Air Quality Projects

History of EME Systems' involvement with air quality measurement

In 2001, Kirk Smith and Rufus Edwards from U.C. Berkeley department of Environmental Health Sciences approached EME Systems about the possibility of converting standard household smoke detectors for the purpose of measurement and recording of household air pollution (HAP). They realized that those detectors offer high technology with great economies of scale, and if possible to convert, would allow economical and extensive coverage at research sites, where standard air monitoring equipment was bulky, hard to deploy, and prohibitively expensive. At the time, a randomized trial was being organized by the World Health Organization to take place in the Guatemala highlands, the first of its kind to attempt to assess the health benefits to be had from the reduction of HAP through innovations in rural stove technology. At EME Systems, we took the challenge and readily tapped into the signal from an ionization smoke detector and fed it to a data logger, EME's own OWL2pe. The log readily showed events in the kitchen. On that basis and upon extensive discussions with detector experts Dave Litton at NIOSH and Jon Wesenberg at First Alert, and with funding from the Shell Foundation, we went on in 2002 to develop, characterize and deploy the UCB-PATS described below. We also went onward to additional engineering projects related to air pollution and exposure.

UCB-PATS (University of California-Particle and Temperature System)

Ionization smoke detectors are best at detecting nano-particles, less than 0.3µm, that come from flaming fires, whereas photoelectric detectors are best at detecting particles 0.3 µm and up that are characteristic of smoldering fires. For this project, we settled on a dual mode smoke detector that incorporates both types, in order to cover the wider range and to have better discrimination of the source. The First Alert SA-301, available in 2002 was especially convenient, because the ionization and photoelectric chambers were located on a separate circuit board. This allowed us to replace the First Alert processor board with our own, based on the Parallax Basic Stamp. It included a stable 10 volt power bias for the ionization chamber, a sample and hold for the photoelectric signal, temperature and humidity sensors for compensation, data logging memory for weeks of deployment, an RS232 interface for download to a PC, and constantly evolving firmware. The firmware had a raw mode for access via any terminal program, and was also programmed to work with a custom management application developed by UC Berkeley for ease of use by technicians in the field. In 2006, due to regulatory concerns with the radioactivity in the ionization detector, we switched with some regrets to an all photoelectric, which still correlated well with pm2.5 mass. With the formation of Berkeley Air Monitoring Group in 2008, the UCB-PATS became commercially available and was used extensively in studies of household air pollution in at least 14 countries worldwide. It had other applications as well, for example, in logging smoke wild-land fires. In 2014, First Alert phased out the SA-3022, and we decided to take a different tack as we looked for a replacement.

PATS+CO Particle, Temperature and Carbon Monoxide System

In 2014-2015, a whole new generation of particle sensor arrived, aimed at the household air cleaner market. These were modules designed for original equipment manufacturers, so they were relatively cheap, available, and did not need to be hacked out like we had done with the original UCB-PATS.  After evaluation, we settled on the Sharp GP2Y1014au0f, due to its small size, ease of integration, and its linear response that we were able to extend to cover high smoke levels that are found in rural biomass kitchens.  The standard Sharp sensor covers up to around 0.8mg/m^3, but with controlled drive to the LED, we were able to improve that to at least 2500mg/m^3. A micro-blower keeps the air moving at a low rate through the particle detector and past the CO, temperature and humidity sensors.  The current PATS+CO has a powerful multicore Propeller microcontroller that logs data to SD card, USB interface for data and for charging the LiPo battery, and new firmware that still offers a  simple terminal interface, but which also talks to Berkeley Air's new PICA data management system.


The miniPATS came about as a small version of the PATS+. It is intended to plug into a parent system that performs a variety of other tasks besides the particulate monitoring. The miniPATS is a tight fit around the Sharp GP2Y1014, and it includes the microblower for air movement, and temperarture and humidity sensors.


