Concept and Development of Air Quality Sensor for Citizen Science ^†

Abstract

This paper presents the concept and development of an autonomous DIY air quality sensor for citizen science. Large civil monitoring projects often rely on air quality calculations based on PM2.5 and PM10 dust readings in combination with some gases and do not cover the full list of air quality indicators. The authors have analyzed existing air quality calculation methodologies and attempted to conceptualize a universal AQI monitoring device for use in citizen science and by volunteers. This device is based on the available ESP32 DevKit v1 platform to which compatible sensors have been selected to monitor AQI indicators such as PM2.5 and PM10 dust particles, ozone, carbon monoxide, nitrogen dioxide, sulfur dioxide, and ammonia. The SD card module was chosen for data storage, the NB-IoT module for data transmission, and a battery pack for autonomy. The housing, sensor design components, and fasteners were also selected. All components are available on the international market. Based on the selected element base, an electrical connection diagram was designed, the device’s design, presented in the form of 3D models, was developed, and the assembly process was described. The cost of the device was also evaluated and compared to the price level of existing DIY devices used in citizen science.

1. Introduction

People, especially those living in large cities and industrial centers, are concerned about what they breathe and whether their health is at risk from chemicals, factory emissions, dust, and gases dispersed in the atmosphere. For many years, the World Health Organization, environmental protection agencies, volunteers, and scientists have been looking for ways to reduce the harmful effects of air pollution on human health.

The urgency of the problem, as well as its comprehensibility among the population and ordinary people, has given rise to a large number of citizen science and volunteer projects for air quality monitoring. The best known among them are “Sensor.Community” (ex Luftdaten) [1] and WAQI [2]. They have also spawned many projects based on them and organized in similar ways, such as Air Matters [3], “Moscow.Breathe” [4], “Nebo” [5], and so on. All these projects use data from low-cost DIY devices of users, data from which is collected into a common network. Some projects allow the connection of any user device to the network, some, such as “Sensor.Community”, offer devices of their own design with ready firmware.

If we take the monitoring station of the “Sensor.Community” project as the most popular and representative example, it represents a certain transition from a simple DIY device to a DIY product for a number of reasons: it consists of a small set of cheap and globally available components; it does not require special technical skills from the user during assembly; and it does not require programming skills from the user to flash the controller. However, this device has a number of disadvantages both in the technical part and in the part of AQI assessment: the device is dependent on the power supply network and cannot operate autonomously; the device requires constant availability of Wi-Fi network, as it sends these data only via the internet to the project website; and it measures only one of the AQI components—PM2.5 and PM10 particles.

If we consider a WAQI project that collects information on different gases besides PM2.5 and PM10 particles, we can see that the sets of AQI components measured at different points do not match each other depending on the sensors connected.

That being said, there are many different cases of building and using DIY devices to construct air quality monitoring stations using available and cheap components.

To summarize the above, a logical step in the development of citizen science AQI monitoring can be the creation of a universal device for air quality measurement, similar to the station of the “Sensor.Community” project, which meets the following criteria:

• The device must be able to measure a universal set of indicators found in most AQI calculation methodologies;
• The device must be able to operate autonomously without a mains power supply;
• The device must have several possible ways to transfer and/or store data: online over Wi-Fi; online over another puncture if Wi-Fi is not available; offline to a memory card;
• The device components must be relatively cheap and available worldwide;
• The device must still be easy to assemble without requiring specialized technical skills.

In this paper, the authors have endeavored to present the concept of a designed device that meets these requirements.

2. Materials and Methods

2.1. Measured AQI Indicators

The measurement of the following components of the air quality index has been selected for the device:

• Dust particles PM2.5 and PM10. Included in almost all existing AQI methodologies.
• Carbon monoxide (CO). Included in AQI methodologies used in Australia, China, and Mexico.
• Ozone (O₃). Included in almost all existing AQI methodologies.
• Ammonia (NH₃). Included in AQI methodologies used in India.
• Sulfur dioxide (SO₂). Included in AQI methodologies used in Australia, China, Mexico, India, the UK, and the USA.

Also, measurements of the following indicators, not directly included in the air quality index, were selected for the sensor:

• Carbon dioxide (CO₂). Not included in any AQI methodologies, but is often evaluated along with it [6,7,8,9,10];
• Air temperature, relative humidity, and atmospheric pressure.

Nitrogen dioxide, which is a component used in almost all AQI methodologies, and lead included in AQIs in India have not found application in the concept. The existing sensors to measure these components are quite expensive and do not fit into the concept of affordable devices for citizen science.

2.2. Sensor’s Components

According to the set requirements and chosen AQI indicators the full list of the sensor’s components can be separated into four categories:

• Sensing components of AQI indicators measurement;
• Main and additional electronic components;
• Enclosure and fastening components;
• Energy store components.

