Thread: Developments in Dust Sensor tech

The photo below shows a telescopic mast I made to be able to safely suspend a recording dust sensor module from various points in the mens shed.

The mast consists of an 600 mm long Al rod that slides into a 600 mm Al tube and a SS mesh cage that holds the Dust sensor and the backup battery.
The highest position above the shed floor is now 2.1m but it can be lowered down to 1.6m for testing purposes (meaning I have to be there to watch it does not get knocked around).


DustS2.jpg


Below shows that the new position is much closer to the middle of the shed as compared t its old position which as over again the far wall at the position marked by X, which was a fixed 2.4m above the shed floor.

The new position is still not idea but its much closer to the action than the previous position.

The new position takes above an area in the shed which has a TS, Belt sander and a drum sander.

DustSensor1.jpg

The large amount of dust on the ducting is mostly from before we had any vacuum switches on any machinery and members failing to turn on the DC or open gates.
There is also a fair bit of dust generated using routers, portable sander etc
There are still 3 main machines without vacuum switches and it's interesting to see their ducting has more dust on top of it than ducting to other machinery.

Cheap dust sensor technology - with careful use would be good for home workshops

Comprehensive display info
Screen Shot 2018-06-16 at 2.13.42 pm.jpg

https://www.ebay.com.au/itm/Househol...kAAOSwevdZ7t4C

Uses the same sensor as I have used in my particle detectors.
Looks like you have to provide your own power supply - works on micro USB charger or 5V battery packs

USE WITH CARE
Do not shove direct under any sawdust stream or you will kill it.
DO not locate too high or too low in the shed
Take regular readings outside shed or in a low dust environment for comparison or there could be a bushfire around.

Looks like a nice implementation.

Bob would you believe that I was looking at that same item about an hour ago during a search for some CNC gear. I thinks that's about 1/2 the price of similar design I looked at a year ago.

I've ordered one and will see what it can do.

It doesn't look like there is a memory or output for plotting the readings over time? Still an inexpensive (I have wine that costs more) indicator for the shop.

Originally Posted by QC Inspector

It doesn't look like there is a memory or output for plotting the readings over time? Still an inexpensive (I have wine that costs more) indicator for the shop.

Correct no data storage.

One important thing to check is how well it deals with the 2.5 to 10 micron data.

These small sensors are really <2.5 micron dust indicators. They pump so air little volume through them that the number of >2.5 micron dust is sort of "estimated" as much as measured. To minimise this the >2.5 micron data needs to collected and averaged over a number of readings otherwise they will not deliver very reliable readings. In my detector I average at least 10 readings before outputting a number.

Well the new Dust sensor turned up today (BTW no GST).
It has no name but I will call it the "Household" detector since it is being sold as a household PM2.5 unit.

The first thing you notice about the Household is how small it is. (85 x 65 x 45 mm).
The acrylic box is well made but open on all 4 sides so the dust can get to the sensors.
The small size means it will be easy to lose or tread on, but it does make it sort of wearable. It would be easily possible to bundle it with a slimline USB battery pack and hang it around your neck although seeing the screen would be tricky while you are doing something.
Otherwise the screen is reasonably bright and easy to see in a regular shed when you are close to it, but would be difficult to see outside or more than about 3m away inside a shed.

Here you can see the blue sensor inside the Household detector box and one on top is the same as the sensors I use in my detectors.
Comp1.jpg

The temperature and humidity sensor is a HDT11 which is one of the the cheapest HT sensors around and is good for only +/- 2º but there does not appear to be any correction of dust data based on HT so it's not all that important.

The dust sensor used in the Household is the Plantower 5003 which is a 5th generation sensor whereas the latest sensor from Plantower is the 7003, a 7th generation sensor and given the 7003 costs about the same amount and is supposed to be more accurate I cannot understand why they did not use a 7003.

IMG_3197p.jpg

The Photo below shows one of two display screens - the other is shown further below.
Lots of potentially useful info.
Notice the AQI - Air quality index - this is based on the PM2.5 concentration data.
Use teh AQI with caution, there could a situation in a shed where the AQI is OK but you are being exposed to lots of dust >2.5microns.
DIsp1.jpg
Apart from size there's not a lot of difference between the 5003 and 7003 sensors - they are both marketed as PM2.5 sensors and a quick comparison shows that the 5003 reports reasonably consistent (in particle counter terms) with the PM2.5 concentrations reported by the 7003s I had running along side it. If anything the 5003 tended to read the PM2.5 low by between zero and 20% but I will need to more checks to verify this.

