Thread: Developments in Dust Sensor tech

Thanks for that

Having a read through the instructions under “Eval Kit Instr Manual, page 2” I noticed what I believe may be an error. They show the GND in the BOB connected to 5v on the Arduino and not GND?

Not knowing this device I presume it’s an error, as other sites show the GND on the BOB connected to Arduino GND.

Originally Posted by dmorse

I just got an email from Digi-Key showing what I think is a new sensor from Panasonic. What's interesting is that, like the Plantower, is outputs PM1.0, PM2.5, and PM10. Unlike the Plantower it specifies a tolerance for those values! You'll have to download the spec to see that, not the datasheet. Here's the Digi-Key page.

Thanks Dan.
FWIW I will use uncertainty in place of tolerance.

The uncertainty for a particle detector output is amongst other things a complex function of the particle size distribution, so a single tolerance figure on a concentration is likely to be an over estimate of some section or other of the concentration range. All pro level particle counters will output actually particle counts across a particle spectrum and allow the operator to input the dust density before providing a final concentration. They also take into account the problem of missed counts at higher count rates which is a problem for all these counters. The net effect will be a variable tolerance across the range of particle sizes and counts rates. In other words if you want to know measurement uncertainty then you have to be prepared to work in detail with the particle size distribution data. If the sensor does not output these particle bin sizes then it cannot be done.

The Plantower counters output particle count/numbers for particles in the size range of >0.3, >0.5, >1, >2.5, 5 and >10 Microns
To determine PM2.5 uses the first 4 of these counts.

At low dust concentrations the numbers of >0.3 micron particles tends to be in the 1000's so this has an uncertainty of sqrt(1000) or ~33 or 3%
The >1 micron range has counts of around 10 to 100, if it is 100 the sqrt(100) is 10 so this has 10% error, if its only 10 then the sort(10) is about 3 so a 30% uncertainty.
If PM10 is required then often numbers of counts of this particle is in the single digits so very high uncertainties at low concentrations.

Absolute accuracy also depends on some form of calibration but short of sending these detectors back to a calibrating lab or comparing their outputs to an instrument that is regularly calibrated there's not that much that can be done about this.

Because the larger particles have by far the greatest mass (but also the highest count uncertainty) they dominate the final concentration value. Because of this, the uncertainty tends to increase dramatically at low concentrations and is probably why the Panasonic specifies a 10% but only down to 10 µg/m^3 below which the Panasonic does not quote a tolerance.

Some improvement in uncertainty can usually be obtained by collecting more counts or data. The best way to do this is pump more air through the sensor. The particle detector at work pumps about 10X/s more air through the sensor that the smaller detectors. An alternative to this is to count for longer an accumulate more counts. This is why all my particle counters operate with 30s count instead of 1s and one of my detectors has an operator selectable count rate from 1 to 100s.

Count accumulation works really well and is essential for taking backgrounds in still air. As fine wood dust will constantly settle out of air and has a half residence time of about 25 minutes, 30s is short enough to obtain a reasonable reading.

But all of the above can often be downplayed because it's the external variability (in both short and long term) of the dust levels in an environment that needs to assessed and this is how I get my final uncertainties.
Typical dust levels drift all over the place so some way of averaging the counts over some time is preferable. Many dust levels usually have to be assessed against a background (ie the ever changing external shed air) so bracketing is required, ie measure external shed air for 90s, then shed air, then external air again, etc.

The net effect of all this is it will be difficult to get readings that reproduce to within 20% That's OK as long as you don't start making decisions based on differences around the 20% mark as you might thing you are making things better when you could be making them worse.

I've compared the Plantower counters with a couple of calibrated counters from my former employer and found the counts agree within counting statistics except for very high count rates where there appears to be a slight lower counts with the Plantower that are just outside counting stats.

I needed another PM2.5 reader to do some dust level monitoring at my local Men's Shed. The one I'm using in my workshop is semi-installed there and a bit fragile being one of those partially open models.

None of the economy models that BobL recommended were available, so I purchased the following one as I also wanted to get some idea of the level of TVOC exposure I'm getting in my workshop when I'm using my wax finish that contains natural turpentine.

8In1 Digital Air Quality Detector Monitor PM2.5 PM10 HCHO TVOC Temperature Meter | eBay

Enclosed economy PM2.5 reader.jpeg

Specification:

- Detection of PM2.5/PM10
Number of Test Particles: 2.5μg, 10μg
Measuring Range: PM2.5, PM10
Sampling Time: 3 seconds
Detection Range: 0-999μg/m3


- Formaldehyde Detection
Testing Items: Formaldehyde and TVOC (containing benzene)
Detection Range:
0.000~2.888 mg/m3 (formaldehyde)
0.000-9.999mg/m3 (TVOC)
Detection Principle: Semiconductor Gas Sensor Principle
Detection Time: 5-10 minutes
Sampling Mode: diffusion sampling
Unit of Concentration: mg/m3


- Power Supply
Battery Capacity: 2200 mAh Polymer Lithium Battery
Input: 5.0V/500mA
Charging temperature: - 10℃ ~ 45℃

Side by side my two readers give very similar measurements (within 1-2μg/m³ at PM2.5), so perhaps equally poor at giving an accurate reading...

