Nice.
Be good to show the safe and not so safe levels on that chart though.
Looking at the AQI, the PM2.5 is the one to watch.
Anything above 35.4 ÎŒg/mÂł averaged over 24h is considered unhealthy for sensitive groups and above 55.5 ÎŒg/mÂł over 24 hours is considered unhealthy for normal groups with anything above 250 ÎŒg/mÂł is considered hazardous.
I have now created a chart with the PM2.5 readings averaged over 24 hours. As I don't have much data yet, this is what I have so far.
I hope the level comes down, my home is quite hazardous at the moment!
Hmm, at first sight - that would seem to be a significant problem.
Do I assume that you were aware of an issue? Is that why you put the sensor together in the first place? That is certainly the reason I starting making temperature/humidity sensor for our house.
I guess I would want to first verify the accuracy of the sensor and then double check that I wasn't mis-counting. Then I'd want to know what was causing that level of particulates - an fire or solid-fuel stove for example?
Or perhaps the air quality outside your home is very poor?
You've got me wanting to make a sensor now! Did you say which sensor you were using? If not, could you let me know?
I agree, it doesn't look good.
I got the sensor just to add to my weather/environment monitoring, purely as 'yet another sensor'.
It is only through testing and setting it up that this has come to light and only since you asked about the safe levels or I would be none the wiser.
I live in Lincolnshire, so not too far from you, but in the countryside. The air quality is pretty good here so outside I would expect very low levels, 1 or 2 for the PM2.5, it was only when I noticed during cooking that the levels shot up.
I have the sensor on my desk in my 'computer room' (man cave) which is just a spare bedroom, so if the sensor is correct, then yes, it could be quite alarming to see these figures although official measurements are taken as an average over 24hours.
I would hate to think what it would register if I put the sensor in the kitchen!
Oh, the sensor is a PMS5003 from Plantower, although Chinese, they are the original makers and looking around the internet at so called professional sites, this does hold up quite respectfully in comparison with units costing ÂŁ1000's yet it cost me ÂŁ10 including delivery and surprisingly only took about 1 week to come.
Here is a link...
They are tiny too, about the size of a match box (the small type not the Swan Vesta type).
Haha, there was me thinking that you lived in China or some other densely populated area! 
OK, rabbit hole time! ....
Looks like it has a 10% consistency level for normal temperatures so you need to factor that into your numbers. If it is 10% per measurement and you are doing lots of measurements, it is hard to predict the final result accuracy. Not sure how you are calculating your 24hr level but I would suggest that an average of all the measurements over 24hrs multiplied by the number of measurements in 24hrs?
Did you also note the requirement for discarding the first 30sec of data after wake-up? Is the unit permanently on?
Looking again at your details though, the example table you gave shows 17-20 ”g/m3 if I read it right? The US EPA says that the short-term standard (24-hour or daily average) is 35 micrograms per cubic meter of air (”g/m3) and the long-term standard (annual average) is 12 ”g/m3 so if those figures are representative, it is a little high for long term exposure.
Do you use candles or joss-sticks at all? That could possibly cause elevated PM2.5. Other possibilities, if I understand the science correctly, living next to a busy road or high usage levels of organic molecules from cleaning products(?)
I'm struggling to convert the ppm figures you have in your chart to the ”g/m3 which seems to be the standard measure.
This document is interesting and indicates that measuring PM2.5 isn't easy even professionally. The standard for UK measurement includes a ±25% error rate. https://uk-air.defra.gov.uk/assets/documents/reports/cat11/1212141150_AQEG_Fine_Particulate_Matter_in_the_UK.pdf
That being the case, your figures could easily be within normal ranges I think?
