Calibrating Carbon Monoxide Sensors
Since we lacked a reference monitor for Carbon Monoxide (CO) and generating CO is a dangerous procedure, we opted to conduct a step-calibration through the use of a 5L chamber and gas standard which we diluted with Zero Air Gas (ZAG).
We performed a step calibration at concentrations of 0, 1, 2, and 4 ppm CO. We achieved these concentrations by diluting a 10 ppm gas standard with ZAG. The setup, process, and outcome are described more below.
Tubing from a 10 ppm CO gas standard cylinder was connected to a mass flow controller (MFC) which then combined with tubing from the device used to generate ZAG from air. A MFC was also used connected to the ZAG line to control the flow. The combined tube was fed into the 5L canister through a steel tube which released the gas mixture near the bottom of the chamber in an effort to generate plug flow.
Three beacons (not pictured) were placed inside the 5L chamber with power cables running out of the top of the chamber. Three beacons were placed just outside the chamber and represent the next set of beacons to be calibrated. A vent hood was placed over the chamber in an event of any leaks or issues with the MFCs that might have released excess CO.
These experiments took the longest to conduct (9-hour minimum) of all calibration efforts and were additionally slow because calibration could only occur in batches of three. The process is as follows:
- Beacons were powered on outside of the chamber and allowed to run for at least one-hour since the CO sensor takes approximately one-hour to start reporting valid measurements. These devices were usually left to run while another set of beacons was being calibrated.
- Beacons were then placed within the chamber.
- The step calibration started with a baseline i.e. pure ZAG. If possible, we would keep the beacons in pure ZAG for a 24-hour period. Otherwise, we maintained a concentration of 0 ppm CO for 2-hours.
- Following this two hour period, we toggled the proportion of CO to ZAG to achieve 1 ppm, 2 ppm, and 4 ppm CO. We maintained each of these concentrations for a period of two hours.
- After the 4 ppm step, we flushed the CO out of the chamber and removed the beacons, replacing them with the new set of three devices.
As with the other parameters, we developed univariate linear least-squares regression models to correct CO readings. However, for CO we developed these models based on four data points, corresponding to each of the steps in the calibration process. For each Beacon, we started by calculating the mean concentration measured during the middle hour period (truncating the first and last 30 minutes) of each step to determine what the Beacon was measuring at 0 ppm, 1 ppm, 2 ppm, and 4 ppm concentrations. The models we created were aimed at correcting these values to the true step concentration. Example results are shown below.
