Glass isn't so bad.

[Guest]

This was just a test, as a precursor to another project I'm considering. The point was simply to see just how much effective attenuation of UV I get through glass (regular soda lime glass). I know the transmission drops off quickly by 350 nM, but then, so does the sensitivity of my camera. So this test was just to see if I would really see much of any difference.

 

Method:

 

Point the camera (fitted with the UV pass filter) directly at the LED array. Take three images: A.) no glass; B.) one plate of glass between the camera and the source; and C.) two plates of glass between the camera and the source. Import the raw images into editing, layer, isolate a sample of each wavelength LED, average, and measure the relative luminosity of A/B/C to see if there is a difference/trend. Get more coffee. Plot the results, and post to the forum.

 

Conclusion(s):

 

The results here show that UV attenuation is minimal, through not only one, but two plates of glass. While this certainly doesn't help, when it comes to UV imaging, it doesn't seem to hurt much. And this is good news, especially since I am not planning to use this in the imaging path, but instead as part of a homemade UV excitation filter in the illuminating path.

[Andy Perrin]

If you keep drinking all that coffee, soon you'll be glowing in the UV yourself!

--

 

Light absorption by glasses varies dramatically depending on how each glass has been doped. You can't really extend these results to any other kind of glass than the kind you tested. So long as that's sufficient for your purposes, it should be fine, but you can't conclude anything about "glass" in general from this test.

[Cadmium]

Hi Mark,

What brand of LED's are these?

What is the filter graph and/or data for the Asahi?

[Andrea B.]

The results here show again not only the camera's lower relative sensitivity to 390 nM

 

I don't understand this? Your Asahi filter ZRR0340 has a sharp slope around 390 nm or so. Thus you can't really conclude anything about recording that wavelength because it is truncated by your filter.

 

I will come back later and re-read. I may have missed something. "-)

[nfoto]

In my country, triple glazing is the norm for windows. Besides the glass is treated by undocumented procedures and the internal spaces filled with dry gas (nitrogen? not sure what is used). However in whatever manner they are treated, I can assure this severely impacts any amount of UV that passes. We're talking about 2 to 4 stops. Thus for indoors UV work, additional illumination with UV-rich sources always are required. The mirrorless camera won't even show any image in their finders.

[Guest]

@Cadmium: I don't have a record of which brand of LEDs were used to build the array, sorry. But the transmission chart for the filter can be found here: http://www.asahi-spectra.com/opticalfilters/detail.asp?key=ZRR0340

 

@ Andrea: You are absolutely right - the filter is cutting the transmission around 390 nM, so its not the camera. Though I can still conclude something - pointedly, the objective of the test... that I can pass UV through the two panes of glass with minimal attenuation. Please let me know if you think this is a wrong conclusion.

 

@nfoto: I agree, that triple glazed windows would severely attenuate UV (and that is by design of course). For this application though I am not using windows. Each pane is just a flat plate of cheap glass, 10" x 12" x 3/32" thick. I really like getting things to work without having to spend a bunch of money - and getting quartz windows that size would have cost a much bigger bunch of money than I would be willing to part with!

[Andrea B.]

Well, Mark, I have some questions about the experiment. :D

 

Like, how exactly did you measure the "relative luminosity"? And, do you have to account for reflection loss? If so, then wouldn't you expect something like 4% per pane of glass? (Fresnel, forgive me, if I'm off base here. I am thinking about filters-on-lenses and that may not extend to glass panes?) Also, one problem with LEDs is that you have these hot spots from them. Do you have any kind of diffuser which would smooth out the light beam? And should the dark (dead) area of the beam be counted in the average? And how dependent are the results on the exposure of the initial photograph? For example, do you determine the initial exposure by spot metering the brightest areas? Because once you blow out an area in the photograph, that area should perhaps not be used for determining any relative intensities. (I think) because you have no way to know by how much you have "oversaturated" the channel (the oversaturation is lost). And, finally, you really should know the output of the LED chips used in the array and how much your Asahi filter is (or is not) attenuating that output. But I could probably let this last one pass because you simply want to show a general transmission drop when going from 0 to 1 to 2 panes of glass.