One last point then I must get back to house cleaning before Mrs K returns and tells me off 
Here is a comment from this article. "Optical particle counters do not measure mass as in standard PM measurements but make an estimate based on light scattered by particles, in particular they respond to water particles that are not measured in PM measurements" - that being the case, are you also measuring temperature and humidity? If you are seeing relatively high humidity levels, this might also impact the readings, especially when humidity is close to the dewpoint (though that is unlikely this time of year). Still, humidity levels over 70% can also cause health issues so are worth checking. The article also uses a BME280 to measure humidity and has some example software too I think - it uses an ESP8266 so that you can position the sensor anywhere and get the values over WiFi which would also be my preferred approach.
This is interesting 
https://blissair.com/reduce-negative-effects-air-pollution.htm
Sorry about the rabbit hole and brain dump. You have tweaked my interest on a fairly quiet Sunday afternoon.
1 Like
Actually, although the chart shows ppm it is not, it was just there as something to put in, but I found the ug/m3 now, so it has been changed. So the values on my chart are actually ”g/m3.
For the averaging, I have left that up to Grafana and hope that setting the mean for 24h it will automatically do the calculation of adding them and dividing by number of readings.
In which case it should not make a difference of how many or how often the readings are taken as long as period between readings is constant.
eg. If I take 1000 readings 500 of them at 10ug and 500 at 50ug it is no different to 250 at 10ug and 250 at 50ug the average is the same.
I am currently taking readings every 10 seconds and it is running permanently.
Please remember, this is only testing and messing about at this stage. My intention is to mount it in a small Stevenson Screen type box outside.
I think all the documents you have linked to or posted I have been there and read them. At first it was just another sensor to play with, but your question about the safe levels has given me a boost to 'investigate' further.
I also pleased it has peaked your interest too, its not often in my life that something I have done or doing has peaked an interest for someone else!
This is my professional installation...
This is one of the first sites I looked at, gave me the idea of making my own cheaply.
Ah, OK.
OK, sounds like that should be OK.
Of course, but it is a fascinating subject 
I think that I would definitely reduce that. It is a lot of data and doesn't really give you more accuracy. Personally, I would take readings every 1-5 minutes at most. Maybe even every 10 minutes. If nothing else, it would be interesting to see if that changes the results at all.
That's what's great about forums like this, many like-minded geeks getting into sensing and monitoring using cheap components and open source. A great hobby even if Mrs K doesn't appreciate it! (She has her dress-making instead which is apparently more acceptable - haha).
I have added this to my list of things to make, sadly not going to be doing it this side of summer I don't think.
One more thought - it is quite possible that the location of the sensor is giving you bad readings. If your desk is anything like mine, it is probably pretty dusty. The article you linked to suggests that the sensor should be at least 1.5m off the ground.
Knowing far too much about house air-flows - after having a number of condensation and damp issues over the years in our old Victorian house - it is clear that positioning will likely make quite a difference. That and checking for humidity. The BME280's are nice or the BMP280 if you also want to measure atmospheric pressure (I think that's the right way round, I always get them mixed up).
Yep, I have quite a few sensors that I play around with.
Got both BMP280 and BME280 both do pressure and temperature but the latter also does humidity.
I also have the DHT11 & DHT22, HTU21D, HTU21 and SI7021 which all do both temperature and humidity with about another 5 X DS18B20's for temperature.
Several light sensors too, the MAX44009, TSL2561 and BH1750 for visible and VEML6070 for UV.
I just play around with them, some I test against each other to see the results. Things like the ambient light sensors all gave massively different results and also the humidity sensors never seem to match up either.
On the particle side, the values I get are normally around 1 - 5 ug/m3, it's only when I (or my wife) has been cooking / frying that the levels rocket up. I suppose it proves that its working. With my windows open, it sits around 1ug/m3.
I will probably look for a box design to put it in, similar to the one from 'GAIA' and get my 3D printer earning its keep. The box they have looks quite large in the photo but I think its only a few inches long with a grill on the front.