The list of the main sensing components according to the chosen AQI indicators is as follows:

• Nova PM sensor SDS011 (Shandong Nova Technology Co.,Ltd., Shandong, China) is used to measure PM2.5 and PM10. This device is used by the monitoring station of the “Sensor.Community” project and considered as one of the most efficient optical PM sensors in the low-cost and DIY sector of the market [11,12,13,14];
• MH-Z19B sensor (Winsen, Zhengzhou, China) is used to measure the level of CO₂. This device is proven to be valid in many indoors and several outdoors measurements [15,16,17].
• MQ7 sensor (Winsen, Zhengzhou, China) is used to measure the level of CO. The MQ is the widely spread series of compact electrochemical gas sensors which is used in many IoT citizen science projects [18,19,20]. It is well calibrated for different use cases [21,22,23,24] and can be replaced by “Prana Air” sensors in those countries where MQ sensors are not available.
• MQ131 sensor (Winsen, Zhengzhou, China)is used to measure the level of O₃ [25,26]. It can be replaced by “Prana Air O₃” in those countries where MQ131 is unavailable.
• MQ137 (Winsen, Zhengzhou, China)is used to measure the level of NH₃ [27,28,29]. It can be replaced by “GS+4NH3-100 DD Scientific” in those countries where MQ137 is unavailable.
• Sensor SPEC SO₂ 110-610 (SPEC Sensors, Fremont, California, the USA) is used to measure the level of SO₂. SPEC sensors are used as a more accurate alternative to MQ sulfur dioxide sensors according to the recent investigations and use-cases [30,31].
• BME280 sensor (Bosch Sensortec Gmb, Reutlingen, Germany) is used to perform additional measurements of atmospheric temperature, humidity, and pressure. It is also used by the “Sensor.Community” project, as well as in a variety of DIY and citizen science devices [32,33,34].

For the majority of gases, the chosen MQ sensors, which are MOS sensors, offer a range of benefits over other types, particularly when it comes to reliability in DIY projects. They have a significantly faster response times compared to electrochemical and catalytic sensors [35,36,37], a longer lifespan [38,39], and they also align with the principles of low-cost devices. With an average price of $1–2, they are not much less expensive than catalytic sensors (which typically cost $2–5), and they are significantly more affordable than electrochemical sensors (the cheapest ones can cost around $60–100).

Furthermore, the approach to bringing these devices to market is crucial. MQ sensors are readily available as pre-assembled modules for integration with a microcontroller, whereas electrochemical and catalytic sensors are sold as standalone products, necessitating the acquisition of additional electronic components or their independent development.

Electronic components have to meet the requirements of connection of the selected sensors as well as receiving, transmitting, and storing the data. The list of used electronic components are as follows:

• ESP32 DevKit v1 (Espressif Systems, Shanghai, China) is used as the main board of the device. The controllers ESP by Espressif is widely used in IoT, citizen science, and DIY projects as an alternative to Arduino board. ESP series controllers possess more processor power and larger choice of board types and configurations [40,41]. The PM monitoring station of “Sensor.Community” project uses a NODE MCU v3 board with ESP8266 chip. This points out usage of the board from ESP series for the developed AQI device. Criteria for the chosen type of ESP board are chip power, built-in wi-fi, and number of microcontroller pins. To ensure the operations of all selected sensors ESP32 chip is used as the most progressive of ESP series [42,43]. Also to ensure operations of all sensors a large number of RX\TX pins (UART interface) is needed, thus ESP32 DevKit is selected as the board with maximum number of UART pins.
• ESP32 expansion board (Espressif Systems, Shanghai, China) is used to increase the number of UART pins. Despite selecting the ESP32 DevKit v1 as the board with the maximum number of RX\TX outputs, their number for all selected sensors is not sufficient and requires the usage of an expansion board for the device.
• MicroSD card module (Shenzhen Fetuoda Electronic Technology Co., Ltd., Shenzhen, China) is used to store and reserve the sensor data on an SD card.
• SMT SIM7080G module (SIMCom Wireless Solutions Limited, Shanghai, China) is used to transfer the sensor data by NB-IoT networks. This module seems to be one of the most efficient SMT NB-IoT devices with optimal price/quality definition [44,45,46].

