dear team,
thanks for a promising product so far.
my goal is to determine human presence in the room using co2 as a proxy. This is why i'm interested in the eCO2 from ZMOD 4410. I Understand the gas sensor is not an NDIR CO2 sensor and will not measure actual CO2, yet I'd hope it would be close enough.
I've been running 3.2.0 library and sensor is providing results.
unfortunately the figures do not seem to correlate with human presence in the room. I know that there will be a delay of at least 15-30 min for the sensor to react. sensor is running for more than 48hso this is why I'd need your help. Please find a few hints about the setup so far.
Is there a testing procedure to ensure that sensor is actually ok, It would be very helpful for you to share a test procedure to confirm that readings in specific test conditions are correct: for example controlled conditions can be based on salt and ethanol concentration in containers. This would help understand whether the sensor is working correctly and we can compare results with factory results.
secondly, I understand there is temperature compensation if we provide temperature. Can you help us understand how important temperature compensation is, especially if variation is 'small' in the ambient air , e.g. between 23 and 28 deg. finally, I see there is new firmware 4.2.0. I have still rouble to upgrade, but can you indicate if v4.2.0. will be substiantially better at evaluating e-c02 ? based on release notes, it seems there is no major improvement with v4.2.0.Thanks again for your help as sensor seems very promising.
Hello,
Thanks for reaching out Renesas Engineering Community!
As far I can understand you are mentioning that the eCO2 concentrations do not correspond to the fact that there is human presence in the room or not?
1. Can you please share with us which MCU target you are using to be sure that you use the correct precompiled libraries for the 2nd IAQ Gen operation mode?
2. You mentioned that you left the sensor been working for 48 hours. That's is correct since these sensors are MOX sensors for proper initialization we suggest to leave the sensors constantly been taking measurements for the very first power-on for at least 48 hours. And after this period passes we suggest to start evaluating the sensor measurements.
Please be aware that initially the sensor for some time needs to be placed in an enviroment with "clean air conditions" for some time from the very first power on, so that you will give to the sensor enough time to calculate the baseline resistor value which will be stored for the algorithm and will be used as the reference resistance value for clean air. When we mention clean air we mean a room, which is not crowded with people and there is good air ventilation (e.g. open windows).
Have you done this? Have you left the sensor working for a while in clean air conditions, so that it calculated correctly the baseline resistance value?
3. Your observation about temperature and humidity is correct. The temperature and humidity will have impact on the final readings. As these sensor are MOX sensors and the behavior of the MOX layer is changing depending the temperature and the humidity measured. So for better accuracy for your results it would be good to have an extra temperature plus humidity sensor in order to feed to the algorithm these data as inputs and have finally better accuracy in the results the algorithm outputs.
4. In general we recommend to our customers to use also our Evaluation kits with our sensors. Which have already been stabilized by us, since they have been left working in the factory for 96 hours. And they can be used with a GUI SW provided by us that will give the final measurement results. And also you have the ability to save the measurements in .csv log files to compare with your device measurements.
Here is our EK for ZMOD4410: https://www.renesas.com/en/products/sensor-products/environmental-sensors/zmod4410-evk-indoor-air-quality-sensor-evaluation-kit
And as you can see it has extra HS4001 (temp &hum) sensor on the sensor daughter board:
So you could initially request a sample from our website for this to use it as reference and start comparing your device measurements with those the EK with sensor board provides.
5. I would suggest that it would be good to use the new version of the library since some improvements have been done as far it concerns reducing the noise with lower gas concentrations. And always the latest versions are improved and suggested to be used. So yes I would suggest to go from 3.2.0 to 4.2.0 version.
https://www.renesas.com/en/document/swo/zmod4410-iaq-and-tvoc-firmware-2nd-generation-algorithms-iaq-2nd-gen-recommended-new-designs?r=454426
Please make sure that as soon as you switch to the new library version you leave the sensor been working in the beginning for some time to clean air so that it can calculate correctly the baseline resistor. And also make sure you leave it being working for 48 hours ( and more if possible e.g. 96 hours). And then start evaluating the measurements. In your case where you are switching to new library version to be sure that you are in the "safe side" to start evaluating measurement more than 48 hours being working is better.
In general we do not suggest powering of the sensor, since for as long as the sensor has been left power on and working the more accurate the measurement results will be. And also in case it gets powered off for a while the less typical start up time it needs if it has worked for a long time period.
Even in applications where low power consumption is a requirement, we do not suggest powering-off the sensor.
Now as far it concerns the comparison between our sensor which is a MOX sensor and gives the estimated CO2 values and compared with the NDIR sensors which provide the absolute CO2 value, please let me inform you that we have a report published. Based on tests which run for two months where the performance of our sensor is compared with other NDIR sensors in the market and the final results showed that our sensor can give quite good results as the NDIR sensor. Here is the report if you would like to have a look:
https://www.renesas.com/en/document/rep/ksi-report-evaluation-renesas-eco2-solution-full-test-report?r=454426
Also, I think it would be good to share with us some measurements to check your sensor behavior while the room you are testing is crowded.
Do the RMOX resistance value gets decreased while the room is getting more crowded?
