RoadTest: Amphenol Air Quality Engineering Dev Kit - Industrial Sensing
Author: ngblume
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?: Similar CO2 sensors with direct gas "feed" and other optical CO2 sensors
What were the biggest problems encountered?: Memory / variable space limitations on the Arduino clone
Detailed Review:
Hey everyone,
first of all, let me apologize for being late with my review of the Amphenol Air Quality Engineering Dev Kit.
I was using the product quite extensively and wanted to compare it to a research CO2 laser spectrometer, which wasn't available for use earlier (details later on...).
I still hope, you find the review quite helpful and if there are any questions, please let me know in the comments.
Enjoy !!
Cheers
Niels Göran
You will find the source code and a PDF version of the review at the bottom of the review.
If you have problems with pictures not showing up, please use the links to the large version of the picture instead or look through the PDF version (at the end of the review).
REVIEW – Amphenol Air Quality Dev Kit
TOC
1. Introduction
Before getting into the review, I want to give you a little background. I’m currently a PhD candidate at Technische Universität Darmstadt in Germany. My focus is advanced laser diagnostics in combustion research, mainly absorption spectroscopy. We are currently in the process of developing a TDLAS-based spectrometer system for exhaust gas analysis. TDLAS in this case stands for “Tunable Diode Laser Absorption Spectroscopy” and refers to a diagnostic approach, which is based on absorption of light by molecules. Instead of using broadband light source, such as halogen lamps or something similar, TDLAS uses narrow-band tunable laser diodes. They are almost identical to the ones used in telecommunication (same wavelength as well). I will go into some details in sections 5 and 6. During this development we looked into alternative systems and their cost and use cases. One very commonly used diagnostic approach is NDIR (non-dispersive infrared). This is the approach used in the CO2 sensor in the Amphenol kit. An NDIR system uses a light source as well, but a broadband source instead. On the detection side, there is a photodiode in combination with a filter, which is specifically tuned to CO2 absorption band(s) in the mid-infrared (MIR). Therefore there is a correlation between the transmitted intensity and the CO2 concentration (the higher the transmission compared to a reference value, the lower the CO2 concentration, since higher CO2 concentrations absorb more light. The behavior of the filter and diode combination can be calculated based on simulated spectra of the gas in question (i.e. CO2) and calibrated quite easily when the complete system is assembled.
For this road test, I planed on using the sensor kit as follows:
1. See, how it easy it is to get going
2. Test various scenarios with the delivered software for the Arduino-clone
3. Modify the software as necessary
4. Use the sensor in my research somehow (wasn’t quite sure at the time, how …)
5. Compare the performance of the sensor to existing systems at our facility, mainly laser based systems, while using a controlled environment (climate chamber)
In general, I was very happy with the performance of the sensor(s). I mostly looked at the CO2 sensor, since the particle sensor wasn’t of any direct use to me (more or less only looking at the value and being fascinated as to when the values goes up or down). For comparison of humidity and temperature measurements; I used a testo 435-2 with a humidity / temperature probe for comparison (https://www.testo.com/en/testo-435-2/p/0563-4352). There were some minor issues that I would like Amphenol to address to make the kit even better. From the standpoint of a researcher / metrologist, it is not really satisfying to be faced with automatic calibration features (that is not consistently documented), but for the general use, this feature proves quite useful.
Let’s get started !
2. Unboxing
The kit was delivered quite fast and properly protected in a box. I can confirm some of the feedback from other reviewers that the display, already mounted to the Arduino-clone in my case, fits only rather closely into the packaging. Given that there is some space left in the package, I suggest that the display is packaged separately in the kit box.
I was pleasantly surprised to see all the necessary things included in the kit, including USB cable. Due to the fact, that the Arduino comes pre-programmed, getting started is a rather quick and easy process (see following chapter). Included in the package was also a “Quick Start Guide”, which was also available beforehand as a download (in combination with a datasheet and some application notes).
All included sensors look quite sturdy and of solid built quality. I’m not fully convinced that the same is true for the Arduino clone, but, in my opinion, the most fragile component is the “AAS-AQS-UNO sensor evaluation shield”. This shield is placed on top of the Arduino and provides interfaces for the different sensors. Due to the nature of some of the sensors (i.e. particle sensor), the mounting interface are rather awkwardly placed (dust sensors being mounted vertically). It is a nice idea to have all the sensors directly mounted on a single board, but especially when testing different mounting orientations this makes it rather difficult to change the orientation of only a single sensor (easily solvable by additional cables). Nevertheless, this makes handling the sensor just a little bit more complicated. It might be worth for Amphenol to invest in a slightly different mounting system.
3. Getting started with the kit
As already mentioned before, getting started is as easy as following the quick start manual, which is quite straight forward:
And, as mentioned by another review, it actually works !! I’m amazed by this as well, since this is rarely the case with any product demo.
