Optimizing LoRa radio performance for embedded devices


Long Range Communications on the Horizon (Source: Pxhere.com/CC0)

Introduction

Whether developing a wearable device or industrial battery powered equipment, maximizing range and robustness while minimizing power consumption is critical.  Optimizing RF performance increases flexibility and enables more appealing trade-offs regarding size, battery-life and RF performance.  

After optimizing RF performance, the product development team can consider lowering transmit power to increase battery life or decreasing battery capacity to reduce the product size or perhaps operating exclusively on harvested power and eliminate batteries completely.

Link Budget & Path Loss

So, what factors determine RF range and performance?  Let’s start by examining the link budget.  The link budget is the difference between the strength of the transmitted signal and the minimum required signal at the receiver and equals the total loss from all sources at the maximum range. The simplest equation for link budget is (Figure 1):  


Figure 1: LinkBudget Basic Elements (Source: Device Solutions Inc)

For a typical LoRa radio implementation:  

This configuration provides a link budget of 150dB.

Before range can be estimated using path loss calculations, there are other factors to consider:

  • Transmitter antenna gain in dB, if positive, increases the link budget
  • Receiver antenna gain in dB, if positive, increases the link budget
  • Loss between the transmitter output and the antenna decreases the link budget
  • Loss between the receiver input and the antenna decreases the link budget

Including all these factors provides the link budget available for path loss (Figure 2):


Figure 2: LinkBudget Intermediate Elements (Source: Device Solutions Inc)

Antenna gain is typically expressed in dB relative to an isotropic antenna (dBi), an antenna that radiates equally in all directions.  Typically, an antenna data sheet specifies the “peak gain”, indicating how well the antenna radiates in the optimum direction and the “average gain”, representing the antenna’s effective radiation averaged over all directions.  Generally average gain should be utilized unless the orientation of the devices can be controlled to realize the “peak gain”.  Average antenna gain is equivalent to efficiency, thus, an antenna with an average gain of -3dB is 50% efficient, which can be a more intuitive way to visualize the impact of the antenna performance.  Antenna gain (transmitter or receiver) of -4dB is typical for a compact LoRa device.  If implemented carefully and compactly, receiver and transmitter losses should be approximately 1dB each.  However, the loss can be much higher if the antenna is not well matched to the transmitter and receiver circuits.

Power can only be transferred from the transmitter to the antenna efficiently if the transmitter output impedance is closely matched to the input impedance “load” seen by the transmitter.  That load includes the PCB trace, antenna and any components in the RF path connected to the transmitter’s output pin.  Typically, there is a matching circuit used to transform the antenna impedance (at the desired frequency) to the transmission line characteristic impedance on the PCB and another matching circuit to transform the PCB transmission line impedance (typically 50Ω) to an optimum impedance for the transmitter.  If the antenna and amplifier are poorly matched, then the transmit signal will not be efficiently transferred to the antenna, reducing range. When poorly matched, the transmitter will consume more current, decreasing battery life, and may generate increased harmonics.  Additional harmonic radiation exacerbates the challenge of regulatory approval and may require additional filtering to mitigate – which increases PCB area, increases loss and increases costs.

Combining the typical numbers with the LoRa example mentioned above yields (Figure 3):


Figure 3: LinkBudget Detailed Elements (Source: Device Solutions Inc)

At least 6dB should be subtracted from the link budget to provide margin for real world conditions and operational robustness.  Therefore, in this example, propagation loss at maximum range is approximately 134 dB.

The development team’s decisions directly impact many of components of the link budget and the team can make trade-offs to increase range or to reduce power consumption. Options include increasing transmitter output power or antenna gain, improving receiver sensitivity or minimizing loss. These choices may increase the size and cost of the radio implementation, the battery or the antenna, but it’s important to deliberately consider the performance impact of each decision. Optimizing performance could make the difference between achieving the desired range within the regulatory power limits or being forced to compromise on range to stay within allowed limits.

These trade-offs can be especially difficult when developing wearables, which are extremely size and cost constrained, demand maximum battery life, minimum size and are further constrained by regulatory (FCC, RED) requirements to minimize the RF energy absorbed by the user, known as “Specific Absorption Rate” or SAR.  Cellular devices are further complicated by carrier and industry requirements which require highly optimized antenna performance and high transmit power (compared to Bluetooth or WiFi) while still meeting the SAR limits.  Meeting these requirements, within in a commercially viable package is extremely challenging.

