One of the most technical challenges in the Internet of Things (IoT) is sensor nodes that can be placed anywhere. These sensors measure parameters such as temperature and humidity (connected home), mechanical stress of motorway bridges (live maintenance monitoring) or the consumption of gas or water (smart flow metering). This data is gathered and processed by servers and needs an extensive coverage area to make a robust network with reliable data. The enabling technology is the wireless transmission of the sensor data to a central host system.
To realize a broad network like this, another critical aspect must be considered, which is that the whole sensor node must feature a very long operational lifetime. The higher the lifetime, the lower its maintenance cost. With the power optimization of microcontrollers and battery-types like LiSOCl2 primary cells, these processors can be powered for 10 years or more.
Until now, radio frequency (RF) transmission of the sensor data for longer distances was not widely implemented. This wireless feature adds another level of complexity to the system’s power considerations. While a wireless sensor node needs to consume the lowest possible average power, it also must be able to deliver high-peak currents for the occasional data transmissions.
From a power perspective, this means a combination of the lowest quiescent currents in the sensor system and an efficient high-power capability for the power amplifier. This requirement is a new challenge for selecting devices, as well as the whole power architecture itself.
Low quiescent current and long life
To ensure that IoT-sensors become a reality, operating the sensor must be cost-effective. Once the sensor is installed and started, it needs to operate as long as possible to minimize the period between maintenance visits and save cost.
This means that, on one hand, durable materials and components have to be chosen. On the other hand, the internal circuits also need to feature lowest current consumption to gain a longer runtime with a given energy from the battery.
Currently, these applications use specific primary batteries. Chemistry types like the LiSOCl2 feature a very high energy density of more than 1 Wh/cm³ and are broadly available on the market. These primary cells bring a very low self-discharge, another important property to consider. This makes them a designer’s first choice for long-life applications.
To benefit from these parameters, the battery current must be limited to less than 5 mA. Currents beyond this increase the self-discharge rate, which reduces the cell’s lifetime. As well, higher currents force the terminal voltage to decrease due to the internal impedance. In addition to the battery, the power-consuming components and power architecture must be optimized to minimize leakage currents.
Ultra-low-power microcontroller system-on-chip (SoC) devices feature several low-power modes to decrease current consumption. An ultra-low-power SoC extends the application lifetime due to its implemented standby mode where the device consumes around 2 µA when connected directly to the battery. Figure 1 shows the supply current of this device in a low-power mode (LPM3). The current consumption depends on the supply voltage (green trace).
Current consumption is further reduced when SoCs are used in combination with an ultra-low-power buck converter to decrease the supply voltage. These are step-down converters with a quiescent current of several hundredths of nano ampere. The blue trace shows current drawn by the application after stepping down the supply voltage to 2.1 V. The higher the battery voltage, the more power is saved due to the efficient step-down conversion. At the typical LiSOCl2 battery terminal voltage of 3.6 V, overall current consumption goes down by 30 percent compared to the direct battery connection.
Peak power for wireless transmission
Besides the low IQ aspect, the sensor must communicate the gathered and processed data to a base station. For example, this can be a local data concentrator, which is common for smart gas sensors in an apartment building. Besides the wireless metering bus (wireless M-Bus), this also can be the available global system for mobile communications (GSM) infrastructure used for field sensor nodes on motorway bridges.
A typical mode of operation is gathering and processing data throughout the day, then transmitting the collected data up to a few times a day. From a power perspective, this means that a low average current consumption in the range of microamperes is mostly needed, with a rare need of higher currents for several milliseconds.
The amount of energy needed for data transmission depends on the range and, therefore, the radio frequency protocol. Widely used standards are wireless M-Bus and GSM.
A comparison of three common standards is shown in Table 1. Each standard has a typical radio amplifier power condition and the required current for transmission duration.
In some cases, currents up to 2.5A are required by the radio amplifier. This amount of current cannot be delivered by the battery types described. Even currents of more than 5 mA should be avoided in order to not reduce the lifetime of a LiSOCl2 bobbin-type battery.
Energy buffering concept
To enable pulsed-load operation as described, new power management concepts need to be considered. Since the battery cannot deliver the necessary current, the required energy needs to be stored when the radio is inactive so it can be used when the radio is active. To achieve this, a new power concept needs to be designed to buffer energy and decouple load peaks from the battery. An excellent medium for buffering energy are storage capacitors because of their high-energy density and large capacitance.
When using a switch-mode power supply (SMPS), a capacitor can be efficiently charged with different voltage than the battery. This can be done in a current-limited operation, which then defines the load current for the battery.
Once energy is stored in a capacitor, voltage is converted to the desired value, for example 1.9 V for the microcontroller SoC or 3.7 V for the radio power amplifier. This conversion takes energy from the buffer capacitor and decouples the load from the battery (Figure 2).
When using a SMPS buffering power architecture, two basic concepts to store energy apply:
- Boost – storage – buck
- Buck – storage – boost
Concept one steps-up the battery voltage to a higher voltage and charges a capacitor. Then the voltage is stepped-down to the desired values for the SoC or amplifier.
This concept uses smaller capacitor values because the stored energy is proportional to the square of the capacitor voltage. The higher the voltage, the more energy is stored in the same capacitor. Once the energy is stored in the capacitor, the voltage is stepped down to the desired value. The energy required for a transmission is extracted from the capacitor, and thereby decoupled from the battery.
The second architecture uses a buck converter that is directly connected to the battery. The voltage is stepped down to charge a storage capacitor. Here the storage capacitance value must be higher because the voltage is lower. However, this enables the usage of electric double-layer capacitors (EDLC), which are broadly available with a high capacity of several Farads. After the storage capacitor, the voltage is stepped up to the desired value (Figure 3).
Besides the higher available capacitance, this concept features three advantages:
- Due to the lower storage capacitor voltage, less safety considerations need to be taken into consideration, compared to a capacitor charged with a higher voltage.
- The already stepped-down voltage can be used to power the SoC microcontroller directly. This reduces overall current consumption with just one SMPS being active at all times.
- The lower voltage enables the usage of EDLC-type capacitors. These capacitors are available with high-capacitance values.
When using a buck-storage-boost concept in a wireless sensor (Figure 3), the lowest EDLC voltage is defined by the minimum required SoC supply voltage. The energy is then buffered by charging the capacitor to its maximum voltage of 2.7 V just prior to radio transmission. This keeps the average supply voltage close to the minimum of about 1.9 V. During radio transmission, the EDLC is discharged to the defined minimum voltage.
Conclusion
The requirement of lowest-quiescent-current devices in combination with high power is a challenge for power architectures. Using the energy-buffering concept of “buck-storage-boost” solves decoupling of the load peaks by storing the required energy in an EDLC. It also achieves lower overall power consumption because of the lower supply voltage of the microcontroller. There are less safety concerns, as the storage capacitor uses a lower voltage. This concept can combine energy buffering in a storage capacitor with a reduced overall current consumption to allow longer application runtime.
