On the PurpleAir map, you may have pulled up a sensor’s graph and noticed two different sets of data points denoted as “A” and “B.” We call these “channels.” In the example below, you can see a graph that shows data from a sensor in the Salt Lake City area (UT, USA).
If you look closely, you can see blue and black lines on this graph. The blue line represents channel A, and the black represents channel B. Each channel is connected to a laser counter in a sensor. Our outdoor sensors include two laser counters, which measure air particulate individually. We compare the correlation of these two channels to create a measure of confidence—called the Confidence Score—that serves as an assurance of data quality.
A and B data are not exclusive to PM measurements; they are a theme with PurpleAir data. Below, you’ll see an example of field descriptions in our API documentation. You’ll see that even the environmental sensor inside a PurpleAir sensor can report A and B data.
This will be the case for all components with multiple outputs. If only one output is reported, such as the single laser counter in indoor sensors and the environmental sensor in certain PurpleAir sensors, only channel A will return data. The average will still be present, but the value will match the base output.
Note that the majority of sensors on the PurpleAir map do not report B data for temperature, humidity, and pressure.
So what does it mean when I have a very low confidence score and the sensors report vastly different values? My confidence score is oven way below 30% and the A & B sensors are vastly different (see graph below).
I don’t see any cobwebs or other detritus in my Classic monitor. Is it near death? Do I need to clean it somehow (suggestions appreciated so I don’t damage it) or do I need to replace it?
It’s possible that there is debris inside the laser counter (channel A) that is blocking particles from moving through or covering the lens. We recommend cleaning the sensor: Sensor Maintenance. As blowing air through the air channel can can the fan to move too quickly, causing damage, we recommend that you apply short bursts of air or use a toothpick to hold the fan in place.
I decided to take Purple Air up on the offer to trade-in my two sensors (outdoor PA-II and indoor PA-I, as arranged with Jacqui Hunt) so they can be refurbished and resold.
I’ve ordered two new Zen sensors. They should be delivered tomorrow.
Hi, I am trying to do QAQC checks on my PA zen sensors before I deploy them. I will just be using SD card data since I will using them in a location without WIFI. When reviewing some of the raw data I noticed there is a difference in the channel A PM data and the channel B PM data. Is this something I should be worried about? Also is there a channel that is more reliable if I am solely hosting these sensors indoors?
It’s typical for there to be some difference, but a large difference could be an issue. Can you share some of the data or a graph? If you prefer to share it via email, you can send it to contact@purpleair.com.
Here is a small snapshot of the differences I am seeing. I don’t know if there is a recommended deviation should be seeing. I tried to highlight all the A and B channel comparisons that seemed to stray too much from each other.
I will say that this is just the raw data and has not gone through the EPA correction factor. Unsure if that has anything to do with the values observed.
Hi Andrew,
Do you think there is an issue if channel A has a hour average of 11.799 and channel B has an hour average of 9.99 or is this a normal deviation?
The level of discrepancy in both sensors appears to be within tolerances for our testing, but if there are issues, they will more clearly present themselves when the pollution levels are higher. We can check the data from when these sensors were tested if you can provide the sensor indexes. Additionally, if you have any examples of data showing higher pollution levels, such as >35.4µg/m3 (above 100 US EPA AQI), please share them.
Channel A hour average of 11.799 and channel B hour average of 9.99 is a normal level of deviation.