Receiver Sensitivity

Less obvious is the influence that the development team has on receiver sensitivity.  Receiver sensitivity is determined by the radio modulation, the bit rate and the details of the receiver implementation.  As always, a larger, higher power and more expensive receiver will typically perform better.  Lowering the bit rate is another way to improve the receiver sensitivity.  

Table 1 below illustrates how modulation and bit rate impacts receiver performance.  Remember, smaller/more negative sensitivity is better:


Table 1: FSK and LoRa Bitrates vs. Sensitivity

The LoRa Spreading Factor (SF) indicates the duration of the physical layer CHIRPs used to transmit the data.  A larger spreading factor indicates a longer CHIRP and lower bit rate.  A detailed description of LoRa can be found in this application note (http://www.semtech.com/images/datasheet/an1200.22.pdf) from Semtech or this web page provides a more intuitive explanation.

The development team can optimize the system design to minimize the required bit rate and therefore improve the sensitivity and range, by ensuring the minimum amount of data is transmitted.  Sensitivity can also be improved by additional investment in receiver power consumption, size or cost. For example, adding additional filtering or a low noise amplifier.  Decreasing the bit rate will increase the transmit time and may decrease battery life.  Minimizing required throughput also minimizes required transmit time (at any bit rate) and allows the team to maximize the sensitivity while balancing range, transmit time, and battery life.  Higher bit-rates yield shorter transmit time but have shorter range, for a fixed transmit power, providing another trade-off the team can use to balance RF performance against other requirements.  If the RF implementation is optimized by maximizing the transmitter efficiency, receiver sensitivity, and antenna gains, excess link budget can also be “spent” on less expensive components or a lower performance antenna that enables a more desirable product appearance or lower transmitter power to improve battery life.

The discussion above assumes that the radio implementation meets the manufacturer’s specifications.  To achieve this level of performance, it is critical to adhere to the manufacturer’s recommendations and to minimize sources of interference which would degrade the performance.   Again, the product development team must trade-off performance against size and costs. Consider common noise sources and mitigation techniques:

  • Sources
    • Processors, especially external memory busses
    • Switching power supplies
    • Isolated RS-485/232 drivers
    • Displays & video drivers
    • Class D audio amplifiers
    • Motor drivers
  • Mitigation
    • Shield cans and shielded cables
    • Additional filters and amplifiers
    • Additional PCB layers
    • Line termination and slew rate control

The majority of these mitigations increase product cost and size, but may be an appropriate choice if they enable improved range or reduce other costs or size – such as a smaller or less powerful battery. Aggressive mitigation of potential issues should also be considered to minimize risk of failing regulatory testing and minimize the time to market.  Addressing and preventing noise will maximize realized sensitivity enabling maximum range and minimum transmit power.

Range and Propagation

Now that we have discussed how to optimize system performance, let’s discuss propagation and range estimates.  In an ideal word, often referred to as “free space”,  the signal propagates out from the antenna in all directions, without reflection, atmospheric refraction or absorption.  Loss in this case is given by (Figure 4):

Where f is frequency in megahertz and d is distance in kilometers.


Figure 4: Signal Loss over Distance and Frequency (Source: Device Solutions Inc)

Note that frequency is a key component of this equation and that lowering the frequency decreases the loss. Decreasing the frequency from 2.4Ghz (Bluetooth, WiFi) to 900Mhz decreases path loss by 9dB and should more than double the range, if everything else remains constant. Understanding this reveals yet another trade-off – lowering the frequency of the signal can increase range.  However, for a given volume, the antenna efficiency decreases as the frequency decreases, potentially offsetting some of the benefit of the lower frequency.

Unfortunately, range in the real world is impacted by many other factors, such as reflection and absorption by various obstacles. There are a variety of real world propagation models, most based on empirical data-sets.  The Okumura–Hata model is a good choice and provides options for various environments (urban, suburban, rural) and a variety of antenna heights. In a rural or open environments, the path loss formula is (Figure 5):

Where:

hB  = Height of base station antenna. Unit: meter (m)

hM  = Height of mobile station antenna. Unit: meter (m)

f  = Frequency of transmission. Unit: Megahertz (MHz)

CH  = Antenna height correction factor

d  = Distance between the base and mobile stations. Unit: kilometer (km).