For all I know the two units may have the same economy detectors inside, but this is what you get in this economy range, ie $65 delivered.

Besides the TVOC readings and being of more robust construction, which is better for transportation and wandering around taking measurements at the Men's Shed or elsewhere, it also has a built in rechargeable battery that is more manageable than carting a USB charger around with the other model.

Anyway, at least I'm getting very similar readings with the two units and therefore at least consistently wrong and not left wondering which one is the more accurate...

The new unit is supposed to have a calibration function... as in, take it outside and let it sample 'fresh' air for 5mins. I'm not sure which of its functions it is re-calibrating, but the air in my workshop is often cleaner (when the cyclone is running) than outside, so I'm not sure how valid that feature is for the readings it gives.

When I'm done with using it at our Men's Shed, for which it was primarily purchased, I might offer it on loan to forum members who would like to get a bit of an idea how well the dust extraction is going in their sheds. Of course, offered with all the caveats outlined by Bob throughout this thread that these are economy devices and can't be expected to give the same accurate readings provided by industrial level monitoring devices that cost many multiples (like 10x) more, but are better than just guessing or ignoring what your PM2.5 exposure levels might be.

Originally Posted by NeilS

...just guessing or ignoring what your PM2.5 exposure levels might be.

I recently dug out the following finding to share with the members of my local Men's Shed.

A 2002 study found that "Each 10 μg/m3 elevation in fine particulate air pollution (PM2.5) was associated with approximately a 4%, 6% and 8% increased risk of all-cause, cardiopulmonary, and lung cancer mortality, respectively." Pope, C. Arden; Burnett, Richard T.; Thun, Michael J. (March 6, 2002). "Lung Cancer, Cardiopulmonary Mortality, and Long-term Exposure to Fine Particulate Air Pollution". PMID11879110.

It is hard enough to get the chaps at the Shed to wear PPI for the many very visible and obvious risks they are exposed to there, so it's an uphill battle to get them to expend funds and effort on what they can't see. I'm hoping some PM2.5 readings bouncing around on the little screen might make that risk more real to them.

PM2.5 readings below about 20 µg/m^3 on budget dust meters are generally inaccurate (as much as by ~100%) but those levels are usually not worth worrying about so it doesn't matter all that much.

The important levels to be able to measure are PM2.5 of 50µg/m^3 (long term exposure level to be avoided by seniors, people with health conditions and children) and >100µg/m^3 (long term exposure limit for any worker.

Being able to measure higher levels accurate is only going to matter if exposure time is monitored and "exposure level x exposure time is calculated", this is what dosage meters (eg for radiation dosage monitoring) effectively do but unlikely for anyone on these forums is in that scenario. If you're consistently measuring >500 µg/m^3 you should just get on and do something about it.

Speaking of high readings, yesterday SWMBO was applying a polish to the stair bannisters and filing the house with a nice beeswax and piney aroma. I also happened to glance at the CO2 meter and noticed it was off the scale (6000 ppm, usually its 415 ppm) and then I remembered the meter also measures TVOC so something in that wax was triggering the meter. It took about 5 minutes to go below 1000ppm and then about another 15 minutes to get back down to 415 ppm.

Originally Posted by BobL

Speaking of high readings, yesterday SWMBO was applying a polish to the stair bannisters and filing the house with a nice beeswax and piney aroma. I also happened to glance at the CO2 meter and noticed it was off the scale (6000 ppm, usually its 415 ppm) and then I remembered the meter also measures TVOC so something in that wax was triggering the meter. It took about 5 minutes to go below 1000ppm and then about another 15 minutes to get back down to 415 ppm.

Yes, it is the solvent (pine turpentine) in my beeswax mix that is the problem for me. A pleasant smell while it lingers in the workshop, but from my repeated exposures to it over the years, and some heavy duty exposure to the solvents used with screen printing in my earlier years, I now get an instant headache from it... indicating that it is having an effect on my central nervous system, probably also my brain, as you may have suspected...

I thought it was about time I got some more precise measurements than a headache to gauge my level of exposure so that I can take some mitigation measures to minimise it. The US OSHA have set some exposure limits.

The Occupational Safety and Health Administration (OSHA) has set the legal limit (permissible exposure limit) for turpentine exposure in the workplace as 100 ppm (560 mg/m3) over an 8-hour workday. The same threshold was adopted by the National Institute for Occupational Safety and Health (NIOSH) as the recommended exposure limit (REL). At levels of 800 ppm (4480 mg/m3), turpentine is immediately dangerous to life and health

I'm definitely getting short duration peaks over that upper threshold of 800ppm or 4450mg/m³ (at mol wt of 136g·mol⁻¹, if that is right).

Occasionally polishing the banister, if that is your only exposure, is unlikely to be an issue, but repeated expose is and I definitely need to do something about it other than change my favourite finishing polish.

