You might have noticed that you can use the AST UPAS mobile app to program the time-resolved particulate matter (PM) sensor included in the UPAS v2.1 PLUS to operate on a duty cycle (i.e., to turn on and off at regular intervals) so that the UPAS can run for a longer duration on a single battery charge. Which duty cycles can be selected?  What are the advantages and disadvantages of each option?  How should you go about selecting the best setting for your specific sampling application?  Keep reading to learn more!

What settings are available?

You can select a PM sensor duty cycle from the "PM Sensor Operation" menu within the AST UPAS mobile app. The 18 available options are summarized in Table 1. If "Continuous" is selected, the PM sensor stays on and measures continuously for the duration of the sample. If the UPAS is set to log at 30-s intervals, measurement data are averaged over each log interval and those 30-s averages are written to the sample log. If "2 Min Interval" is selected, the sensor warms up for 30 s, measures for 30 s, and then sleeps for 60 s; data from the sensor are averaged over the 30-s measurement period and written to the sample log every 2 minutes. The sensor operates similarly when the 3- to 15-minute interval options are selected. In each case, the sensor warms up for 30 s, measures for 30 s, and then sleeps for the remainder of the interval.  Data from the sensor are averaged over the 30-s measurement period and written to the sample log every 3 to 15 minutes.

If "15s-5s-10s Interval" is selected, the PM sensor warms up for 15 s, measures for 5 s, and then sleeps for 10 s; measurement data are averaged over the 5-s period and written to the sample log every 30 s. If "15s-5s-40s Interval" is selected, the sensor warms up for 15 s, measures for 5 s, then sleeps for 40 s; measurement data are averaged over the 5-s period and written to the sample log every 60 s. Finally, if "20s-10s-30s Interval" is selected, the sensor warms up for 20 s, measures for 10 s, then sleeps for 30 s; measurement data are averaged over the 10-s period and written to the sample log every 60 s.

How will the PM sensor duty cycle affect my data?

When selecting one of the interval operation modes in Table 1, your primary consideration is probably battery endurance, but it’s important to consider how your selection will affect the PM concentration data captured by the sensor. Consider the two scenarios below. In each scenario, 16 UPAS v2.1 PLUS were programmed to measure PM_2.5 for 5.5 hours. Each UPAS logged data with the PM sensor operating on one of the following eight settings; two UPAS were programmed with each setting:

Scenario 1: PM_2.5 concentration changes gradually over time

In this scenario, the UPAS measured PM inside a single-car garage where the door was closed for the whole 5.5 hours and a candle was burning for the first 2.5 hours. In Figure 1, the data recorded by each UPAS in which the PM sensor operated on an interval are compared to data recorded by each UPAS in which the PM sensor operated continuously. All UPAS captured the same trends in PM_2.5 pollution inside the garage over time. PM_2.5 data logged from the sensors that operated on the 15s-5s-10s, 15s-5s-40s, and 20s-10s-30s intervals are noisier than those logged from the sensors that operated continuously because the former three were 5-, 5-, and 10-s averages, respectively, whereas the latter were 30-s averages. Still, the relative standard deviation of the 5.5-hour-average sensor-reported PM_2.5 concentration was 8% across the 16 UPAS.

Figure 1. PM_2.5 concentrations measured over 5.5 hours inside a single-car garage with the door closed and a candle burning inside. In each panel, sensor-reported PM_2.5 concentrations logged using two UPAS in which the PM sensor operated on the specified interval (black lines) are compared to those logged using two UPAS in which the PM sensor operated continuously (red lines).

Key takeaway: If you are measuring in an environment where you expect the PM_2.5 concentration to only vary gradually over time, you should be able to capture that variation using any of the intervals listed in Table 1. Longer intervals will allow the UPAS to run for a longer duration on a single battery charge.

