[UVC SAFETY WARNING] UV-C light sources & filters

[colinbm]

The last one looks very UV, is it UVA or UVB ?

 

[lukaszgryglicki]

Im doing all to get UV-B only... I'm quite sure it is UV-B in mono shots, in color it is probably a mix of IR leak and some UV-B - ratio is unknown.

Note totally black windows (everywhere) and totally white skies while objects still casting shadows (this is in sunlight around 1PM).

 

So it is full sun UV-B, typical sunny 16 gives f=16 time=1/100s (ISO 100), I'm getting f=11 time=30s (ISO 100) this gives 1 f difference and 11.5 stops from t (12 stops is from 1/128 to 32s), overall 12.5 stops which is 2^12.5 = 4096*1.41 = 1/5775 of sunlight = 0.017% of sunlight...

 

 

 

[dabateman]

There is very little UVB in sunlight,  especially this time of year.

Also your Nikon with stock coverglass on the sensor will be blocking UVB.

So the long exposure times are expected. 

Black windows is a good sign, as there shouldn't be UVB passing through a window,  if its from a reputable company. 

So looks like you are imaging the dark world of UVB.

[lukaszgryglicki]

But probably not in CFA version or maybe just a tiny bit at a very very very hight exposure cost and IR leaks... but mono seems to be very OK and usable.

Will see how this changes with quartz mono GFX and update here.

 

Quote

 

Also your Nikon with stock coverglass on the sensor will be blocking UVB.
 

 

Actually I don't know if coverglass is removed stock or what...

I know that IR/UV cut glass is removed and CFA is physically or chemicaly removed by brute force what you can see on edges of the frame.

 

I wonder how transmission curve looks like for U340 4mm + IV308, for IV308 I have a chart (the exact chart of mine filter): https://docs.google.com/spreadsheets/d/1y2XAs5Y12vt_tVL7wj4EZregpfbQ8yrxny7_OY7qyrY/edit?usp=sharing, if I could have similar data for U-340 I could create a chart... I just wonder what the peak weavelength for such a combination is what is approx FWHL and what is OD level of vis/IR blocking...

 

 

[Andy Perrin]

Lukas, instead of bulb mode, an alternative is to average together multiple exposures to reduce noise. I would do multiple 30 sec exposures and then take the median at each pixel across all photos. 

[Stefano]

Photos look fine, although the orange-yellow color is a bit weird for me. Usually (based on my personal experience and Andrea's tests, this color appears at ~340 nm, and UVB usually looks green, as I said before). But this may not mean much.

[lukaszgryglicki]

This is probably leak?

I think it is next to impossible to get UV-B on a camera where CFA is still in place...

Or some other error that I don't know...

This should be mostly around 310 nm from approx transmission curve of U-340 (4mm) + my IV308...

 

[Stefano]

It's not impossible, but you need very good blocking. I use two bandpass filters stacked (each rated at OD 5), because one was not enough.

[Andy Perrin]

It’s most likely 340nm-ish, based on what we have seen in other people’s previous tests and the image colors you got. Occam’s razor: the simplest explanation is leakage unfortunately. 
 

Unfortunately the camera sensitivity is much higher at longer wavelengths. So even if your filtration maximized the 310nm light, it’s so much less sensitive there than at 340nm. I think sensitivity goes down exponentially or nearly so. 
 

https://en.m.wikipedia.org/wiki/Urbach_tail

[lukaszgryglicki]

Yes, but monochrome is just fine, so I will focus on mono - CFA was just a test to see what happens.

 

[Stefano]

Try to take a photo of something made of glass (such as a magnifying glass). You can have the sky in the background as a source of light. In my tests, a magnifying glass is totally opaque at 310 nm and mostly transparent at 340 nm. See if there is an obvious difference between your color camera and the monochrome one.

[lukaszgryglicki]

Good idea, but already did with windows and through my window in live view mode (with mono) - window is black - several posts above it I check it blocks 6 stops or more.

 

[Stefano]

Yes, I remember that, but I think you didn't test this with your color camera (I can't find a post on that). Some windows (like mine) already block at 340 nm, so they may not be a good test subject (they can appear black with both your cameras, even if they see different UV bands). You need something which looks noticeably different at 340 and 310 nm.

