On a cold and windy day in mid January, we went out to the garden to run two tests on the ozone monitor. We had been getting unusual noise and spikes in our data, and wanted to see if there was something wrong with the ozone monitor. We ran a lamp test, which sees if the voltage of the UV lamp is between 0.6 and 2.2 volts, which is recommended. The ozone monitor works by pumping air past a UV lamp with a UV photodiode sensor on the other side. Ozone will absorb the UV light, so by measuring the amount of UV light that passes through the other side, we can measure the amount of ozone. If the voltage is too high, the monitor is overheating, and if it is too low, the lamp is burning out and needs replacing. We also ran a zero test that determines what the reading corresponding to zero ozone concentration is. Neither test showed anything significantly wrong with the monitor.
We decided to send the monitor back to the manufacturer for repairs. Meanwhile, we did some of our own analysis of the data from the ozone monitor during the spikey time period. Along with the amount of ozone, the monitor also records cell temperature, cell pressure, flow rate, and the photodiode voltage (pdv). The pdv is simply the current voltage of the ultraviolet lamp. The absorption cell is the part of the ozone monitor where air travels through while passing the UV lamp. We brought all these sets of data into Igor and began to graph them.
We identified time periods of suspicious spikes in the ozone concentration data, including 1 hour both before and after each period. Then, we pulled diagnostics from the laptop and isolated the entries for those time periods. After some work, we finally inputted all this data into a table in Igor, and made a single graph with multiple y-axes showing all of our variables vs time. From this diagnostics graph we saw that the lamp voltage was very low whenever it was really cold outside. This lamp does not operate well at temperatures below freezing, and tends to “flicker”. This will affect the level of light that is being emitted by the lamp, and thus absorbed by the ozone. The second issue that makes the measurement noisy is that our flow rate and cell pressure was low, indicating that our lines might be dirty and there might be particles in the system. These particles can also interfere with the light, changing the amount of light a little bit, and causing a bad ozone measurement.
More recently, the ozone monitor came back from the manufacturer, who had replaced the UV lamp. Before re-installing the monitor outside, we had to clean the tubing and calibrate the monitor. There will be more about cleaning and calibrating in upcoming posts.