The Wynd Halo 1st Edition uses a proprietary PM sensor with a blue-light laser instead of a red-light laser, which presumably makes it more accurate and precise. Interestingly, I remember during the Kickstarter campaign of the monitor, backers were asking the company if they will sell the sensor as a component part to third parties, and they said yes because it was about to revolutionize the industry.
Although I don’t have a reference monitor, the comparison I conducted is considered a field evaluation in a real-life situation. AQ Monitors are designed to operate inside apartments and buildings for this reason I personally value more a field evaluation in situ than a lab evaluation where every parameter is controlled.
This kind of evaluation won’t tell us how accurate is the sensor, but it will reveal the correlation against other monitors in order to determine if measure something respectable. For example, there are many field and lab evaluations for the commonly found Plantower PMS5003 sensor that demonstrate a correlation of 0.78 – 0.93 against various reference monitors.
I had the devices running side by side for 2 weeks. I picked a date randomly, but I made sure humidity didn’t surpass 70% to avoid hard interference and although I had thousands of measurements, I ended up creating 1-h average values for all the parameters to avoid time inconsistencies. Some monitors take measurements every 10 seconds, others every 5 minutes and others every 15 minutes.
Thanks to the free software RStudio and some packages I ran various Pearson correlations tests between four monitors. I am not going to reveal the name of the competitors only the sensors they use.
It is important to know which sensors are used in order to make the right conclusion in the end. The numbers we see in the table below help us understand better the correlations. Numbers closer to the value of positive 1 mean that we get the best correlations. Negative values indicate that both variables move in the opposite direction.
(1) Although they use the same VOC sensors by Sensirion, the results are different. I think this is a design issue, in particular, the VOC sensor inside the AQ2 gets hotter than the normal as it is placed under the CO2 sensor, as a result, the readings are inconsistent.
(2) It seems to me now that AQ3 has an issue with the temperature sensor as across all the correlation results it scores the lowest. It is placed in a good location inside the device away from interference.
(3) Wynd Halo’s proprietary PM sensor seems to be the worst. Very bad PM2.5 and PM10 correlations. The log file kept registering 1μg/μ3 for PM1.0.
The Air ID feature came with the version 2.1.02 of the iOS app (currently v2.1.03). Supposedly, Air ID combines raw sensor data with contextual data from the cloud and tells you if the air has pollen, forest fire smoke, or smog from the nearby industrial plant. Unfortunately, it was again a disappointment. I saw some inconsistencies and the information it provides you with is simple and frankly some areas don’t support the feature. An Air ID that I got followed the following logic: If VOCs are present, it tells you Poor Ventilation. Nothing contextual in my opinion.
I have tried to contact the company and ask them to comment about the results, unfortunately they don’t seem to care. The idea behind the technology is great, however, I do believe that you cannot create an amazing product by selling it that cheap. US$149 is cheap especially when you need to invest in R&D for a sensor. Maybe in the future they will be able to pull it through.