The total numbers of particles reported by the the Household are also low by around 25% but sometimes as much as 50% showing the greater efficiency of the 7003.
The PM10 is also reporting low by the Household by same amount.
I know my 7003s underreports the true PM10 concentrations by 10-20% so the Household detector is really struggling to be a valid PM10 detector. I don't consider mine are that good either but there is a way to make it more reliable.

The main problem with the household implementation is it's output is too fast (about one reading every 3 seconds) so low level dust numbers are never going to be reliable and even at high levels when differences between readings are small making decisions on values that jump around all over the place is unwise. This is why all my sensors collect data and average it over 10 readings (I can alter this by a switch on the outside if I need to).

disp2.jpg

The need to integrate particle count number readings is illustrated by the following.

The uncertainty of a count of events is the square root of the count.
In the above photo the >10 micron data is 3 counts and the square root of that is 1.7 so its uncertainty is >+/-50%.
This only gives the uncertainty to 66% confidence, to get 95% confidence you need two times the square root so this takes you over the +/- 100% mark.
Even the >2.5 micron particle counts is only 13 so a +/- of about 7 or 50%
This means ALL low level counts by this detector with this counting speed are basically not worth get excited about.
You need to see at least 100 counts to be getting +/-20% and some sort of reading you can rely on.

The problem with both of these sensors is they only sample a very small amount of air per unit time, around 100ml per minute, so a 3 second count is based on about 5 ml or air, whereas a pro level particle sensor will sample at least 1000 ml/min and even then they will only report their data about every 30 seconds so tat means they have sampled 500 ml of air.

When you go to purchase and use this sensor beware of a the following

• Dust sensors need to be close to "you" to record your exposure - I's no use having the sensor 3-4-5 metres away from where you're making dust as most of the bigger dust particles will fall out of the air before they even reach the sensor.
• OTOH Do not and I repeat DO NOT place the sensor in or under a direct stream of saw dust or you will kill it.
• Ideally you should have it it on your shoulder but that's hard to do and see, so somewhere between 1 and 2 m away is probably a starting point and then bring it in closer if the counts are not too high.
• Periodically you should let the detector clean itself out by running it in a clean place like in a still room with minimal dust sources for a couple of hours.
• Ignore all pm10 readouts based on than <100 particle counts in the >5 and >10 micron bin readouts and and anything less than 1000 particle counts when making decisions - you could end up making decisions the wrong way.
• It is really only a PM2.5 sensor so it will monitor a small fraction of the fine dust in your shed. It simply will not accurately monitor the dust >2.5 microns.
• The dust between 2.5 and 10 microns is significant and has been implicated in nose and throat cancers in woodworkers.
• To be on the safe side multiple all PM10 readings by 2.

If you want to start making decisions with it then you will need to accumulate readings and average these and work with the averages - the longer you can record data for the better.

Give the lowish price, the Household is definite better than nothing but it needs to be used with some care or you will need to buy another soon.

EDIT:
I remembered a website that shows an ongoing real time comparison of three 7003 sensors and four 5003 sensors.
For PM2.5 the 5003s are consistently reading 32% low, for PM10 its 35% lower and number of particles its 50% lower so my quick comparisons mentioned above are consistent with the results on this website.

OK, I took the plunge and ordered one of these 'Household' PM2.5 5003 units. It was a lower entry point for me that didn't require me to do any electronic component assembly to get something going. Mine also arrived yesterday and I gave it a first run in my workshop yesterday afternoon.

Before reporting on my results I need to stress that I don't pretend to understand any of the technical implications of particle size counts and OH&S thresholds other than it is the very tiny particles of wood dust that are the bad guys. As a chronic asthmatic I know they are the ones I need to minimise in my workshop where I spend a couple of hours turning on most days.

I also need to point out that there were quite a few inches of shavings (actually more like splinters) from very old redgum lying on the floor around and on top of my lathe when I took the following readings. So no cleaning up beforehand.

First reading is from outside of my workshop. This is the highest reading I got. It ranged from this to about half of that, depending on the wind and its direction.

[Apologies for some of the photos that are less than crisp. I forgot to take a better camera and had to use my mob]

PM2.5 5003 sensor 1 - outside WS.jpg

That outside reading is probably less significant in my case as I'm not (as yet) venting my DC system outside, which then draws back in the 'fresh' air from outside as displacement. To be clear, I'm not suggesting that venting harmful wood dust laden air out of the workshop and replacing it with outside air with a higher, but less harmful, particle count isn't a good idea. I'm just pointing out why the outside count isn't so relevant with my current setup.