Scenario 2: PM_2.5 concentration changes rapidly over time

In this scenario, the UPAS measured PM_2.5 inside a small laboratory enclosure in which high concentrations of sodium chloride aerosol were introduced every 20 to 40 minutes. PM was continuously removed from the enclosure by an exhaust pump connected to a HEPA filter. We designed this experiment to approximate the transient PM concentrations that can be observed if lots of PM is emitted over a short duration (for example, due to a brief cooking activity) but the PM concentration decays quickly after the emission event because the environment is well-ventilated.

The data recorded by each UPAS in which the PM sensor operated on an interval is compared to the data recorded by each UPAS in which the PM sensor operated continuously in Figures 2 and 3. All UPAS captured the same trends in PM_2.5 pollution over time when the PM sensor operated continuously or on an interval in which data were logged every 30 or 60 s (i.e., 15s-5s-10s, 15s-5s-40s, or 20s-10s-30s). The relative standard deviation of the 5.5-hour-average sensor-reported PM_2.5 concentration was 10% across the 8 UPAS included in Figure 2.

Figure 2. PM_2.5 concentrations measured over 5.5 hours inside a laboratory enclosure in which high concentrations of sodium chloride aerosol were periodically introduced using a nebulizer. In each panel, sensor-reported PM_2.5 concentrations logged using two UPAS in which the PM sensor operated on the specified interval (black lines) are compared to those logged using two UPAS in which the PM sensor operated continuously (red lines). Only interval settings for which PM sensor data were logged every 30 or 60 s are shown.

When the UPAS was programmed so that the PM sensor turned on and measured at less frequent intervals, fewer of the peaks in PM_2.5 pollution were captured (see Figure 3). For example, the sensors programmed to operate on the 5-minute interval largely missed the pollution events that occurred shortly after 14:30 and shortly before 15:00. The sensors programmed to operate on the 10-minute interval missed the pollution events that occurred before 12:30, shortly after 12:30, shortly after 13:00, shortly before 14:00, shortly after 14:30, at 15:00, and shortly after 15:00.

Figure 3. PM_2.5 concentrations measured over 5.5 hours inside a laboratory enclosure in which high concentrations of sodium chloride aerosol were periodically introduced using a nebulizer. In each panel, sensor-reported PM_2.5 concentrations logged using two UPAS in which the PM sensor operated on the specified interval (black lines) are compared to those logged using two UPAS in which the PM sensor operated continuously (red lines). Only interval settings for which PM sensor data were logged every 120 s or less frequently are shown. Solid and dashed lines distinguish the two replicate UPAS programmed with each setting.

Because some pollution events were missed, some of the PM sensors programmed to operate on the intervals shown in Figure 3 measured lower 5.5-hour-average PM_2.5 concentrations than the two PM sensors that operated continuously. Conversely, because the low baseline pollution level was measured less frequently, which caused peaks in pollution to appear artificially wide, some of the PM sensors programmed to operate on the intervals shown in Figure 3 measured higher 5.5-hour-average PM_2.5 concentrations than the two PM sensors that operated continuously. Overall, 5.5-hour-average PM_2.5 concentrations calculated from the data logged by the UPAS operating on the 2-, 5-, 10-, and 15-minute intervals ranged from 16 to 69 µg m^-3, while two PM sensors that operated continuously measured 5.5-h-average PM_2.5 concentrations of 31 and 33 µg m^-3, respectively.

Key takeaway: A PM sensor operating on a long measurement interval (for example, 5-, 10-, or 15-minutes) might do a poor job of capturing short-duration pollution events, whereas the “15s-5s-10s”, “15s-5s-40s”, and “20s-10s-30s” interval settings will do a better job. If you want to measure PM_2.5 in an application where you expect a lot of pollution events to be short in duration, and you have enough battery capacity to run the PM sensor for 50% of the sample, you might be better off using the "20s-10s-30s" interval rather than the 2-minute interval. Similarly, if you only have enough battery capacity to run the PM sensor for 33% of the sample, you might be better off using the “15s-5s-40s” interval rather than the 3-minute interval.