PurpleAir: IoT Sensor Soars To 13,000 feet

Air pollution sensors find IoT niche.


Startup PurpleAir makes a living counting dust. Its consumer-level sensors measure particles of dust in air—airborne particulate matter (PM)—using laser particle counters that can count particle sizes 0.3µm up to 10µm. Add power, WiFi, the Internet, and PurpleAir pulls all the data from the devices into an online map showing extremely local air quality around the world, wherever devices are installed. The relatively accurate sensor, reasonable cost and convenience to own and set up, is growing the user base.

Concern is an underlying factor growing the user base, but what accelerate additional sensor installations is fire. Especially in California where colossal fires have in the past few years have tragically destroyed towns and lives, the smoke produced has been another long-distance danger. The smoke from the 2018 Camp Fire in Butte County, California, lingered over the San Francisco Bay Area for a least a week. A chart on an Energy Institute article shows a spike in PurpleAir sensor installations after the fire.

“It’s a real…I don’t think it’s just a fad. I don’t think it’s going to go away,” said Tim Dye, TD Environmental Services, LLC (Petaluma, Calif.). Dye is an independent advisory consultant who works with community groups, schools and other organizations to understand and install the technology. He also does data management and data interpretation from these systems. “And part of that is because some of the organizations that I’m working with are keenly interested in this.”

Even a mountain is being monitored. A PurpleAir PM sensor was installed in its highest altitude location on July 29, 2019, on the side of a small astronomy dome in California’s White Mountains, at the White Mountain Research Center, at approximately 13,000 feet.

The data the company and other IoT air monitoring companies is helping cover a gap in data, both quantity of data and type of information. PurpleAir’s sensors measure mostly PM 2.5, which are 2.5 microns or smaller. “Those are the small particles that get into our lungs,” said Dye. “It can’t measure larger, like dust particles. It doesn’t do well.”

But this is okay. “PurpleAirs are good at detecting smoke dust in the air,” said Dye. He said even though some new, temporary reference monitors went online during the wildfire emergency in California along with the existing ones, “there are only so many reference monitors around the state. But these sensors are good enough to fill in some of the gaps and give an indication of how much smoke is in the air. Is it a little, as a moderate amount, is it a lot, is it, you know, a whopping amount? It will give a relative, you know, really good relative indication. So that allows people to figure out where it is clean right now or where is it smoky, and where’s that going.”

While not the only consumer IoT air-quality sensor of this class, PurpleAir compares favorably to the regulatory sensors that are the professional grade air quality sensors used by government agencies costing $15,000 to $20,000 each.

But PM sensors are not perfect yet. “Some of them can be inaccurate. But there are others—and PurpleAir is one of these—that detect particles in the air and do a good job at that,” said Dye. “Out of the box [PurpleAir’s sensor] reads a lot higher than the reference monitors run by state and local and agency air-quality agencies, like professional grade,” Dye said. “But having said that, you can definitely follow the trends. When there are more particles in the air, it’s going to be increasing. When there are fewer, it’s going to be decreasing. And if you follow the trends very quickly, you can see that [the PurpleAir sensor] does a decent job at measuring certain types of particles in the air.”

Sensor data is sometimes hard to clean up. “A number of groups are starting to work on algorithms and methods to clean it, or to normalize it so that it’s more like what’s really occurring,” said Dye “One of the things that was interesting is during the fires last year, the PurpleAirs read high. They read about 45% high. And there are ways that are fairly straightforward, where you can read, you can normalize that, and you can make it closer to what’s really happening.”

PurpleAir knows its sensors read high sometimes. In an FAQ on its website, PurpleAir says “PM2.5 from wildfire smoke will have a different density than PM2.5 from dust blowing off a gravel pit. This means the mass concentration reported by a PurpleAir sensor can vary depending on the specific composition of PM for a given area. Thus making the sensors appear to “read high”. So far, 2 different research groups have completed studies for their areas to create conversion factors for PurpleAir specific to the composition of the air shed (AQandU and LRAPA on the PurpleAir map).”

More algorithms are to come.