Enclosure and some fastener components have to meet the requirements of providing sufficient Ingress Protection codes, enclosures also have to give enough space for all placed-in components. The list of enclosure and fastener components is as follows:

• Spelsberg AL 2212-8 (Günther Spelsberg GmbH, Schalksmühle, Germany) is used as an enclosure for the device, it has 220*120*80 parameters. It provides IP66 protection, made from aluminum, thus more resistible for hits and falls and it is accessible for purchase in the large variety of countries. Of course, it could be replaced by some analog.
• Cable channels ID 3.49*6.09 OD Square Silicon Shielded Wire (Guangzhou Nafini Technology Co., Ltd., Guangzhou, China) is used to protect wires from the board to external components of the device. They provide IP68 protection.
• Rubber protective covers ZF-65-U (Optima Premium, Moscow, Russia) are used to house the external sensors of the device.
• Gland fasteners PG-9 4-8 DKC 52600 (Dielectric cable systems, Tver, Russia) are used to fasten the cable channels to the enclosure on the one side and to the rubber protective covers on the other side. Glands provide IP68 protection.
• Silicone Tube ID 6*8 OD (Shenzhen Tiasen Technology Co., Ltd., Shenzhen, China) is used to provide air intake of the SDS011 sensor.
• Screws of different diameters (M1.6, M2, M2.5, M3, M6) are used to fasten the components inside of enclosure.

Energy storage components must fulfill the requirements for optimal autonomy of the device without a high increase in its cost and size. The list of energy storage components is as follows:

• Four Li-ion Panasonic NCR18650B 3400 mA*h Li-ion rechargeable batteries;
• 4X 18650 battery holder to connect the batteries;
• TZT DDTCCRUB (Shenzhen TZT Technology co., Ltd., Shenzhen, China) device is used to balance the charge level between batteries.

3. Results and Discussion

3.1. Electrical Circuit Desing

Based on the selected electrical components and sensors, the electrical connection diagram of the device was developed. The schematic is shown in Figure 1. A detailed pinout is presented in Appendix A.

3.2. Sensor Design

3.2.1. General View of Device

A general view of the device design is shown in Figure 2. The main electronics are located in the housing of the Spelsberg AL 2212-8 device. External sensors are located in the ZF-65-U protective covers. They are connected by wires with electronics inside the housing, located in cable channels ID 3.49*6.09 OD. For protection against penetration cable channels on the side of the housing and on the side of rubber, protective covers are fixed with glands PG-9 4-8 DKC 52600.

3.2.2. Components’ Placement in the Enclosure

The internal components of the assembly are distributed on the bottom and four walls of the Spelsberg AL 2212-8 enclosure. The enclosure layout is shown schematically in Figure 3.

There are no components located on sides one and three of the enclosure.

The Elanzeme DEVKIT V1 30 pins expansion board, ESP 32 DevKit v1 board, Nova Fitness SDS011 dust sensor, and SO₂ converter module from the SPEC SO₂ sensor are installed on the bottom of the Spelsberg AL 2212-8 enclosure using screws. The location of components on the bottom is shown in Figure 4.

Dust sensor SDS011 is fixed to the housing by three-cylinder head screws, the screw thread diameter is chosen to be smaller than the diameter of the hole for the fixing element, which is equal to 3.2 mm. The length of the screw is chosen so that the screw does not pass through the enclosure. With a case thickness of 4 mm and a thickness of 4.1 mm for the SDS011 fasteners, a screw length of 7 mm is optimal. The screws A.M3-6g*7 are therefore suitable.

The selection of fixing screws for the other components is similar. The Elanzeme DEVKIT V1 30 pins expansion board is fastened to the enclosure using four A.M3-6g*7 cap screws. We attached the SO₂ AD converter module to the enclosure cover using four Allen screws A.M1.6-6g*6.

The microSD card module and the SIM7080G sim card module are installed on side 2 of the Spelsberg AL 2212-8 enclosure. The location of the components on side two of the enclosure is shown in Figure 5.

The microSD card module is attached to the housing using four A.M2-6g*6 cap screws. The SIM7080G sim card module is attached to the case using two A.M1.6-6g*8 cap screws.

In the bottom of edge four of the Spelsberg AL 2212-8 enclosure, we pre-drilled 15.5 mm diameter holes for the installation of PG-9 4-8 DKS 52600 glands, which are used to attach the external sensors to the components of the assembly by means of a cable channel, and for the installation of the measuring tube to the Nova Fitness SDS011 dust sensor.

Side four of the housing with holes and gland mountings is shown in Figure 6.

3.2.3. Components’ Placement on the Cover

The charger board TZT DDTCCRUB is installed on the Spelsberg AL 2212-8 case cover using double-sided tape; four Li-ion Panasonic NCR18650B batteries, which are inserted into the battery holder SZEKS 18650 Battery Case 4X, are fixed with screws.

The SZEKS 18650 Battery Case 4X battery holder is fixed to the lid of the case with eight A.M2.5-6g*5-cylinder head screws.

The location of the components on the enclosure cover is shown in Figure 7.