As per quick start guide, the preprogrammed source code is available via Github (https://github.com/AmphenolAdvancedSensors/Telaire), which is very nice and rather “modern”. Providing open-source code and versioned software is not what sensor companies usually do. Big thanks to Amphenol for this!! This is, what gets developers going really fast!
In the picture above you can see the system in action after the source code modifications. The CO2 sensor is currently taking a reading, which is visible by the activate illumination of the detector.
4. First impressions and first modifications
I’m not really a fan of showing measurement values after each other, since I want to see all values at the same time, especially when they are captured simultaneously. Upon looking into the example code, I saw that there is an option for also publishing the data via serial console in a more suited manner for simultaneous measurements. And that’s when the first problem arose: The provided Arduino lacks dynamic memory for global variables! I’m not sure, if this is due to it being a clone or just the way the Arduino is designed. My aim was to modify the software to show all variables on the display at once and simultaneously provide the values for further processing via serial console. I’m not an expert with Arduinos, but I was able to make it work by significantly modifying the original source code. You find my source code attached to this review. It appears that the rather large display (128 x 64 pixels) uses a very large amount of the dynamic memory. This in combination with the way the example code forwards recorded values to the main function caused a large chunk of dynamic memory used. Due to this, the Arduino IDE issues the warning “Low memory available, stability problems may occur.”, when compiling the source code. In the end, getting it to work, involved some restructuring of the code and a lot of trial and error, seeing how much dynamic memory can be used, before the program becomes unstable. An unstable Arduino has problems executing program code reliably, causing for example random restarts or a program repeating a certain part of the program continuously. Since I planned on using the serial console mainly, while the display being a quick indicator, if the system works, properly, I would have gone with no display in the end, if both at the same time wouldn’t have been possible. I think this issue is rather minor, since most system using the Amphenol Telaire sensor would not necessarily use them with Arduino or at least not with a display directly connected to the Arduino but rather forward the measurements to a PLC or something similar.
SOURCE-CODE-ARDUINO - see at the end of the review
Modifications:
ATTENTION: Adafruit library needs to be set to be expecting a 128x64 px display. Modify the “Adafruit_SSD1306.h” in “C:\Users\UserName\Documents\Arduino\libraries\Adafruit_SSD1306” by making sure, that only the line “#define SSD1306_128_64 ” is NOT commented out. All other options regarding pixel number and size need to be commented out. Path might need to be changed according to your installation.
Since we are using LabVIEW heavily in our lab, I also attach a source code project for LabVIEW 2017 (other versions on request), that records the values from the serial console in TDMS files and shows the necessary values and plots in the user interface.
SOURCE-CODE-LABVIEW - see at the end of the review
5. Using the sensor kit in research
The first possibility to use the sensor in research occurred when we were doing comparison measurements between two CO2 laser spectrometer and needed a sensor for the ambient OC2 concentration. Due to its diffusion sampling, the sensor is aimed at slow changes in concentration with a specified response time of less than 3 minutes for a 90% step increase. The concentration in a gas cell was measured with TDLAS at a rate of 125 Hz. Simultaneously a second TDLAS system with added WMS (Wavelength Modulation Spectroscopy) was taking readings as well. WMS is a modification of TDLAS aimed at better signal to noise ration and therefore more reliable measurements at very low concentrations. This is achieved by adding a high frequency modulation to the laser, which in turns moves the actual information in the laser signal to higher frequency further away from most noise sources (which operate at lower frequencies). After applying a filter, this information is then extracted.
In the experiment a gas cell was filled at the beginning with a rather high OC2 concentration. A vacuum pump was used to pump gas from the cell, while a needle valve on the input tubing of the gas cell limited the speed at which the air in the gas cell was replaced with ambient air. The following picture shows the setup. The blue tubing is connected to ambient air, while the stainless steel tubing at the bottom of the cell is connected to the vacuum pump. Within the blue tubing the needle valve and one pressure transducer can be seen.
This procedure ensured a constant pressure (as opposed to evacuating the cell) and a steadily decreasing CO2 concentration down to ambient level (which was measured by the Amphenol sensor). The aim of the experiment was to test if both TDLAS system would be able to resolve the ambient CO2 concentration. The following plot shows the results. It can be clearly seen, that only the WMS is capable of reliably measuring ambient CO2 concentrations. The standard TDLAS system is disturbed by the noise on the signal and measures significantly lower, therefore non-physical, concentrations of CO2. Shown along the graph are snapshots of the different systems. Only the WMS signal provides reliable information regarding the CO2 concentration. The ambient level, measured with the Amphenol sensor is shown as the grey box (range including specified uncertainty of the sensor).
This is a perfect case for using the Amphenol sensor in research, since is diffusion-sampling limits the resolution of steep gradients or direct exposure to rather hazardous environments (i.e. high temperatures).