The next reading is the baseline reading in the workshop (next to the headstock of my lathe) before commencing any activity in the workshop, other than walking in, and 24 hours after the last activity in there. The count bounced around a little bit, but settled around about this reading.

PM2.5 5003 sensor 2 - baseline inside WS.jpg

I then switched on my DC system and respirator unit (think of that unit as a small air scrubber with approx. 7" fan and HEPA filter). The particle count immediately jumped.

PM2.5 5003 sensor 3 - start DC.jpg

Then began to slowly drop. Next reading at 4 minutes of DC run time.

PM2.5 5003 sensor 4 - DC 4mins on.jpg

Next at 8 minutes of DC run time.

PM2.5 5003 sensor 5 - DC8mins on.jpg

Next at 9 mins 30secs of DC run time.

PM2.5 5003 sensor 6 - DC9.5mins on.jpg

Next at 10 minutes of DC run time, where it stabilised.

PM2.5 5003 sensor 7 - DC10mins on.jpg

With the DC still running (it mostly does when I'm in the workshop) I then turned some very old and very seasoned redgum. The sort that produces more dust and fine splinters than curly shavings. Yes the count was still dropping.

PM2.5 5003 sensor 8 - turning - DC on.jpg

I then sanded the redgum piece, starting around #120 and going through to #400. Still some dust in the sub-one micron range, but less of it.

PM2.5 5003 sensor 9 - sanding- DC on.jpg

This was only a small 7inch diameter piece. It would be interesting to turn a 22 inch piece and see if that flung more dust out of the immediate reach of the dust port. That will have to wait for another time.

I then took a measurement in the furthermost corner of the workshop away from the DC intake after 30mins of run time. Not unexpectedly there were more particles lurking there out of reach.

PM2.5 5003 sensor 9b - DC 30mins on at far end of WS.jpg

With the DC still running at the 15min mark and immediately after turning and sanding I took a reading at the DC enclosure outlet where it vents back into the workshop. For me this was quite informative.

PM2.5 5003 sensor 9c - DC outlet at 15 mins.jpg

Without a logging function I couldn't get a reading there while I was turning and sanding at the same time, but none the less still informative. I was hopeful of a good number as I have been running this DC system for about 10 years and never seen evidence of any dust on the horizontal surfaces inside the filter enclosure.

My DC-cyclone setup is a key component of my fine dust minimisation, but I combine this with a tethered positive pressure full head respirator running of a large surface area HEPA filter. The next two readings are taken inside the respirator (looking through the shield) next to the scrubbed air intake. I like the first reading, but what to make of the second reading? I will leave that to the experts. A stream of quite clean air directed at the intake of the 'Household' PM2.5 5003 probably just confused it, there not being sufficient air dwelling long enough in the sampling chamber to pick up anything.

PM2.5 5003 sensor 9d - inside respirator - DC on.jpg PM2.5 5003 sensor 9e - inside respirator - DC on.jpg

Anyway, as I said to begin with, I don't pretend to understand any of the above readings and what they might mean.

Taking Bob's advice I will double all of those numbers as my starting point. But at least I now have something as a point of reference and comparison, even if they are not absolute values.

Good work Neil!
Can you remind me what your DC is?
Also how big is your shed.

The high dust levels outside could be a humidity issue - check the humidity next time and see if there s a difference.
The fact that your shed levels are so low means any residual dust has well and truly settled overnight.
A still room will gradual clear itself of dust by settling so it means you shed is also not drafty or leaky.

The fact that your DC is not outside and the levels go down with continual use means
- you are collecting the fine dust at source
- and that your DC is well sealed.
Even though it does not measure the >2.5 micron dust all that well this demonstrates a really good use for that dust sensor.

It also clearly puts the issue of "DC location" back on the table.
Using one of these sensors running all the time this may provide enough monitoring to enable DCs to be re-admitted back into sheds. Although for noise reasons and the fact that that DCs take up space is enough of a reason for me to locate my DC outside.

Rather than next to the headstock a more reliable location would be at head height next to you but not directly under the spray of sawdust that typically comes off a lathe.

Putting the sensor inside the respirator provided with clean air is an excellent way of cleaning up the sensor especially if it accidentally gets hit with a high dust load.

Just a general comment on testing. It's tempting to perform tests at the initial stage of wood working and this is still a good idea to get a starting baseline but a better test would be late in the afternoon after you have been turning all day. This will capture any sort of accumulated dust issues.

Keep up the good work.


I assume more members will start purchasing and testing these and maybe other sensors - I will start a new thread to post results in.

