Skip to first unread message
Bob Holmstrom
Jul 10, 2013, 12:09:50 PM7/10/13
to airqua...@googlegroups.com
Several Air Quality Egg (AQE) owners/users have expressed concern that the values for NO2 reported by AQEs do not seem to be in line with expected values. Some of the differences come from the lack of calibration of the sensors and lack of temperature compensation of the data, but there are additional unexplained problems with the data obtained from almost all AQE's. I suggest that most AQE's NO2 values seem to be too high. This is an issue if one intends to use an Egg to get absolute measurements or to compare with measurements of another Egg. These two constraints are likely of interest to many users and it should be possible to rectify but will take some attention.
[Note: Below - Wired refers to single eggs (1) that have the Nanode module and sensor board in the same egg directly connected to the internet. Wireless refers to the paired eggs (2) that are of the base + sensor configuration - note the Nanode in older vs recent wired AQEs seems to have changed a bit - Wired eggs currently report data in floating format (e.g. digits after the decimal point), Wireless eggs report data in integer format.]
I am concerned that AQEs constructed in different ways produce different NO2 values. At this point it is not known what is causing the difference. Below is a link to the output from several eggs in serial number order. They are marked as wired or wireless based on floating or integer output. I am currently showing only those wireless eggs that I know are running version 32 software and have had the NO2 sensors recently deoxidized. It would be very useful to identify more such sensors.
http://www.hsn161.com/AirQualityEgg/NO2day.html Click on 'week' and 'month' to see longer intervals.
It is interesting that a few of the low serial number eggs located here in Portland seem to to working well and giving reasonable NO2 values similar to those predicted by Washington State University's information from their modeling site for the location where the AQE are installed. See > http://lar.wsu.edu/airpact/gmap/ap4.html [go to species animations and select NO2 and zoom in on Portland in the NW USA]
A few other low serial number wired eggs are exhibiting frequently occurring 'spike' behavior. This often happens in synch with sets of Eggs plugged into the same outlet. What is causing this abnormal behavior?
Note: The currently recommended procedure for an AEQ with high NO2 values is to 'deoxidize' it - e.g. put the NO2 sensor in the CO socket and power it up for 90 seconds) This improves things, e.g. reduces the NO2 reported values but not sufficiently, and not for a significant period of time.
The three high serial number wired eggs in Portland look more like wireless eggs that have been deoxidized - values significantly higher than his low serial number eggs and suggested values for Portland. Curve is similar in shape to low serial number eggs, but an order of magnitude higher and not showing oxidation yet.
Results of the 'deoxidation' procedure do not seem to solve all of the issues we are seeing.
Data from my two eggs, before and after deoxidation are shown below in more detail - they started out in the 200's after deoxidation, but have climbed to non-useful values after a few hours. See the link below for images of the graphs, before and after deoxidation. Results for an egg in Quebec are also shown.
http://www.hsn161.com/AirQualityEgg/NO2deox.html
Much to learn!
What more do we need to know about deoxidation?
Why do the low serial number wired eggs give such different and perhaps better readings ?
What other eggs should be added to the page linked above? If you have an AQE that is giving good data (e.g. representative of values appropriate for your location), please let me know so that I can add it to the linked page in this post.
If you have anything to add to this discussion, please join in!
Bob Holmström
Note - Victor Aprea added the the following note during review of the material in this post.
---------
"The only thing I can think of is it's conceivable that a small number of early on Eggs had an incorrect voltage divider resistor installed on the NO2 chain. Correct hardware should have a resistance reading of ~244200 ohms between NO2 S- (the socket pin) and GND with the NO2 sensor removed.
Other than that I don't think there are any hardware differences between wired and wireless eggs. I could imagine a wired egg generates a bit more heat in the enclosure because of the Ethernet controller.
With sensors that don't hold their deoxidization I might speculate they have become 'poisoned' in some way. The manufacturer literature suggests that is possible, but I don't know enough to discuss that topic intelligently. If I were to automate the deoxidization in the next revision of the egg shield I might do it as often as every 24 hours according to some manufacturer application notes.
I want to emphasize that the point of the egg shield in the architecture was largely to insulate the differences in system configurations. The hardware and software surrounding the actual sensors is identical between wired and wireless systems. The only differences (given the same relative software baseline running on the Nanodes) is in the data transport layer I believe."
--------------
[Note: Below - Wired refers to single eggs (1) that have the Nanode module and sensor board in the same egg directly connected to the internet. Wireless refers to the paired eggs (2) that are of the base + sensor configuration - note the Nanode in older vs recent wired AQEs seems to have changed a bit - Wired eggs currently report data in floating format (e.g. digits after the decimal point), Wireless eggs report data in integer format.]