On GAIA the advertise this
The Gaia Ax series is a semi-professional Air Quality monitoring product
Gaia A11: Personal Sensing Station
The Gaia A11 is the most simple version of the series, but also the most affordable. It comes with 2 integrated sensors, one for Particulate Matter (PM2.5, PM10) as well as one for Relative Humidity and Temperature.
Particulate Matter Sensors: 1x PMS 5003
Meteorologic Sensors: DS18B20
Power Supply: 5V (USB compatible)
Connectivity: WIFI (with embedded antenna)
Dimensions: 65 * 95 * 55 mm
Not sure how they get Relative Humidity from the DS18B20 as its just a temperature sensor so probably an error as the A12 uses the BME280.
Anywayl, break time, off to play Battlefield 4 for a while.
Most of them are rubbish as you've seen. The DHT's in particular. However, the newer digital sensors are generally pretty good. The BME/BMP-280's and the HTU21D should give very similar results with minimal drift.
I've only used the BH1750 which seems pretty good. I have one on the sensor platform in our rear hallway that can get quite dark. It switches on the light there if it falls below a certain level.
Ah, well that is fine from my reading.
Never have managed to convince myself to spend the money on a 3d-printer. Probably as well, I don't think I've anywhere reasonable to put one at the moment.
I tend to collect old boxes of various types. Some cosmetics come in useful plastic boxes. I also collect smallish tins which can also be useful though they tend to rust if kept in damp conditions of course. Plenty of cheap project boxes from China on my bookmarks in case I need them. I just need to find time and energy to move my various sensor platforms into boxes 
I may get round to it one day.
Anyway, thanks for the discussion. Interesting.
When I originally read this thread, I was disappointed to read about WeatherUnderground disappearing and to read later that it had ceased for personal use.
I never bothered to check as just took it as read, but tonight I couldn't sleep so fired up the computer to do a little reading and happened to hit my link to WU.
I logged in as I used to and everything is there, complete with my API for uploading data.
According to my settings I have a 'paid' membership until March 2020. (I have not paid for any service with them, so I assume the free account just says paid anyway).
So, I'm now looking to update my station to send data to WU, I already have a station set up, but never got around to uploading anything because everything had to be in those strange old-world figures (imperial), which would mean yet another conversion and I didn't have Node-Red back then.
An update on my particle monitor, I made a small ventilated box for it and placed it outside a few metres from the house. Itâs been running about 3 weeks now and data looks good.
I also placed a BME280 (temp, humid, press) sensor in with it, but it is faulty, the temperature indication is higher than it should be and doesn't really drop much. Its not a ventilation problem as I have also mounted a separate temperature sensor inside and that reads fine.
I have a new BME280 on order, should be here anytime.
I also got an old Davis Instruments anemometer from a friend, the reed switch was broken and required replacement. Unfortunately, the unit turned out to be a very early model that could not be opened up as described in the many articles on the internet. I had to saw the unit in half to gain access to the wiring and dig out the old reed switch which was epoxied into position. I replaced it with a Hall Effect transistor which was a pain to get the orientation correct as the magnet in the cup part is a vertical (not horizontal) rod which meant the orientation of the magnetic field was not well suited, but I managed it.
Itâs now epoxied back together with a nice 12m lead, so I will be testing it and then adding the code to the ESP8266 to include that in my station.
My next problem if anyone has any ideas is how to combine all the data in Node-Red to a single string. Now the particle data is sent in a different message to the other sensor data because of the MQTT message limit.
In the âInfluxDBâ database, although all the data is recorded, the particle data is on a separate line to the other data, not a problem there, but I want to export the data to other platforms such as WU or Thingspeak and I need it all in 1 single string.
So, I would like to take data from two or more separate inputs at different times, hold them until all sets are received then put them together into a single string before passing them to âInfluxDBâ and the other platforms. Any guidance would be appreciated.
Anyway, sorry for the long story again.
This is my dashboard so far... https://snapshot.raintank.io/dashboard/snapshot/xgjf6GFoFIwCmIuAnUcxp9JbSxUQ0wAW