 

It is possible that analyzing the actual raw data of the 3 photographs would give you some quantitative channel data to work with that might help elucidate (word play unavoidable) matters. This assuming other problems are addressed.

 

Be sure to edit your experimental write-up to at least mention the UV-pass filter properties and remove the "wrong" conclusion about the D70 and 390 nm. And add something about your measurement techniques for that "relative luminosity" even if other questions above are not answered in the write-up.

 

[This entire comment reminded me of the olden days in which I occasionally had to attend a dissertation defense for some poor guy who wanted to get hired by Bell Labs and had to present his dissertation all over again. What about this? What about that? How do you account for such-n-so?? Why didn't you consider this over that?? Whew!!]

 

**********

 

For years I could UV-photograph small subjects indoors in my south-facing dining room because the old windows from about 1960 passed sufficient UV on a sunny day. Our new Anderson windows, however, do not support this effort. :P

[Cadmium]

Hi Mark, I don't know your intended use, but maybe you could use UVT.

UVT is plastic, it is clear in UV, and transmits UV down to below 300nm.

I have some of the sheet I have tested and it works great. The sheet I have is 1/8" (0.12", 3mm) thick.

http://www.polymerpl...ents_uvta.shtml

[Guest]

@Andrea: I've updated the post to remove the references to the camera sensitivity. Thank you for pointing that out. I did not address measurement of luminosity because it is a basic graphics application operation, and is best handled in countless tutorials already online for Photoshop, GIMP, etc. I did not address reflection loss/other causes because for the point of this test I was not so much concerned about the reason for any attenuation, but only checking if there is, or is not attenuation, and roughly to what degree; pointedly for the sake of simply determining feasiblity for my application. You are again right about the hot spots in the LEDs. To account for this I used a starting exposure time which was sufficient to generate images without high value clipping. And also to your point, I averaged the images of the LEDs in lieu of a diffuser to generate a single value for each test condition; which includes dark and light areas, relative from one condition to the next. Again, the results are relative to an initial image, which was not spot metered, but manually set (thus, no blow outs (clipping)).

 

It certainly would be possible to quantify data from the raw files in one or more other ways. I didn't take that route here because it was a simple enough exercise to just check the apparent difference from one image to the next. I would like to get abolute quantitative data, but that would take equipment and control conditions I do not have the means to acquire.

 

I understand this can apply only to the glass I have in-hand (@ Andy, above), and that the test sought only relative results. i hope this doesn't detract value from the results I showed, even if only for general information, or interest. It certainly was very useful to me (to prove to myself feasibility of hardware/equipment/etc). And when/if I get to the application I intended in the first place I'll be sure to share photos from that too!

 

Thank you for the reviewing!

[Guest]

@Cadmium: The link you shared points to the acrylic rods, but I did find on their site the acrylic sheets. Which look like they would be perfect for my application (likely much less brittle than the glass I have). But, at $36.35 / sqft its definitely out of my price range, even with the 'discount' - especially since they only offer it in 4'x8' sheets. I think for now, the glass plates I have - at about $4 / sqft - will have to do. I'll just be sure to handle them very carefully (I've already cut myself on one of the edges!). I may actually write to that company to see if I can buy just a couple of 1 sq ft pieces.

[Cadmium]

I noticed it now says 'rods'. I don't know if that vendor still sells the sheet or not, but they sold me less than a sheet.

[Andrea B.]

I hope this doesn't detract value from the results I showed, even if only for general information, or interest. It certainly was very useful to me (to prove to myself feasibility of hardware/equipment/etc).

 

I think it was a cool experiment. It is good to try such experiments to "prove" the facts to yourself and to any interested readers. :)

 

I'm still not sure whether it is ok to average the LED images. Shouldn't you leave out the dark area of the light beam?