Originally Posted by NeilS

Occasionally polishing the banister, if that is your only exposure, is unlikely to be an issue, but repeated expose is and I definitely need to do something about it other than change my favourite finishing polish.

SWMBO applies the wax about twice a year (usually when I'm not around) so I'm not going to be to worried about it.

The other interesting fume/dust test I did accidentally yesterday was when I was milling some steel using misted kero and AFT as the lube coolant.
Usually I run a ventilation fan but this time I forgot and after a couple of minutes of milling I noticed a faint blue tinge to the shed air.
I looked at the particle counter about 6m away and noted the PM2.5 was up at around the 50µg/m^3 mark and went to turn on an exhaust fan and even though about not turning it on to see how high it would go but didn't really want to fog the whole shed out with that stuff.
So I turned on the exhaust fan and within 2 minutes the PM2.5 was down to background levels even though I did about 10 minutes more of milling.

This morning SWMBO was doing some baking and I glanced at the particle counter (like the ones in the picture below) that was in the kitchen and the PM2.5 reading was reading between zero and 10 µg/m^3 which is low for when she's backing as it usually goes over 100 µg/m^3

Screen Shot 2021-12-22 at 3.25.43 pm.jpg

On closer inspection I spied a thin film of spiders web over the fan exit on the sensor. I wiped it away and then blew gently into the air inlet and this must have cleared the air pathway as the readings jumped to over 400 µg/m^3 and then settled down to about 120 µg/m^3.

I also compared the readings to those from my other sensors and found it showed the usual bias (1 ~20%) for PM2.5 so I was fairly confident I has cleared the air way.
So just a heads up that low numbers could be misleading.

....and high numbers appear to be bad breath.

Pete

Originally Posted by QC Inspector

....and high numbers appear to be bad breath.

Pete

The high numbers are more likely to be moisture droplets in breath plus backed up dust inside the sensors air pathway.

I know Bob. Just making a little joke. You can remove the post along with this one if you prefer to stay on topic.

Pete

A good mate of mine bought one of these cheap PM2.5 particle counters to monitor dust during his carving activities. While moving the counter around he dropped it from bench height onto a hard concrete floor and it stopped working so he bought another (he can afford it and I asked him if I could look at the broken one. Turned out it was OK - it was the USB power adapter that was malfunctioning. I offered to buy the dropped one from him but he just gave it to me.

So I have been running the 2 counters (mine and the mates) side by side and I have noticed a consistent difference. This morning SWMBO said she was going to do some baking so I thought it would be good to run the detectors side by side during the bake as during baling the PM2.5 counts (with no extraction fan running) typically goes over 100 µg/m^3.

PM25detectorcomp.jpg

I recorded the numbers over the period of the bake and they are shown on the graph below.
PM2.5 counts never got above about 65 µg/m^3 - cake being baked was a "PANATONE" which is low temp bake.

"A" (blue bars on graph) is the mates old detector, B is mine, and its pretty clear that A is showing about 20% (average of 22.5%) higher readings across the hour or so over which readings were made. This is not new to me but just haven't recorded is systematically like this before

BTW, in the photo above you can see the Adetector is straddling a gap (between fridge and freezer) while B is on top of just the freezer. I did think that might make a difference to ir currents etc but I did swap their positions back and forth and it didn't seem to make any difference.

Screen Shot 2022-04-16 at 12.10.00 pm.png

You night notice the temperatures and humidities also differ (Temps on average differ by about 1.6ºC but humidities vary by 17% RH
Maybe dropping the A sensor affected it?
The Temp/Hum sensors used in these particle coulters are very "budget level" sensors and temperature wise are only really good to +/- 1ºC anyway so that is not an issue but the Humidities should be closer.

I compared these T/H readings to a better quality T/H sensor (I'll call this one the standard sensor) which is supposed to be good to 0.2ºC and the Humidity to +/-2%
The standard reads any temp in between A and B
My Mates Temp sensor reads closest (<0.6Cº) to the Standard, while B is within 1ºC - no issues.

In terms of humidity both A and B read too high, A by 10% and B by ~27% !!

Apparently the humidity sensor can be affected by things like the amount of kitchen grease floating around in the air and I ned to look at this closely.

I've done this to show that these detectors aren't that accurate especially at low levels and even taking multiple readings may not give very precise results. The problems come if you were to be making dust extraction decisions based on readings from these detectors. you probably need to be operating at the +/- 10% level, and to be "safer", at the 2+/-20% level,

I threw mine in the bin because the humidity reading never changed.

Originally Posted by Chris Parks

I threw mine in the bin because the humidity reading never changed.

If these cheap TH sensors are exposed to stuff like finishing vapours or cooking oils these sorts of problems can arise. The sensors (DHT11) used on the detectors are very cheap (a few $$) are commonly used in Arduino projects. The DHT22 sensor is a bit more accurate and precise (and a nit more $$) and are supposed to be more rugged although I have not made a side by side comparison about how vapours affect them.

Breathing on the sensor is one way of testing them (they should indicate a change within seconds) while the other way is just to put a clean finger on them, the small amount of water vapour coming of skin and raised temperature should also indicate a change quite quickly