ezPyro SMD detectors for Non-Dispersive Infrared (NDIR) sensing of Carbon Dioxide (CO 2 ) The use of ezPyro sensors for NDIR gas sensing of CO2 based on a waveguide reference design, completed in co-operation with the University of Strathclyde Also see IEEE Sensors Journal, 19(15), 6006-6011. (https://doi.org/10.1109/JSEN.2019.2911737) for more details.
ez ezPyro sen sensor sors f s for or d detec ecti tion on of of C CO 2 2 us using ng NDIR This presentation describes the use of ezPyro sensors for NDIR gas sensing of CO 2 based on a waveguide reference design. Key NDIR module design features investigated here include: § Size of the module form factor design …. Utilising a SMD IR Pyroelectric detector and SMD IR emitter to make it is small as possible § Low power consumption, fast data collection § Single channel/gas measurement capability and stability. § Small optical path length (~ 2.5cm), but capable to detect <50ppm CO2 . All measurements were done in a controlled gas chamber at the University of Strathclyde. 2 12/1/2019
Pe Performance Monitoring The performance of the both the ezPyro detector and the NIDR module were monitored as a function of various key NDIR module design features. Key NDIR module characteristics investigated include : § Spot measurement capability and limits of detection for CO2 sensing (in ppm) § Effect of emitter type on gas sensing capability § Effect of flow cell vs passive cell (gas flow rate) § Effect of temperature § Behaviour over different incoming CO2 concentration regimes The stability of the device for long-term measurements was also determined by an Allan Variance analysis. Key performance parameters include the limits of detection(LOD) for the module and the spot measurement resolution in parts per million (ppm). 3 12/1/2019
ez ezPyro NDI NDIR Gas Mod odule Setup The CO 2 NDIR gas module was designed to use a single detector, ie no reference detector was included. Pyreos ezPyro™ SMD detectors (ePY12231) and breakout boards/stamps (ST Nucleo-F303K8) were used to read out and collect data using ezPyro™ software. Axetris LabKit IRS G1 drive board was used to drive emitters, with peak currents of 85 and 67 mA for the Filament(tungsten) and Axetris(EMIRS50 AS01T) sources respectively. All measurements were conducted at emitter frequency of 5Hz due to instability of some emitters (filament lamps) above 5Hz.
Gas Gas Chamb amber er Setup Constructed from high pressure/high vacuum components sealed with copper gaskets and con-flat knife edge connections ensuring gas tightness. 5 12/1/2019
Con Concentration on M Measurements : : 0 t 0 to 10000p o 10000ppm Averaged peak-to-peak values at • increasing gas concentration. Zero analyte measurements show • difference between vacuum and dry nitrogen. Larger concentration range exhibits • non-linear behaviour To record these measurements the timed data-saving option was used.
Con Concentration on M Measurements : 0 to 500ppm Lower concentration • measurements follow a linear decay in amplitude. Zero analyte taken with nitrogen. •
Conce centration Measurements: Combined ranges Combined data plotted as concentration V’s • averaged ezPyro output with best fit.
Temperature Dependence The temperature of the sensor was also monitored to investigate periods of signal • instability. A type-K thermocouple was positioned to be in contact with the ezPyro. • Temperature data was recorded using a PicoLog temperature logger.
Temperature Dependence During this test the lab temperature was increased during the • measurement, allowed to settle at a mid point then increased again. It appears the signal is responding to a 0.4°C variation in the sensor • temperature.
Al Allan V Variance In e Introd oduction on An Allan variance plot visualizes the effect of signal averaging, drift and noise types. A plot of variance ( σ 2) or standard deviation ( σ ), often referred to the Allan Deviation (units of ppm) in optical signal intensity as a function of averaging time for the detection system. At shorter averaging times the measurement resolution is limited by the white noise and therefore the variance is inversely proportional to averaging time. However, after some optimum averaging time, drift effects can be seen to start with a subsequent rise in variance as averaging time is further increased. The point at which this happens, and the extent to which performance deteriorates thereafter, are both application– and installation– specific for a given detection system and setup. (DT0064, ST)
Allan Variance with Filament bulb emitter • Single measurement detection was ~50 ppm with a LOD of 8ppm after an integration time of 110 min before systematic drift began to dominate the measurement. 12 29/04/2020
Allan Variance with Blackbody emitter Controlling the temperature with Axetris SMD black body source the system achieves much better results! • Single measurement detection was 22ppm with a LOD of 3ppm after an integration time of 150 min before systematic drift began to dominate the measurement.
Flow Measurements By comparing the response of the ezPyro output to increasing flow rates • it is clear that flow affects the output by cooling the system. The difference between the response of the ezPyro output signal may be a result of • variation in the gas temperature. This was not monitored but it may be appropriate in further experiments.
Flow Measurements Recording data for 1 second every 2 • seconds over a total of 15 minutes and at a flow rate of 600 ml/min. Each concentration step was recorded for 2.5 minutes finishing with nitrogen until the recording period ended.
Flow Measurements • As an approximation of human breath calculations showed that a flow rate of 600 ml/min would be greater than normal human breathing through the greater diameter flexible tubing used in a ventilator system. In real life the exhaled CO 2 concentration • in breath is ~5%.
Temperature Response With a static sample of gas in the cell the sensor was heated using a • temperature controlled flexible heating element. The sensor temperature was monitored independently. As the ezPyro has a built in • temperature monitor in it. it would sensible to use this as a way of compensating for temperature variations.
ez ezPyro CO CO 2 NDI NDIR Measurement Results Summary We successfully demonstrated long-term measurements of CO 2 using ezPyro in various gas environments/temperatures up to 1000mins. Despite measurements being highly sensitive to surrounding/system temperature effects, good stability of signal was observed with increasing temperature. Using the ezPyro™ detector for CO 2 measurement at 5Hz yielded the following limits of detection/spot measurement performance : Filament lamps = 4ppm / 48ppm § § Axetris EMIRS50 = 3ppm / 32ppm (22ppm achieved with t=35oC) Flow measurements lead to cooling of the system and therefore indicate the need to compensate for this heat loss for accurate measurement. It is likely that improved limits of detection/spot measurements can be achieved by : § operating the emitter/detector system at higher frequencies which are optimum for the detector (eg 20- 30Hz) § Including a reference in the optical module design § Optimized optical design for the gas module. 29/04/2020 18
Thank you to David Wilson and Michael Lengdon at University of Strathclyde Please see IEEE Sensors Journal , 19 (15), 6006-6011. (https://doi.org/10.1109/JSEN.2019.2911737) for more details. Visit www.pyreos.com ! 19 4/29/20
Recommend
More recommend