3.2.4. Dust Sensor Tube Fastening

To supply air to the Nova SDS011 sensor, a silicone rubber hose with an inner diameter of 6 mm and an outer diameter of 8 mm is used, which is connected to the outlet of the dust sensor and led outside the housing through a pre-made through hole with a diameter of 15.5 mm under the PG-9 4-8 DKS 52600 gland. The place of hose outlet from the box is secured with a gland.

The attachment of the tubing to the SDS011 sensor can be seen in Figure 4. The tube outlet through the hole in the housing can be seen in Figure 6. Securing the tubing with a gland on the outside is shown in Figure 2.

3.2.5. External Components Fastening

The SPEC SO₂ sensor, MQ131 ozone sensor, MQ7 carbon monoxide sensor, MH-Z19B carbon dioxide sensor, MQ137 ammonia sensor, and BME280 temperature, humidity, and pressure sensor are placed inside housings using rubber headlamp covers with a diameter of 45 to 65 mm and a depth of 25 to 100 mm. This is to ensure that the sensors are protected from environmental factors such as water, moisture, dust, dirt, etc. In the tops of the protective covers are pre-made holes with a diameter of 15.5 mm, for installation of glands PG-9 4-8 DKC.

The sensors are held inside the protective covers by the connection wires to which they are connected. The connection wires are placed in the conduits and led out of the housings. The fixing point of the cable duct and the hood is fixed with glands, which ensure a tight fit and prevent moisture from reaching the sensors. The glands are selected by the inner diameter corresponding to the outer diameter of the conduits. The selected PG-9 4-8 DKS 52600 glands are suitable for diameters from 4 to 8 mm, the outer diameter of the conduits is 6.09 mm.

The other side of the cable duct is fixed to the bottom wall of the junction box, in which through holes with a diameter of 15.5 mm were previously made.

The protective covers are placed at different heights to prevent them colliding and damaging the sensors inside. Thus, a SPEC SO₂ sulfur dioxide sensor and BME280 temperature and humidity sensor are placed at a distance of 64 mm from the bottom edge of the box, a Prana Air O3 sensor and Prana Air CO sensor—at a distance of 190 mm, and a MH-Z19B CO2 sensor and GS+4NH3-100 Ammonia Sensor—at a distance of 129 mm.

The arrangement of sensors on the outside of the device in protective covers at different heights, suspended on cable ducts, as well as the fixation with glands can be seen in Figure 2.

The location of the sensors inside the protective covers is shown in Figure 8.

3.3. Sensor Operation Modes and Lifetime

The question of the life cycle of a device until the batteries are replaced will always be related to the frequency of measurements. The more frequently measurements are required, the shorter the battery life will be. In any case, the user should set the operating mode of the device depending on the research being carried out, which should determine the need for the frequency of data measurement. Since the gases listed above are not measured frequently, and in electrochemical devices the number of measurements is directly related to the duration of the sensor operation, the autonomy will essentially be related to the measurement frequency of PM2.5 and PM10 dust particles.

For the same continuous measurement mode, such as in the mains-powered device of the “Sensor.Community” project, a preliminary estimate of the autonomy time will be about 4.5 days. For dust measurement modes in urban green infrastructure (1 measurement per 30 min) in standby mode outside of measurements, the autonomy can be increased up to 15 days. By setting the device to deep sleep mode outside of measurements, the autonomy can be increased up to 60–65 days.

3.4. Economic Assessement

Since the purpose of the device is to make it available to citizen scientists and volunteers, the cost of the device should be in line with the market for such devices. The cost estimate of the device, including all its components, is presented in Appendix B, for estimation we took average values on the market. The final cost of the device is estimated at $241.48. This is much higher than the cost of simple DIY sensors (for comparison, the cost of the dust monitoring station of the “Sensor.Community” project is about $35–40), but generally fits into the market with more complex devices of citizen science. To give an example, the actively used in citizen science monitoring acoustic open hardware sensor AudioMoth [47] costs about $249.98 ($149.99 for the sensor and $99.99 for the housing).

4. Conclusions

The presented concept of air quality sensors for citizen science aims: to produce a final concept for a marketable device for scientists and volunteers that integrates the measurement of key air quality indicators, as well as versatility in information transfer (Wi-Fi, NB-IoT, redundancy, or recording to SD in the absence of all networks) and independence from the power grid. It is based on existing devices and instruments used in citizen science, DIY developments, and the experience of their application in the assessment of AQI components.

Subsequent research in the prototyping and realization of this concept should aim for actual simplicity and ease of assembly of the device by people without appropriate technical skills; ease and simplicity of microcontroller firmware not requiring the skills of an experienced user; and investigation of the most commonly required modes of operation of the device to allow user selection and customization of the mode.