Low power MOX
Gas sensor technology is evolving towards smaller and more power-efficient sensors with digital interfaces. Thanks to micro-machined technology (MEMS), metal oxide (MOX) gas sensors can be fabricated nowadays with a very compact form factor and power consumption of mere tens of mW. High power efficiency is a key requirement for gas sensors to be integrated into mobile or battery-operated devices, such as portable gas detectors, smartphones or wearables. Gas-sensitive mobile devices have a direct application to the food industry, industrial safety and biomedicine, among others. However, even the 15 mW of the Figaro TGS8100 (one of the most power-efficient MOX sensors in the market) might be considered too high for certain battery-operated applications. For instance, if we integrate this sensor into a Samsung Galaxy S7 smartphone (which consumes 462 mW under non-intensive use) the battery lifetime would be reduced in 45 min. Considering that several MOX sensors are needed in a real application, e.g., to increase the selectivity of the system, the situation worsens. For example, the multi-sensor unit SGP by Sensirion AG (power consumption of 48 mW) accounts for more than 10% of the smartphone’s power consumption or, equivalently, a reduction in battery lifetime of 2 h and 15 min. It is known that smartphone manufacturers demand sensors with power consumption lower than 1 mW.
The main source of power consumption in MOX sensors is the heater resistor that heats up the sensing material where the chemical interactions between the target gas and the metal oxide take place. Many solutions to the problem of power consumption in MOX sensors have been proposed using the technique of duty-cycling. Duty-cycling saves energy by periodically switching on and off the heater resistor. The amount of time the heater is powered with respect to the duration of the cycle is known as the duty cycle (DC). The DC is directly proportional to the average power consumption. However, a less studied problem is how the DC affects the accuracy of the sensor measurements. It is expected that low DCs (i.e., short ON periods and long OFF periods) degrade the accuracy of the sensor with respect to the standard continous power mode but this has not been characterized properly in the literature.
In the following, we describe our efforts towards the reduction of power consumption in MOX sensors by exploring different duty-cycling strategies, and the characterization of the sensor accuracy under each strategy.
In this paper, we propose a 10% DC operation of 10-min periods in order to reduce the energy consumption of FIS SB-500-12 sensors by 90% as compared to the continuous operating mode. We evaluate the method in a scenario of on-demand measurements of carbon monoxide (CO) at low concentrations with background interferences of humidity and temperature. This scenario is motivated by a myriad of applications requiring the detection of low concentrations of CO under variable humidity and temperature conditions, such as breath analysis or the food logistics chain. We compare the performance of this strategy against the most power efficient method (on-demand operation) and the most stable one (continuous operation) during two weeks of measurements.
The results illustrate that the proposed DC operation represents a trade-off between on-demand and continous operation in terms of prediction error. Compared to the on-demand mode, the DC approach reduces the prediction error by a factor 2.5 (2.2 vs 0.9 ppm) but only increases the power consumption by 10 %. Compared to continous mode, it doubles the prediction error (0.45 vs 0.9 ppm) but reduces the power consumption by 90 %. The average error of the DC approach thjrought the 2 weeks remains below the 1 ppm threshold required by many CO sensing applications.
In this other work, we applied PTO operation to ultra-low power (ULP) consumption MOX sensors to fulfill the requirements of a Radio Frequency IDentification (RFID) flexible tag application. The developed prototype is the first ISO 15693 compliant semi-active tag prototype, including low power control electronics, RFID antenna, 4 ULP sensors, memory and a thin film battery. The ULP sensors were provided by the Institute for Microelectronics and Microsystems, Italy.
Under continous heater excitation, each ULP sensor consumes 14.5 mW of power. As a result, a 25 mAh battery can only supply the required power to the set of four sensors for 40 minutes. To minimize the power consumption and increase the battery life, we explore ten aggressive PTO modes leading to power savings of 91 - 99.9 % with respect to continuous heater stimulation. These PTO modes represent power consumptions in the range between 1.4 mW (least aggresive mode) and 6.8 uW (most agressive mode).
We compare the different strategies based on four figures of merit: power consumption, sensitivity, limit of detection (LOD) and stabilization time. In the proposed test scenario, we expose the four ULP sensors to ammonia, ethylene, and acetaldehyde (relevant gases in the food logistics industry).
Results show that the stabilization time can increase from 10 minutes (most conservative PTO mode) to 10-20 hours for power consumptions smaller than 200 microwatts (most aggresive PTO modes).
Regarding the sensor sensitivity, PTO schemes with higher power consumption resulted in higher sensor sensitivity. Similar trends in sensivitity were observed for the three studied gases. In the case of ethylene, the most aggresive PTO mode dramatically reduced the sensitivity.
As a result of the change in sensivitity with power consumption, the LOD of the sensors was also affected by the PTO scheme. In the case of acetaldehyde, the LOD increased by a factor of 5 when going from the most aggresive to the least aggresive PTO modes. On the other hand, ammonia shows a robust LOD that remains constant for all the tested PTO schemes. This is probably due to the higher sensitivity of the sensors to ammonia and robustness to noise at lower concentration levels.