some directions, which is known as radiation pattern.
Some key Antenna facts:
– High Gain Antennas: Antennas with high gain (>3dB) can focus the signal in some specific directions and as a consequence, they can achieve a longer range (distance). A longer-range in some directions, means a shorter-range in others. So, if the direction of the signal to receive/transmit is well known, then a high gain antenna (pointing to the desired direction) is recommended.
– Low Gain Antennas: Antennas with lower gain (0-3dB), in practical terms radiate fairly evenly in almost all directions (over a plane). Their radiation pattern follows a doughnut shape. These antennas are useful when the direction from where the signal is coming from, is unknown.
It is important to clarify that antennas are reciprocal devices, which means that they transmit and receive with identical radiation patterns.
A product’s regulatory certification such as FCC, CE, RCM and IC, is linked to the antenna a product has been certified with. It is therefore advisable to check the regulatory certification of a product before changing antennas, as some procedures may need to be followed.
In a typical LoRa gateway scenario, where many end-devices are intended to connect to the gateway from different directions, lower gain antennas may be more advantageous. This applies especially in a network system like Helium where “witnessing” is an important requirement.
A matching network circuit “matches” the impedance between a source and a load (like an RF amplifier and an antenna, or vice-versa). If the source and the load are not matched, then power transference is not optimized and some power will be lost due to reflections in the connection.
Manufacturers design their components (RF front-end, RF transceivers, filters, among others) for 50Ω input and output. However, it is important to consider that the actual impedance of the RF pins on a transceiver or RF module will depend on the final PCB layout (which won’t be necessarily identical to the one designed by the chip manufacturer or reference design).
It is therefore good practice to study and measure the input/output impedance of the RF modules when placed in one’s own PCB. It is also very important that the interconnection between those components, as well as the connection to the antenna, be done with a 50Ω impedance transmission line, to keep losses to a minimum due to mismatching.
Good PCB layout is important as well as routing, and impedance-matched interconnections in an RF PCB. The selection of the passive components placed on the PCB for the matching circuit (inductors & capacitors) are also relevant, as well as the ESR (Equivalent Series Resistance) or SRF (Self-Resonant Frequency) needs to be considered when selecting the appropriate component part numbers.
As the amount of radio devices using the LoRa Gateway increases, the radiated signal for one device or service becomes the interferer signal for another device or service. Therefore, it is important to filter frequencies that are outside of the intended operating frequency range. This is especially important for co-located radio modules, as in LoRa gateways, that usually use also WiFi or Cellular LTE.
A typical LoRa gateway uses an Low Noise Amplifies (LNA) in the reception path to amplify the weak incoming signal while driving down the noise figure of the reception chain. An LNA will show high amplification gain for the frequency range that is intended to operate but is not fully immune to other frequencies that could arrive at its input. When high signals (at other frequencies) arrive at the input of an LNA, the unwanted signal can saturate the LNA, making it enter into compression mode. This reduces the LNA effective gain of both, the high-powered unwanted signal and the small-signal gain at the desired frequency. This phenomenon can prevent the receiver from “receiving” the desired signal, or at least a considerable reduction of the sensitivity.
RF filters allows us to suppress these parasitic signals and consequently improve the selectivity of the reception by keeping the LNA into its linear operation.
It is important to remark that when there are no undesirable signals driving the LNA to compression, there may not be a performance advantage for having a filter before the LNA, so this is highly dependent on the environment where the LoRa module is installed. However, typical LoRa gateways co-locate LoRa with WiFi or LTE transceivers… which can affect the LNA performance and make it operate in compression.
There is always a trade-off, as filters also introduce insertion loss, so it is important to pay attention to the insertion loss at the desired operating frequency when selecting a filter as this has an important impact on the receiver’s overall sensitivity.
Front-End/ RF Switch and LNA
LoRa gateways usually contain a front-end with embedded RF switch (to share the same antenna for TX and RX) and also an LNA to amplify any weak incoming signals towards the RF receiver module and reduce the overall noise figure of the receiving chain, as mentioned above. The capabilities of the LNA, such as gain and Noise Figure (NF), have a high impact on the overall sensitivity of the receiver. Typically, RF front-ends used for LoRa systems have a gain of 16-18dB with a noise figure of less than 2.5dB.
Transceivers (I and Q modulator/demodulator) and peripherals
LoRa transceivers, like SX1257, SX1250, are highly integrated RF to digital I/Q modulator and demodulator chips. They incorporate a built-in LNA as the first component in its reception chain. The gain of this LNA also has a high impact in the overall sensitivity of the system. The noise figure of this LNA doesn’t impact considerably the noise of the system, as a previous LNA was employed as part of the external front-end (discussed above).
For the RF to IQ transceivers, the crystal oscillator is their main timing reference. It provides the reference source for the transmit and receive frequency synthesizers and as a clock for digital processing. In gateway implementations, it is recommended to use a Temperature Compensated Crystal Oscillator (TCXO) to achieve better frequency accuracy and avoid frequency drifts due to heating during some concentrator’s operation (if not properly temperature controlled). The oscillator or TCXO must also be selected to accomplish a good absolute accuracy in ppm for frequency variations, to ensure that LoRa nodes with higher frequency variation (during transmission), can still establish communication without sensitivity degradation with the gateway receiver.
As mentioned before, there is a sensitivity related to the coding scheme, which is processed in the concentrator chip, but those differences do not impact on the sensitivity as much as the rest of the elements exposed above.
Furthermore, the sensitivity of LoRa gateway reference designs/datasheets is generally defined for specific combinations of concentrator+transceivers+external LNA under nominal temperature, voltage conditions, and using other peripherals components that aim to optimal performance. So, the sensitivity of a gateway cannot be described using a specific concentrator, especially because the concentrator itself is the clement that impact less on overall sensitivity.
Some concentrators will generate large amounts of heat when operating, so when using them, a good heat dissipation design is required. Semtech recommends mounting a heat sink on top of those chips for safe operation in all conditions.
Using metal shielding for the LoRa gateway reduces the amount of noise being radiated by the module, as well as the noise that could affect the module (basically leaving the antenna as the point for the entrance of noisy signals). However, it is important to consider heat dissipation design when placing concentrators and transceivers/front-end inside the same metal shielding structure, as this may cause frequency drifting, especially when not using TCXO for the transceivers.
Whilst not necessarily the most critical element impacting gateway performance, it is nevertheless an important part of a successfully designing a gateway. Regular firmware updates are important to keeping your device healthy and as such and a Premium Gateway will have Firmware-Over-The-Air (FOTA) support. All Operating System and Chip Set vendors will regularly issue updates including patches and therefore the gateway manufacturer will make regular updates of the firmware available. The maturing of the firmware code and this ongoing process will make gateway performance optimal and much more predictable.
In summary, there are many elements to consider when developing a LoRa Gateway and comparing the sensitivity/performance of two designs, even when they are based on similar reference designs. The designs must be analysed to provide a proper comparison.
We hope you found this informative and feel free to send us your comments and of course questions
Pycom’s RF Team