Dylos Air Quality Monitor

Dylos Corporation came out with its DC1100 air quality monitor in 2010. It was and is primarily intended to help persons who suffer from respiratory sensitivity, or to help folks such as woodworkers who have need to track conditions in their workshop. We were immediately intrigued by its possibility for research purposes. In initial tests, we found it to be extraordinarily sensitive. As a laser particle counter, it displays a count of particles in two size bins. While we recognized that it would not be ideal for third world deployments due to its limited range at the high end and its power requirements, it was a great candidate for local use in first world settings in California.


wSUMS, wireless Stove Use Monitoring System

It is desirable to collect data about the use of improved cookstoves in homes without having to disturb the occupants. With an innovation grant from the Vodafone Foundation, and in cooperation with UC Berkeley EHS, stove developer Biolite Corporation, and Mexican NGO GIRA, we developed a data logger that could be interrogated wirelessly using short range radio (XBee 802.15.4)  The wSUMS collects data on stove use. Occasionally, the research technician passes by and unobtrusively, without entering the house, wirelessly offloads the data using a device built specifically for that purpose. The data is compressed to daily summary statistics, in order to fit in small packets that can be transmitted quickly and reliably. The device was adapted to use with both rocket stoves, and with efficient Mexican Patsari stoves.

Time-Activity Monitors (TAMs)

In the context of improved stoves and health outcomes, it is difficult to assess how much exposure individuals actually have to sources of air pollution on a daily basis within the household micro-environments. Family self-reporting of time in the cooking area versus elsewhere is unreliable. Babies and small children are especially susceptible, and they spend a lot of time with their mothers while meals are prepared. TAMS is an ultrasound-based system for monitoring exposure time and duration. Each of up to 4 participants wears a small tag, which at intervals of about 30 seconds, emits a coded 40kHz ultrasonic identification code. A receiver mounted on the wall in the area of interest picks up the signal and logs the time and the presence of the that tag. We chose ultrasound for this project, over alternatives of radio or light, because it does echo within a room, but it does not go through walls. Participants carry the tag in a cloth pouch chosen to minimize attenuation of the ultrasound. The decoding algorithm took substantial development and proving time, to handle reverberation in the room, extraneous noise, and overlapping signals.

kSUMS, thermocouple data logger

While Berkeley Air Monitoring Group provides iButton SUMS (Stove Use Monitoring System) for monitoring the temperature and usage of stoves, there are situations such as open fires ("three stone") and some traditional fireplaces build of mud and local materials that are not suitable for attachment of iButtons. Additionally, the temperatures are too hot and varied. And there is need for a reference ambient temperature and simultaneous monitoring of multiple points. The kSUMS accepts 3 to 6 k-type thermocouples and logs the temperatures in long-term memory so they can be downloaded at intervals. The kSUMS can be configured for other purposes and can accommodate additional senors. It is build around the Propeller multicore processor with a USB interface for data and battery charging.


There is a growing understanding that that stoves that burn biomass, even though improved, are still prolific polluters. One alternative is the LPG (Liquified Petrolium Gas) stove, which, although clean burning, requires an easy supply of gas. In India, there is an initiative to subsidize the use or LPG. In a controlled study, the local researchers want to provide incentives for the use of the LPG stoves, especially by pregnant women. The subsidy can come in the form of conditional cash transfers, dependent on actually using the stove as verified in some way, or simply via free fuel. The PinkKey, the round pink object in the photo, is a small data logger that plugs into a socket mounted on the front of the stove. It uses cheap silicon diodes as sensors attached thermally to the two burners. The pinkkey after some time is unplugged and its data transferred to provide a total amount of usage and a count of "meals" based on an algorithm developed at by Ajay Pillarisetti at U.C. Berkeley Dept. of Environmental Health.

ARMS, Air-Exchange Rate Monitoring System

In homes that use biofuels, there is danger of carbon monoxide and particulate buildup. An extensive study in Nepal and Mongolia sought to quantify air exchange rates in traditional homes. This was done by releasing carbon dioxide into a room, and tracking the dispersion and decrease of that gas through time. To accomplish this, we build data loggers for the project that were called the "Squid" in addition to the more formal ARMS. This logger connected via cables to between one and four COZIR NDIR carbon dioxide sensors distributed around the room. (Non-Dispersive InfaRed). The sensors are powered from 3.3 volts, and a small fan is mounted on the CO2 entry port to help force in air and speed up the response. Without the fan, the sensor with diffusion alone was too slow. The typical experiment in the field lasted less than an hour, so gradual drift of the sensor was not important.


2229 5th Street, Berkeley, CA 94710 • (510) 848-5725 •