Hope it helps.
Best Regards,
IK
Dear team,
thanks a lot for reverting with valuable information. I believe I have a finding that can explain the strange values, but before rushing to conclusions, let me share the details:
we have been starting evaluation since nearly 2 years and it's been quite difficult to make SW work correctly. We have succeeded with ESP8266, the firmware v3.20 is working correctly.Initial evaluation was done without temperature and humidity compensation. Since the result of the study done in 2020 was done on earlier SW, the SW level alone does not explain the strange values.
We have tried to introduce temperature compensation first. The compensation was calculated, but yet the values were not correlating with human presence. Then we introduced temperature and humidity compensation and we were still getting strange values, i.e. values dot not correlate with human presence. we were seeing that as soon as cooling was introduced, the air was reported better, although air quality has not improved. FYI air cooling means that the same air is cooled but no fresh air is introduced.
We have left the sensor in clean air indeed (window open) to help stabilise the baseline conditions.
about EVK, I have US-800XXX but without the MCU and Windows SW. But since there is ok response from the chip, I am not suspecting the EVK to be issue. I checked on Mouser, the new EVK with both chips is not available yet. I appreciate this option since the MCU enables to run immediately the application on windows without SW development to the target MCU.
Thanks also for sharing the experiment, it is very detailed and it helps understand the study and the conditions.
Since i was still not able to replicate the experiment despite these detailed explanations, I started to compare the method. I have a theory which could explain the difference between study and our experiment.
The theory is the below:
the experiment is conducted in heating scenario: i.e. the building for the experiment is in Germany in october november is heated.
our experiment are conducted in cooling conditions, i.e. the air needs to be cooled using air conditioning. That means a coil where 100% of the air will eventually go through.
I understand that MOX sensor will mostly react to TVOC. I could not see the list of exact TVOC that ZMOD4410 uses, but I understand that ethanol is one (if not the major TVOC). The theory is that heating (building heating machine at temperature of e.g. 40 degree) does not modify substantially the gas composition (ethanol and CO2). But when we are talking about cooling, the situation can drastically different. The cooling element is at temperature of 3-10 degree. Condensation temperature of ethanol is 78 deg and most other TVOC will be in this range. This means that ethanol and TVOC will condense (maybe not fully but at possibly 50% of the ethanol will condense as it passes through the cooling element), based on documented experiments in literature. However co2 which does not condense will remain at exact same composition in air before and after the cold coil.
based on the theory above, this would mean in cooling case: - the CO2 concentration remains the same and air quality remains the same - the cooling process reduces the ethanol and TVOC actual concentration substantially without improving air quality. However because the concentration of ethanol reduces, ZMOD4410 will indicate that TVOC values are less and thus display a significantly better air quality than it actually is.
I have tested with air conditioning turned off. The results are not 100% perfect but much better and much clearer correlation between human presence and air quality.
If this theory is correct, this would still mean a quite important reduction of the ZMOD4410 ability to correctly evaluate CO2 in cooling case. This would also explain why the results have been unreliable so far but correct in other cases.
Are you able to revert on the specific cooling case and confirm the chip mode of operations? if this is true indeed, i'm not sure how the chip could be used on our case. maybe there is a SW evolution to make the chip react to gases that do not condense as easily as ethanol.
Thanks again for your support and hope to read from you soon. you can revert privately if that is easier.
Hello Rasmus,
Apologies for my delayed response, due to the big workload and amount of requests we receive per daily basis.
So as far I can understand you main concern is if ZMOD4410 can detect human presence according to the eCO2 concentration.
First of all let me point out a few thing I would like:
As you already know the ZMOD is a MOX sensor that can detect the present VOC gases in to an indoor place (and then gives the eCO2 value).
So it will be able to always react to all present VOCs in a room (not only ethanol). It might not be able to to have selectivity in specific gases present in a gas mixture as it is a non selective sensor, but it will react to all the VOC gases present. That's why it is called TVOC sensor.
Please check:
https://en-support.renesas.com/knowledgeBase/20684478
https://www.renesas.com/en/document/apn/zmod4410-application-note-tvoc-sensing?r=454426
Also indeed there are some researches that associate the concentration of the VOCs with the temperature and the humidity of a room. As far I can understand and you mentioned you are testing the sensor in rooms that are cooled by air condition and not heated.
Indeed according to some researches that related the VOCs with the room temperature showed up that in rooms with higher temperatures the concentration of the VOCs might be bigger. And especially as far as it concerns the VOC gases that are produced by building materials (carpets etc.)
However I think the crucial think is to check these parameters:
1. Size of room testing sensor
2. How many people entered the room.
3. In some cases also the condition of the people (sleepy, sportive etc).
In any case you are concerned about the fact that ZMOD due to the fact it gives an eCO2 value and not the absolute CO2 value doesn't respond immediately to human presence. Let me also suggest you as alternative our new NDIR CO2 sensor RRH47000.
https://www.renesas.com/en/products/sensor-products/environmental-sensors/intelligent-sensor-modules/rrh47000-single-channel-ndir-co2-module-humidity-and-temperature-sensor#overview