RoadTest: Sensirion Environmental Sensor Shield - Industrial Sensing
Author: mloik
Creation date:
Evaluation Type: Development Boards & Tools
Did you receive all parts the manufacturer stated would be included in the package?: True
What other parts do you consider comparable to this product?: There are numerous CO2 sensors available. They vary tremendously in accuracy and cost. One close example is the Figaro CO2 NDIR Sensor or the many other similar products from CO2sensors.com.
What were the biggest problems encountered?: I was not able to recreate CO2 concentrations levels in the Sensirion platform that I was able to measure with a laboratory research-grade infra red gas analyzer tuned for carbon dioxide and water vapor.
Detailed Review:
Several other reviews have summarized unboxing and certain aspects of operation. I will provide my review on the Sensirion’s Environmental Sensor Shield (ESS) from the perspective of someone who has professionally worked with environmental sensors since the mid-1980s.
The board readily fit into the Arduino Uno that I used. At one point, I removed the board and the pins bent somewhat easily: my fault for being impatient, but a word of caution to the wise. I used CoolTerm to display and save the serial port signal.
Ideally, a board like this would have some flexibility in terms of measurement frequency. In some cases, I need data loggers to monitor sensors only once a day, for something slowly changing like soil moisture. Other things can be measured hourly, such as air temperature over the course of a day or several consecutive days. Still other parameters (e.g. light) can have faster rates of change in the environment and thus require even faster sampling times. The Sensirion ESS is currently set up to display one measurement per second. I contacted Sensirion and spoke to Pooja Bhadrappanavar (Field Application Engineer) who was quite friendly. I was informed that unfortunately, the sampling times cannot be changed on the ESS. I was told that this is because the baseline calibration uses the fixed sampling time. I am not sure this answered my question, but it suggests that the firmware controller for the SGP30 on the ESS is not accessible to the user.
To test the accuracy and sensitivity of the Sensirion ESS to CO2 concentration (aka the partial pressure of CO2, pCO2), temperature, and relative humidity, I mounted the board on a ring-stand in an 0.75 m3 acrylic calibration chamber in my laboratory. 
Next to the ESS (within 5 cm) I placed the sensor of a Li-Cor (Li-Cor Inc., Lincoln, NE) Model LI-7500 Open-Path Infra Red Gas Analyzer (IRGA). This sensor is a highly accurate and rugged device for measuring CO2, water vapor and temperature under all kinds of environmental conditions. I have used this instrument since the late 1990s and am very familiar with its operation. I included a small battery-powered fan to ensure mixing of the gas phase in the chamber. A small plastic tube was used as a port to introduce pulses of air into the chamber. The chamber was allowed to leak CO2 from the pulses of introduced air such that pCO2 would undergo increases and decreases over time on the scale of minutes.
I monitored the outputs of the Sensirion board (on CoolTerm) and the LI-7500 signal (on Li-Cor’s software) during periods of no external air addition, and during multiple pulses of high-CO2 and water vapor concentration. I repeated the pulses six times but show only one example below; the trace for the other pulses are essentially the same. The minimum baseline CO2 concentration signal of the Sensirion was 400 mol mol (= parts per million, ppm). Before introduction of any CO2 pulses, the Sensirion and LI-7500 agreed well and were off by only ca. 1.5 to 2 ppm from one another.
Upon introduction of the CO2 pulse, the values collected by the LI-7500 responded almost instantaneously. It took 43 seconds longer for the Sensirion to begin to indicate an increase in pCO2. The signal of the LI-7500 peaked at 1227 µmol mol-1, however the Sensirion’s signal only recorded a maximum pCO2 at the same time of 456 µmol mol-1. The Sensirion’s signal for pCO2 decayed back to the baseline after 83 seconds, whereas the LI-7500’s signal suggests that the CO2 took about ten minutes to leak out of the calibration chamber.
The air temperature signals were off by about 2.25 degrees. There was much more variation in the LI-7500 compared to the Sensirion signal. The air pulses did not affect the air temperature signals.
The patterns for relative humidity and water vapor concentration (cwv) were similar, but the magnitude of the change in RH was not the same for the ESS as the LI-7500.
The Sensirion might be useful in deployments where pCO2 changes slowly. It does not appear to be practical for situations where measurement frequency is longer than 1 sec-1. Likewise, the Sensirion does not respond rapidly to changes in pCO2 so it may not be suitable for all needs. Because some of the other reviews showed good performance in response to direct breath, I blew directly onto the board. Even under this scenario, the response was much lower than for the LI-7500. Could I have received a board with operational problems? It would be interesting to see if the apparent discrepancy in CO2 concentration detected by the Sensirion in comparison to the Li-Cor LI-7500 is borne out in more realistic settings. Thus, I plan to continue to test this board side-by-side with the LI-7500 in controlled CO2 environment chambers, greenhouses, offices and classrooms to determine whether it acts as a robust sensor of pCO2, temperature and humidity in those settings. This review will be updated based on those results.
Top Comments
Keep in mind that the ESS does not measure CO2 directly, its tuned to measure H2 and assumes a constant H2-CO2 ratio. Its ment to be used to measure CO2 generated by human respiration. There is a paper…
Nice review.
DAB
Indeed, the response time is constrained by the reaction. I am doing some follow-up measurements with both sensors side-by-side in an occupied-room setting that I hope to report next week. Thanks for your…