I am concerned that AQEs constructed in different ways produce different NO2 values. At this point it is not known what is causing the difference. Below is a link to the output from several eggs in serial number order. They are marked as wired or wireless based on floating or integer output. I am currently showing only those wireless eggs that I know are running version 32 software and have had the NO2 sensors recently deoxidized. It would be very useful to identify more such sensors.
http://www.hsn161.com/AirQualityEgg/NO2day.html Click on 'week' and 'month' to see longer intervals.
It is interesting that a few of the low serial number eggs located here in Portland seem to to working well and giving reasonable NO2 values similar to those predicted by Washington State University's information from their modeling site for the location where the AQE are installed. See > http://lar.wsu.edu/airpact/gmap/ap4.html [go to species animations and select NO2 and zoom in on Portland in the NW USA]
A few other low serial number wired eggs are exhibiting frequently occurring 'spike' behavior. This often happens in synch with sets of Eggs plugged into the same outlet. What is causing this abnormal behavior?
Note: The currently recommended procedure for an AEQ with high NO2 values is to 'deoxidize' it - e.g. put the NO2 sensor in the CO socket and power it up for 90 seconds) This improves things, e.g. reduces the NO2 reported values but not sufficiently, and not for a significant period of time.
The three high serial number wired eggs in Portland look more like wireless eggs that have been deoxidized - values significantly higher than his low serial number eggs and suggested values for Portland. Curve is similar in shape to low serial number eggs, but an order of magnitude higher and not showing oxidation yet.
Results of the 'deoxidation' procedure do not seem to solve all of the issues we are seeing.
Data from my two eggs, before and after deoxidation are shown below in more detail - they started out in the 200's after deoxidation, but have climbed to non-useful values after a few hours. See the link below for images of the graphs, before and after deoxidation. Results for an egg in Quebec are also shown.
http://www.hsn161.com/AirQualityEgg/NO2deox.html
Much to learn!
What more do we need to know about deoxidation?
Why do the low serial number wired eggs give such different and perhaps better readings ?
What other eggs should be added to the page linked above? If you have an AQE that is giving good data (e.g. representative of values appropriate for your location), please let me know so that I can add it to the linked page in this post.
If you have anything to add to this discussion, please join in!
Bob Holmström
Note - Victor Aprea added the the following note during review of the material in this post.
---------
"The only thing I can think of is it's conceivable that a small number of early on Eggs had an incorrect voltage divider resistor installed on the NO2 chain. Correct hardware should have a resistance reading of ~244200 ohms between NO2 S- (the socket pin) and GND with the NO2 sensor removed.
Other than that I don't think there are any hardware differences between wired and wireless eggs. I could imagine a wired egg generates a bit more heat in the enclosure because of the Ethernet controller.
With sensors that don't hold their deoxidization I might speculate they have become 'poisoned' in some way. The manufacturer literature suggests that is possible, but I don't know enough to discuss that topic intelligently. If I were to automate the deoxidization in the next revision of the egg shield I might do it as often as every 24 hours according to some manufacturer application notes.
I want to emphasize that the point of the egg shield in the architecture was largely to insulate the differences in system configurations. The hardware and software surrounding the actual sensors is identical between wired and wireless systems. The only differences (given the same relative software baseline running on the Nanodes) is in the data transport layer I believe."
--------------
jeanrobe...@yahoo.ca
Jul 10, 2013, 1:27:28 PM7/10/13
to airqua...@googlegroups.com
Very, very interresting post !
Concerning my NO2 sensor, (1232794855 in St-Etienne-des-Gres Quebec) well i just recived a new NO2 sensor that i ordered 2 weeks ago. So, my NO2 sensor is brand new. I think i was the problem ! I have to confess this: when i installed my sensor, a few weeks ago, i put my WEED EATER 6 inches from the sensor, just to see if he react !!!! SHAME ON ME !!!
So, today at noon, (10 th of July), i installed the new sensor. Values look a lot lower (i think normal value should range between 0 and 300 ppb). So lets see and hope for the best.
Jean-Robert
jeanrobe...@yahoo.ca
Jul 10, 2013, 5:43:47 PM7/10/13
to airqua...@googlegroups.com
By the way, Vic told me that it is possible to change a wireless unit to a wired unit ans i am considering this option if it solve the problem. May be the 433 mhz signal influence the data in a way we don't know.
Jean-Robert
jeanrobe...@yahoo.ca
Jul 12, 2013, 8:24:50 AM7/12/13
to airqua...@googlegroups.com
On Wednesday, July 10, 2013 12:09:50 PM UTC-4, Bob Holmstrom wrote:
My Egg sensors will be off line for a few days, time Vic will transform from Wireless to Wired.