If you purchase a Household dust sensor and want to test that it will actually respond to high dust levels then it is tempting to bring it closer and closer to a source of sawdust. Because you cannot see the fine saw dust you will have no idea if the sensor is close to high concentrations of dust that might damage the sensor. A safer way of testing is to use either a soldering iron and solder, or an burning incense stick. That way you can see the smoke and know where it is. However, don't poke the source direct into the air intake of the sensor, instead light the stick or melt the solder about 100 mm away and then use a fan or wave a bit a paper about to stir the smoke through the air. These smokes are sticky so I wouldn't do this all that often.

The following diagram illustrates the limitation of these dust sensors.

The results are for 80grit sanding of pine and is the same graph as that posted back in post #135, although this time the X-axis is linear and the curves are now labelled correctly.

The pale red curve shows the percentage of the number of dust particles produced <~10 microns in size - this fine dust represents 80% of the total numbers of dust particles.
The red curve shows the same but for particles >10 microns in size - this coarse dust represents 20% of the total numbers of dust particles.

The pale purple shows the percentage of the mass or weight of the dust particles produced <~10 microns in size - this fine dust represents 10% of the total mass of dust particles.
The purple curve shows the same but for particles >10 microns in size - this coarse dust represents 90% of the total numbers of dust particles.
Screen Shot 2018-07-04 at 3.43.06 am.png

I nave also marked 4 dashed lines on the graph.

The particles to the right of D are easily seen and fall very quickly (immediately) out of the air so are not be picked by these sensors. This is fortunately because they would quickly clog the sensor internals. They are of no known concern to health so they can be ignored.

The particles between lines C and D also fall out of the air fairly quickly but some will make it to the sensor. Unfortunately these sensors are not that good at picking these up. Particles close to the 30 micron mark would only be rarely detected and those closer to the 12 micron mark would only be detected with some 10's of percent efficiency. The sensor manufacturers do attempt to allow for this inefficiency but this is not as successful as we would like. The dust around the 9 to 12 micron range can be lodged in the nose and upper throat and is of concern to health being implicated in nose a throat cancers.

Particles to the left of line C will stay suspended in air long enough for these sensors to pick some of them up. Those between lines B and C will be detected with ~50% efficiency, Those between A and B with >50% efficiency. The sensor manufacturers corrections for this inefficiency is much more successful than for larger particles, especially around the 2.5 micron range (since this is what these sensors were specifically built to detect). The newer sensors are much more efficient and corrections for the remaining inefficiency is better than the older sensors.

So for Wood Workers there remains a bit of a gap between what these sensors detect and what we would like to know.

If you already have good dust collection the most likely dust remaining will be the fine stuff so these sensors should be adequate for monitoring residual dust but if your DC system is poor then the sensors may not adequately monitor the residual dust.

Originally Posted by BobL

Can you remind me what your DC is?
Also how big is your shed.

Bob, my workshop has a floor area of 32sq.m (8M x 4M).

My cyclone is the Pentz design with a cone ratio of 3:1, pulled with a 3HP 3Ph motor, 12" impeller, pleated filter (1m h x 600mm diam). All ducting (other than final flex hose) and filter housing are custom made with max air flow (minimum of 6" on the inside) and max dust tightness on the exhaust side, in mind.

I consulted Bill Pentz extensively at the design stage. He would have preferred that I have a larger impeller, but after investigating having a larger impeller made for me (there was nothing available off the shelf here in Oz) or able to be imported at a reasonable price, I had to forego that because of the cost. So, the compromise was to go with the longer cone ratio on the cyclone, which is more efficient at dust separation, and that partly offset having the smaller impeller. As you know, most implementations don't use the longer cone ration because of the height restrictions in sheds.

That was back twelve years ago and the system I built is theoretically running close to the required limits.

Although not absolute, the readings from the 'Household' PM2.5 5003 gives me some assurance that my system is contributing to fine dust control in my workshop.

I would certainly make some changes if doing it again now with the other options that are currently available, like the impeller and control unit from Clearview.

Here a is photo of my cyclone with a 3:1 cone ratio.

PIC00057.jpg

And, my customised plenum chamber that connects the impeller housing outlet to the pleated filter that sits on top and the drop bag below (which has never looked like it needed to be emptied during regular use over the last twelve years!).

PIC00077.JPG

PIC00072.JPG

So now that those 5003 units (and similar) are cheaply available, is there any point in paying $300+ for a Dylos?

Originally Posted by Spyro

So now that those 5003 units (and similar) are cheaply available, is there any point in paying $300+ for a Dylos?

Not really.