 

This test is just to make sure that the orange color is from a leak, and to see if your monochrome camera is actually seeing UVB.

[Stefano]

Some concrete examples (the wavelengths are indicative, my filters are not that tight). TriColours as a bonus:

 

310 nm:

 

340 nm:

 

387 nm:

 

TriColour:

 

310 nm:

 

340 nm:

 

387 nm:

 

TriColour:

[lukaszgryglicki]

Nice idea, will try to test this, but I think for CFA it will just be a pure noise while for monochrome this should be OK.

Need to wait for a sunny day, tomorrow I'm driving to a soccer championship with my son (on my motorbike) so I have my weekend busy...

Hmmm I may take my UV-Nikkor and filters with me but I'mn a bit scared if they "like" travelling 350 kms with motorbike...

LOL

Motorbike is worth about $5k max now

UV-Nikkor is around $7k, IV308 is $800 and cheapest are my cameras - two differently converted Nikon D600s....

 

[Stefano]

I recently recorded some 340 nm videos on the beach (some in the afternoon, without direct sunlight and at 60 fps), so it's doable.

[lukaszgryglicki]

Yep, but now I speak about UV-B only - stack of Hoya U-340 4mm + IV308nm - only mono D600 makes any sense.

 

[lukaszgryglicki]

Got data from Edmund for UV-C mercury filter (designed for 253.7nm line), here it is: https://docs.google.com/spreadsheets/d/1lREgR82Tkv06je7GN9yVzUxId1tqAfupE_liIpYkvVE/edit#gid=117646081

Does it look as a good candidate to make UV-C only photos with low pressure mercury lamp?

I'm concerned about leaks starting around 650nm+ - will they dominate with low pressure lamp? 

Last two lines listed on wiki are:

578.2		yellow-orange
650.5		red

This filter gives around 4.8 OD at 578 and 4.6 OD at 650 but OD4 around 658...

While the main line 253.7 is OD 0.27 so this should really isolate that lime IMHO - correct me if I'm wrong?

 

[ulf]

It is not enough to just look at the filter OD for this.

You have to include the difference in sensor sensitivity and light source intensity at UV-C compared to the offending light in VIS.

To not see any contamination of unwanted wavelengths you will need a detected difference of 4 stops or more. 

 

I have no personal experience with deeper UV-C imaging, especially not with a monochrome converted camera, but my gut feeling, for what it is worth, say that this might be a difficult borderline case.

With the UV-Nikkor at least the limitations from a lens is not present.

 

[lukaszgryglicki]

I know sensor sensitivity drops down there... but this is literally the best option I've found so far. Of course only with UV-Nikkor.

I guess I need to take the risk and just try...

To be clear: I want to do photos under low pressure lamp ONLY - darkness otherwise, I want to just shine it and do long exposure, probably 10 minutes or so... with ISO 100, f=8 or 11, just wondering how much non-UV-C contamination I will get when using this lamp as the ONLY light source, nothing else. Sunlight has no UV-C...

[Stefano]

If the filter leaks, a simple but expensive fix is to buy another one and stack it on top of the first one. This way you will reduce the overall transmission (especially if it was already low), but you will double the OD.

[lukaszgryglicki]

Well, very expensive...

 

[lukaszgryglicki]

My Edmund filters arrived, bot with 52mm lens mount device.

- 253.7nm one is leaking a lot of visible, I can see through it but only at sharp angle.

- 220nm one is pitch black and my mono Nikon D600 sees nothing - so this one seems to be really blocking ... also that D600 has no quartz glass so it is not able to see 222nm KrCl light - will check with 50R mono quartz (of course I'm talking about UV-Nikkor).

[Lou Jost]

It's well known that pure quartz objectives and lenses (those with no fluorite or other elements) have no way to focus all wavelengths onto a single plane. So with these lenses, the IR leakage will be de-focused a lot, and may not even be visible except as flare, which could be subtracted out if it is not too strong. So for those of us who don't have such good lenses as a Nikon UV lens, maybe we can get away with less-than-perfect filters?

 

I was surprised that for my quartz lenses, even the focal difference between 254nm and 365nm is huge.

[lukaszgryglicki]

Yes, pure quartz lenses have a lot of chromatic abberations, I've even posted some tests for my UV5035BK