My feed number is: 1232794855
Thanks.
Jean-Robert
Adam Laskowitz
Jul 15, 2013, 3:02:01 AM7/15/13
to airqua...@googlegroups.com
When you say "It is a little high" are you referring to the value of 311 ppb or the jump from 18 to 311 ppb is high?
jeanrobe...@yahoo.ca
Jul 16, 2013, 8:26:53 PM7/16/13
to airqua...@googlegroups.com
On Monday, July 15, 2013 3:02:01 AM UTC-4, Adam Laskowitz wrote:
> When you say "It is a little high" are you referring to the value of 311 ppb or the jump from 18 to 311 ppb is high?
> When you say "It is a little high" are you referring to the value of 311 ppb or the jump from 18 to 311 ppb is high?
Hi
In fact sometimes NO2 values goes as high as 20,000 ppb !!
So, what i did last week, i send back my wireless base and remote and they gently accept to modify it to a wired one !
As soon as it will come back here i will see the change.
Thanks
Adam Laskowitz
Jul 18, 2013, 11:31:03 AM7/18/13
to airqua...@googlegroups.com
Remember though Jean, these data are not scaled or calibrated. So for your particular egg, 20,000 ppb may not be high at all.
Bob Holmstrom
Jul 18, 2013, 3:42:06 PM7/18/13
to airqua...@googlegroups.com
I think something more complicated than scale/calibration issues is happening here. As my initial post in this thread states, not all eggs are acting the same. Some wired eggs have been giving stable reasonable values for NO2 for a long period of time without requiring deoxidation or sensor replacement. A few wireless eggs are showing reasonably stable NO2 but at an order of magnitude above expected values. I have been unable to find any wireless eggs that are showing significant periods of reasonable values. [readers - please help me find any 'good' wireless eggs running version 32 software for NO2 data - note there are some version 16 units showing low values of NO2!] Note - Vic states that sensor shields and shield software are identical for wired and wireless units.
Both Jean-Robert and I think we have showed that de-oxided sensors show good values for a while and then revert to the pre de-oxided behavior. Jean-Robert also tested a new sensor and had the same result. Note - during our experiments, no scale, calibration, software or hardware changes were made.
Vic - you did experiments with sensors outside of eggs while preparing the last software update - what values did you see?
NeilH and David Holstius - your discussions during early egg development showed you had some knowledge of these sensors - your thoughts please?
Does anyone have contact information for someone at e2v that might help?
Bob Holmström
Both Jean-Robert and I think we have showed that de-oxided sensors show good values for a while and then revert to the pre de-oxided behavior. Jean-Robert also tested a new sensor and had the same result. Note - during our experiments, no scale, calibration, software or hardware changes were made.
Vic - you did experiments with sensors outside of eggs while preparing the last software update - what values did you see?
NeilH and David Holstius - your discussions during early egg development showed you had some knowledge of these sensors - your thoughts please?
Does anyone have contact information for someone at e2v that might help?
Bob Holmström
David Holstius
Jul 19, 2013, 12:00:27 PM7/19/13
to airqua...@googlegroups.com
Bob,
It would be helpful to see a figure illustrating the data you're talking about, or, absent that, some machine-readable dataset (CSV?). Even if all eggs acted the same (they won't---variability being contributed by, e.g., 5% tolerances in nominal resistor values), all sensors are not going to act the same. That said, it would be really interesting to see how much variability there actually is!
Re: advice from e2v: The belief I hold at the moment is that few, if any, people have a truly solid understanding of how metal oxide (MOx) sensors behave in ambient air. e2v simply doesn't sell them for this purpose, so I would also think they're unlikely to have/dedicate resources to answer questions about the particular application of measuring trace atmospheric gases. But, you could try!
From what I'm aware of at the moment, without information on ozone concentrations, you'll be hard pressed to say whether a change in MOx sensor resistivity implies a change in ozone concentrations or a change in NO2 concentrations. Generally speaking, they have opposite effects (on the same MOX substrate), so to infer anything about NO2, you're going to have to selectively remove ozone. If you don't do it physically, you're going to lose a LOT of precision in your statistically "adjusted" estimates. Which are already pretty imprecise.
Some researcher/commercial teams are doing great stuff with MOx-based ozone sensing. For example, David Williams (University of Auckland) and Geoff Henshaw (Aeroqual Limited). Unfortunately, a lot of sensor research is behind paywalls.
In the BEACON project (http://beacon.berkeley.edu), which I'm working on with Jill Teige (who is more of an expert than I) and other fantastic students, professors, and partners, we're working to collect data from a bunch of MOx sensors over a long period of time (months, maybe years). It's not the primary goal of BEACON, but other, more reliable measurements are collected at the same time. These combined datasets may be useful, but the BEACON team (including me) needs more confidence before publishing them.