Fig. 1: PurpleAir’s sensors overestimated smoke during the 2018 Camp Fire in California, when compared with a regulatory sensor, but it is consistently picking up the spikes in PM2.5. Source: TD Environmental Services

Fig. 2: The astronomy dome sits at approximately 13,000 ft above sea level, above the University of California’s high-altitude research center White Mountain Research Center, near the Ancient Bristlecone Pine Forest in Inyo County, California. The PurpleAir sensor is mounted on the opposite side of the dome as pictured above. Photo: Susan Rambo/Semiconductor Engineering

Fig. 3: PurpleAir sensor mounted to the astronomy dome at White Mountain Research Center in California. The sensor is the highest PurpleAir sensor as of July 29, 2019, according to the installer. In the background is White Mountain, a 14,000 foot mountain in the White Mountains. Photo: Susan Rambo/Semiconductor Engineering

Fig. 4: Looking up into the PurpleAir sensor.

Fig. 5: The PurpleAir sensor at White Mountain Research Center at approximately 13,000 feet.

Professional grade vs. consumer IoT
The expensive regulatory grade monitors always will be needed. “Every airport has the weather station. That’s an expensive weather station,” said Dye. He explained that if you’re landing planes based on the weather info, you better have reliable data from a rugged system, not some $100 sensor the airport bought on Amazon. The point is accurate, reliable, expensive monitoring systems will always be needed.

“The decisions that are made, whether it’s weather, or whether it’s air quality, really require high-quality information. So we have, you know, a network of 2,000 or more reference stations that are run by various air agencies. And we need that data for policy, to measure changes in air quality, to look at, you know, emission reductions, and are those effective,” said Dye. “But this new technology is, I think it’s game changing, because it allows individuals, other groups, research institutions, to begin to install monitors to look at more local air quality issues, and problems.”

PurpleAir says on its site that the way its sensors measure PM and how they do averages time of data collection are two main differences between its sensors and regulatory sensors. “There are two major difference for PurpleAir sensors when being compared to regulatory sensors, (1) the method of measuring particulate matter and (2) the averaging time of the data collected.”

“Federal reference sensors typically measure mass concentration of PM directly. This method is expensive, difficult to install, requires a specialist to maintain the sensor, and report on an hourly scale,” says PurpleAir’s FAQ. “Because of this many cities have a limited number of these sensors (or none at all) and it’s not feasible for the general public have their own.”

The PM sensors have two lasers in them made by Plantower, which also makes the counting algorithm. A fan brings in air so particles can be examined.

“They have a device that measures the light that scattered from particles, so it has a tiny laser, as a photo diode or light receiver. And as particles flow through that, that interrupts the light from the laser,” said Dye. “And you can count effectively or make an estimate of the number of interruptions, and that’s proportional to the amount of particles in the air.”

A chip, the ESP8266, with Arduino code, handles the data. And an antenna or a SD card are used for moving data.

Growing market
PurpleAir is not the only company making air-quality sensors. You can go on Amazon and buy a reasonably priced PM sensor to test air quality. “In the last five years, it’s been real progress in improving these devices, making them more accurate. It’s getting out there. More, more people looking at the data and deploying and using these for variety of things,” said Dye. “It’s not just two person startups.” Intel and Bosch are involved making air quality sensors systems.

“And it’s only going to increase. This is not a trend that’s decreasing over time,” said Dye. The reference sites in California only went from under 200 in 2016 to 1,500 in 2018. “The enormous increase in the number of those [reference sites] out there is primarily due to their costs.”

Fig. 6: The growth of air pollution sensors in California from five PM2.5 sensor vendors. Source: TD Environmental Services

The environmental sensor business, especially the air-quality sensor, is one to watch. “In five years, we’ll be looking at 10s of thousands of these devices across the U.S. and maybe even more, probably even more. I think they’ll be a lot better,” said Dye. “I think we’ll see the emergence of both outdoor, and especially indoor air monitors. Keep in mind that we spent a long time indoors. Sometimes indoor pollution is more polluted than outdoors. So having that kind of understanding of what’s inside and what’s outside is important.”

Yet how big the demand will be for these sensors is unclear, he notes. “Some of the big companies—they’re just beginning to get into this. How big the market is and how big the demand is—it’s a little bit uncertain.”

Leave a Reply

(Note: This name will be displayed publicly)