Public Lab Research note


a first look at dust with consumer scanners

by mathew | June 11, 2014 02:58 11 Jun 02:58 | #10550 | #10550

What I want to do

See if I can't see PM10 or even PM2.5 particles on a consumer scanner. I purchased a Canon canoscan 9000F MK II, which allegedly does 9600dpi, or interpolated 192000 dpi.

My attempt and results

Sticky pad prep

Using packing tape (probably polypropylene with acrylic adhesive) I made a quick sticky pad and left it out for 30 hours near my window. following the sticky pad method) I covered it in plastic wrap labeled PVC and BPA free, which means its probably LDPE. I got a lot of bubbles and wrinkles trapped, which was bad. Better technique would help, but this seems to be a bit of an issue for getting good sticky pad readings. I might go to a shellac instead.

IMG_20140610_163246.jpg

Scanning

Then I went to scan the sticky pad, first in visible light to determine particle size, and then hopefully with polarized light later on a modified scanner. The scanning experience was underwhelming.

Screen_Shot_2014-06-10_at_5.13.39_PM.png

despite its advertised 9600dpi resolution, the scanner can't even fake it-- it crashes the software to select more than 4800dpi. After running around support websites, installing different Canon software packages, and a good measure of cursing those who build the TWAIN specification into software on consumer scanners (commercial scanners don't need drivers), I discovered a link suggesting my work might all be futile anyways. The effective resolution, according to these testers, is about 1700dpi, not the 9600 advertised.

I gave up and just played with the 4800dpi image.

Image Processing

the raw image made it hard to zoom in and look at dust:

Screen_Shot_2014-06-10_at_6.06.40_PM.png

But pull the contrast severely, and the particles start to pop right out:

Screen_Shot_2014-06-10_at_6.14.56_PM.png

Still, its hard to actually make out very small particles. On a 4800dpi scan, each pixel is about 5.3µm across. In the image below one can barely make out two dots of particles that from their pixel dimensions are 15-30µm across. If the effective resolution is 1700dpi, then the actual image pixels don't get smaller than 15µm, or 3x3 4800dpi pixels. This test seems to confirm the results from Patrick Wagner at Scan Dig, that I mentioned earlier, and the effective resolution of 1700dpi. Bummer.

Screen_Shot_2014-06-10_at_6.21.27_PM.png

Questions and next steps

So much for getting pm 2.5 measurements on a stock consumer scanner. That is pretty disappointing. I learned more about scanner resolution generally, and have some leads on slide scanners that have an effective resolution of 3300dpi that ought to have a chance of picking up PM10, but it looks like PM2.5 is out of the range of any scanner currently on the consumer/entry level professional market.

At this point, my choice is to switch optical instruments, lower my standards to PM 10, or build my own instrument. I'm still pretty keen on the latter option. Having built several projectors before I'm confident in the ability of hacked-together optics to get good results, and there are a whole century of well made lenses on Ebay for cheap because they don't fit new stuff.

Back to the drawing board...


7 Comments

I had a similar experience when I was figuring out how to copy old film negatives. The best (very expensive) scanner at Middlebury College made a blurry mess, so I just took a digital photo with a close up lens and good camera. The resolution was much better, but I only got about 100 pixels/mm (100 µm/pixel). With another extension tube on the lens this could probably be improved by a factor of three or four (20-30 µm/pixel). That's not good enough for PM 2.5, so a longer extension tube would be required. But that might be doable.

Reply to this comment...


right-- and with an SLR the image capture would be much faster than waiting for a scanner.

Reply to this comment...


Hm.. disappointing! I have faith though, keep on geekin on :)

Reply to this comment...


Mat, thanks for working on this crucial issue. lung cancer sucks, and you should apply for an NIH grant for your work to cure it. <3

Reply to this comment...


I may not be understanding this method fully but keep in mind that PM2.5 indicates particulate of aerodynamic diameter less than or equal to 2.5µm. The "resolution" needs to be capable of distinguishing particulate less than 2.5µm not just about 2.5µm. This is especially important when checking for silica where it is the smallest particulate that is said to be most of concern.

Reply to this comment...


From the literature on stickypads, concentrations of particles smaller than the resolution of the scanned image can be roughly estimated by albedo changes in the pad. PM2.5 indicates the median aerodynamic diameter of the particle. My contention would be that being able to see 75% of the PM2.5 range and measure albedo changes for the remaining area ought to provide some useful information.

Whether this can ultimately be correlated to volumetric standards, however, is an open question. I'm getting more skeptical about accomplishing that.

Reply to this comment...


Hi Mat, have you thought of using a microscope? you can get ones that connect to smartphones, which depending on your phone should give you good resolution, I saw from one of your other posts that you have used polarizing filters before as well, can I suggest that a 90 degree off set might not be the most ideal, most crystals don't rotate light by that much. Myself and Amy are currently looking at a similar method for fast asbestos detection and would be grateful if you would rekindle this project since they are both very similar. https://groups.google.com/forum/#!topic/plots-airquality/9HzdtJM-UT0

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


Login to comment.