David
It would be helpful to see a figure illustrating the data you're talking about, or, absent that, some machine-readable dataset (CSV?). Even if all eggs acted the same (they won't---variability being contributed by, e.g., 5% tolerances in nominal resistor values), all sensors are not going to act the same. That said, it would be really interesting to see how much variability there actually is!
Re: advice from e2v: The belief I hold at the moment is that few, if any, people have a truly solid understanding of how metal oxide (MOx) sensors behave in ambient air. e2v simply doesn't sell them for this purpose, so I would also think they're unlikely to have/dedicate resources to answer questions about the particular application of measuring trace atmospheric gases. But, you could try!
From what I'm aware of at the moment, without information on ozone concentrations, you'll be hard pressed to say whether a change in MOx sensor resistivity implies a change in ozone concentrations or a change in NO2 concentrations. Generally speaking, they have opposite effects (on the same MOX substrate), so to infer anything about NO2, you're going to have to selectively remove ozone. If you don't do it physically, you're going to lose a LOT of precision in your statistically "adjusted" estimates. Which are already pretty imprecise.
Some researcher/commercial teams are doing great stuff with MOx-based ozone sensing. For example, David Williams (University of Auckland) and Geoff Henshaw (Aeroqual Limited). Unfortunately, a lot of sensor research is behind paywalls.
In the BEACON project (http://beacon.berkeley.edu), which I'm working on with Jill Teige (who is more of an expert than I) and other fantastic students, professors, and partners, we're working to collect data from a bunch of MOx sensors over a long period of time (months, maybe years). It's not the primary goal of BEACON, but other, more reliable measurements are collected at the same time. These combined datasets may be useful, but the BEACON team (including me) needs more confidence before publishing them.
David
Bob Holmstrom
Jul 19, 2013, 2:31:15 PM7/19/13
to airqua...@googlegroups.com
David,
Thanks for responding!
http://www.hsn161.com/AirQualityEgg/NO2day.html shows NO2 data for several eggs I have been tracking - most here in Portland where there will be quite a few (25+) eggs in one residential neighborhood that borders an 'industrial sanctuary' and a freeway.
http://www.hsn161.com/AirQualityEgg/NO2deox.html shows the results when three eggs were recently 'deoxidized' - all three quickly returned to previous behavior.
I have added data from Adam Laskowitz's site to the bottom of the 'day' page of the first link above.
I will explore your other suggestions.
Bob
David Holstius
Jul 19, 2013, 3:19:08 PM7/19/13
to airqua...@googlegroups.com
Bob, no problem. Can you or Adam point to code/API for extracting that data from Xively? I have a Cosm library for R, but not for Xively. Thanks in advance -- David
> --
> You received this message because you are subscribed to a topic in the Google Groups "#AirQualityEgg" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/topic/airqualityegg/_S4GRGRVY-Y/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to airqualityeg...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
> You received this message because you are subscribed to a topic in the Google Groups "#AirQualityEgg" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/topic/airqualityegg/_S4GRGRVY-Y/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to airqualityeg...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
Bob Holmstrom
Jul 23, 2013, 9:25:46 PM7/23/13
to airqua...@googlegroups.com
David,
Sorry, I missed your code request on the 19th. The code for my NO2 graphs uses some code that was posted to this list a few weeks ago. Mine can be seen using 'view source'. Each graph uses a statement like:
<img src="https://api.xively.com/v2/feeds/81686/datastreams/NO2_00-04-a3-27-c3-21_0.png?width=1000&height=300&c=F53527&duration=1day&title=Nitrogen Dioxide ppb 81686 Portland - wired&show_axis_labels=true&detailed_grid=true&scale=auto&timezone=Tijuana">
<br>
Where feed number and channel number for each sensor is as it appropriate.
The last graph from Adam Laskowitz's site is a screen shot - you will have to ask adam about his code.
Bob
Message has been deleted
dasnervt gewaltig
Jul 27, 2013, 2:51:26 AM7/27/13
to airqua...@googlegroups.com
There seem to be several issues going on with no2 readings. I taught my egg reasonable no2 readings by covering 3/4s of the fan hole with gaffa tape. The no2 sensor does not seem to like air flow.
But since some eggs seem to be operated with no fan at all, there must be more to it.
Taping the fan hole did not seem do affect the other sensor readings at all.
But since some eggs seem to be operated with no fan at all, there must be more to it.
Taping the fan hole did not seem do affect the other sensor readings at all.
Reply all
Reply to author
Forward
0 new messages