Figure 5: Signal Loss with Multiple Factors (Source: Device Solutions Inc)

Building on the previous LoRa example and using the IEEE worksheet available here, the HATA model predicts a path loss of 134dB at 3km with antennas 2m off the ground.

Power Consumption Trade-offs

In addition to the first order trade-offs mentioned above (transmitter power, bit rate vs. transmit time & sensitivity, noise reduction, cost, size) there are many other considerations to minimize power consumption. Minimizing radio on-time, in either receive or transmit mode is key to maximizing battery life. Although it is intuitive that transmitting uses a significant amount of energy, power consumption of many modern receivers is on par with transmitter power due to the significant signal processing required.  Careful design of over the air protocols and synchronization algorithms is necessary to ensure quick and reliable synchronization, frequency alignment, and minimum on-time. Use of higher precision crystals can minimize the risk of temporal or frequency misalignment and ensure the radio “locks” more quickly, minimizes noise and minimizes retransmissions, especially when the temperature and aging is considered. Careful attention to initial accuracy, accuracy over the required temperature range, and frequency drift due to aging must be paid to ensure your design will work over the long term.

Optimizing the over the air protocol is only one example.  All events which trigger the device to deviate from the minimum power consumption state should be carefully considered, including interaction with all inputs, outputs, and any “indicators” or UI elements.  Whenever possible, multiple events should be handled each wake period to minimize the frequency of the wake periods.  Similarly, a power consumption trade off must be made regarding higher clock speed, which results in higher power consumption but for shorter duration versus lower clock speeds resulting in lower power consumption for longer durations.

All aspects of the power supply design must also be considered.  State of the art switching power supplies have improved tremendously but can still be inefficient when the load is only a few micro-amps, such as when the device is in a sleep state between transmissions.  However, very low quiescent linear regulators often have surprisingly poor transient response characteristics, so it is imperative to carefully consider these components.

Often, in low power states, many sub-circuits are switched off, however, the state of each IO line and connection between sub-circuits must be examined to ensure there are no active signals connected to powered down components or there will be unintended leakage current, perhaps several milliamps, and unexpected behavior can occur due to leakage currents partially powering some components.


Figure 6: Select Considerations and Trade-off’s in Optimizing Radio Design (Source: Device Solutions Inc)

In conclusion, it should be clear that to maximize range and battery life, almost all aspects of the device must be considered.  The size of the device bounds the antenna efficiency, battery volume and PCB area for an optimum RF implementation.  The care with which the RF circuits are designed bounds the range and if done poorly, will degrade the battery life. Similarly, care invested in design of the operational states to maximize sleep time and minimize on air time can increase range and improve battery life. Real-life product development requires constant negotiation to achieve technical optimization and commercially viable size, cost and performance (Figure 6).


Chris Lamb , co-founder and Chief Technology Officer of Device Solutions, has over 2 decades of experience in commercial product development and  mobile devices and applications.  Chris drives end to end product and technical strategy for Device Solutions and their customers, finding solutions which ensure technical, regulatory and commercial success.  Through broad technical involvement, including work with the NSF ASSIST center at North Carolina State University, he is able to identify unique solutions and form key partnerships.  He is recognized worldwide as an authority on advanced mobile devices, certifications, protocols, and messaging. Chris has managed the product acceptance testing of mobile phones for Ericsson and Sony Ericsson, and maintains strong relationships with vendor managers at major North American operators.  He has worked directly with global third party developers of mobile device applications and content to help drive products to market. Chris is a graduate of the Massachusetts Institute of Technology, and holds an MSEE from North Carolina State University. He serves his community as a little league baseball coach and board member, volunteering with schools and as assistant scoutmaster for Troop 461.

Rod Williams is Director of Engineering at Device Solutions. On the management team at Device Solutions, he has over 3 decades of engineering experience in the wireless product development industry with companies ranging from IBM, to Ericsson, to Garmin and HTC.  Rod has provided technical leadership to engineering efforts at Device Solutions since 2012 and has been pivotal in design reviews, testing, certification and approval processes for almost all of the embedded products Device Solutions has developed for clients and their custom applications.  As Systems Engineering Manager at HTC in the Corporate Strategy and Innovation office, Rod drove product concepts to life for smartwatch wearables and high performance tablets.  Rod served on the GSMA committee that standardized the nanoSIM.  Outside of work Rod enjoys TIG welding for art and his passion for old pickup trucks.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.