Despite the aim of the sensor being air quality monitoring, I wanted to see how far one can push the sensor. Therefore the next experiment involved steep CO2 gradients and a fast CO2 reference measurement system for comparison.
6. Comparison with TDLAS in climate chamber
To provide a rather challenging environment fo the CO2 sensor, I choose our climate chamber, which allows me to create environments with stabilized temperatures and proper mixing of the gases inside (fan at the back of the chamber). Due to its rather thick insulation, the system reaches almost steady-state conditions. By injecting amounts of CO2 into this closed environment, the response to steps in the CO2 concentration can be tested. The injected amount of CO2 is mixed by the constantly running fan. The reference system is an in-situ fiber coupled system directly placed next to the Amphenol sensor (see following pictures). The shape of the measurement volume of the reference system if tube-like and the Amphenol sensor is placed directed at the end of the tube. The air flow through this tube-like structure therefore reaches the Amphenol sensor at the same time.
The climate chamber is roughly 75 cm in diameter (~30 in.). The fan can be seen in the background (not yet turned on). Behind the Amphenol sensor the optics of the TDLAS system are placed (see following picture for details).
In this picture, you can see the Amphenol sensor package directly positioned next to the TDLAS system. The OC2 sensor is currently taking a measurement (illumination visible). At the top of the picture, you can see both temperature- & humidity probes (Amphenol probe on top with white cover) and testo probe right below. Right next to both is the thermocouple.
During the experiments temperatures up to 317 K (44 °C, 111 °F) with humidity levels up to 53 %rH were reached. The following plot shows the curves for temperature (measured with a thermocouple for reference), temperature measured with the Amphenol sensor and for humidity (Amphenol as well). Furthermore the water vapor content shown is measured with the TDLAS system. The values are consistent, since the humidity level is dropping with rising temperature until a water vapor source in form of a cup with hot water is added to the climate chamber. With the added water vapor (constantly releasing more water vapor) the humidity levels rise with increasing temperature. Each drop in temperature is caused by opening the door of the climate chamber for a short amount of time. It is clearly visible that the thermocouple follows rather immediately, while the Amphenol temp- &humidity probe needs a certain amount to start following the change in temp and therefore humidity. This caused by the cover over the probe inlet (see following image), which creates a micro-climate inside the probe.
When comparing the CO2 concentrations measured by the TDLAS system and the Amphenol sensor, the corresponding curves are shown in the following graph. As expected the Amphenol sensor follows significantly slower than the TDLAS system. When comparing the Amphenol sensor to its specifications it appears that the sensor reacts faster than 3 minutes. Furthermore there seems to be an offset between both systems which is correlated with the concentration itself (comparable to a scaling factor). As mentioned before I’m not fully understanding the automatic calibration and I’m not certain, if the system was running constantly for the required time frames as specified for the automatic calibration.
When looking more closely at the graph (see following plot as a zoom of the previous), it appears to be the case, that the Amphenol sensor also performs quite well outside its specification regarding maximum CO2 concentration. The sensor is specified up to 5000 ppm CO2, but performs quite well up to at least 12500 ppm. Exceeding the specification might also be part of the reason for the offset between both systems. There is good chance, that after a full recalibration cycle as specified by Amphenol, the values are in better agreement, especially below 5000 ppm CO2. As mentioned before, the CO2 sensor from Amphenol uses NDIR technology. Therefore higher CO2 concentrations reduce the recorded light intensity. This might be also part of the reason, since the signal-noise-ratio decreases with increasing CO2 concentration (and therefore the chance for errors). TDLAS systems on the other hand encounter increasing SNR with increasing CO2 concentrations.
7. Conclusion
When comparing the CO2 sensor from Amphenol sensor with our TDLAS reference system, the results look quite promising. The sensor is not really suited for our intended purpose (exhaust gas analysis) due to the high temperature and the response time. Nevertheless, that is something that can be overcome, since I’m convinced that the hardware itself, when separating the electronics from the gas itself, can withstand the temperatures. One option might be a separate measurement cell that utilizes fiber coupling to and from the cell. This might require a different filter (MIR wavelength don’t travel so well through fibers), but is still a viable option. Furthermore, such a system also allows for forced flow through the cell instead of diffusion-sampling, which is a requirement for exhaust gas analysis.
Overall, given the price point of the Amphenol sensor, this is a great package to get started with gas diagnostics and is already more than enough for several applications (ventilation etc.).
Top Comments
Dear Niels,
You wrote an outstanding review. Congratulations to You and good luck for Your work with the Analitical Device Development and writing Your PhD. Thesis. You could even include this review in…
Thank you for a good review of our product lines.
Comparison with high end metrology is always interesting, I'm pleased we faired well.
We do offer CO2 modules without auto-calibration, however as this…
Hello DAB,
thanks !
I checked on two computers and it appears to be the case that it shows up on some, but not on others.
There should be a link right where the picture should be which gives you a large version…