Hellow from Poland (Minsk Mazowiecki)

[lukaszgryglicki]

Well, I *think* this is UV-B

This is the 1st filter: https://docs.google.com/spreadsheets/d/1y2XAs5Y12vt_tVL7wj4EZregpfbQ8yrxny7_OY7qyrY/edit?usp=sharing

Second filter is Hoya U340.

So I don't think there are IR leaks in this stack - I already presented other photos from thsi stack elsewhere and people agreed there that this is UV-B.

Also lens is passing UV-B (UV-Nikkor) and body is UV-sensitive (50R converted to mono with quartz glass).

If you think there are IR or visible leaks, thats totally possible, I would like to know what looks like a leak and I would be happy to discuss and then try to mitigate this.

Currently I'm qute sure that this stack produces UV-B only photos, even UV-A is surpassed by this stack IMHO.

 

[ulf]

You might be right here. I just wanted to point out that modern car windows are are black also for UV-A.

 

One thing to consider is that the sensor's sensitivity drops for shorter wavelengths and so do the UV-content in daylight.

 

The slope of those drops varies by altitude and sensor type and type of conversion.

After monochrome conversion the sensitivity drop is less steep than a normal FS-converted camera.

Combined those two factors still put a very high demand on UV-A rejection by the filter stack.

 

[lukaszgryglicki]

Yes and this is so true, see:

- F was around 8-16 (UV-Nikkor).

- ISO around 800-1600 (maybe one image was 400).

- Exposure was at least 15s, but rather 50s-2min.

- It was during the day, when sun was shining.

 

In UV-A I would very easy do thsi handheld - tried more than once. If IR leaks would dominate then glass wouldn't be that black. Also my home windows are totally black/opaque in this stack, while with Hoya U-340 + S8612 they are still transparent (and the latter passes some UV-A and also blocks IR), in IR (Hoya R72 - glasses are transparent, with IR850 too - maybe a slightly bit darker, but with this stack they are completely BLACK, even with ISO 12800, f=4.5 and t=1min).

 

[Andy Perrin]

Ulf, if you look at his outdoor photo as a whole, nearly everything but the snow went dark, including car bodies (not just windows), sides of buildings, etc. That is not typical of UV-A results. So I’m leaning towards this being UV-B but it would be nice to have a UV-A of the same scenes as a benchmark. 

[ulf]

I am sure you already know this, but I want to clarify for others too.

 

 BG glass like S8612 or BG39 have a wide transmission peak. When used in an UV-filter stack the main purpose is to surpress IR.

 

An unwanted effect of the BG Glass is that the left rising slope also limits the UV-content. The rising slope gives a peak shift to the right of the U-glass filter as the a peak in the BG-glass rising slope's range.

A U-340 has it's peak around 340nm, but the remaining peak when combined in a stack with any of the BG-glass is closer to  365nm, as the lower UV-range is cut away.

 

This phenomena is working the same way even if the slope is not by a filter transmission, but by a gradual sensitivity change or a decaying UV-content in the light source.

As combinations of two or more decays are multiplicative the resulting decay can fast become very steep.

 

Another thing to be aware of is that your 308nm filter only fulfil the rejection specifications with light of normal incidence.

Used as a camera filter it's blocking is a bit worse. 

 

At our latitudes this time of the year the sun is not that high in the sky even at noon and more atmosphere is filtering the UV light more.

https://www.pveducation.org/pvcdrom/appendices/standard-solar-spectra

 

Your exposure data above say that there is not much light left to become an image.

That light might be upper UV-B or short wave UV-A that leaks though the filters.

 

It is very tricky to know what you really have recorded, but I am sure of that it is not any IR-leakage.

How thick is the U-340 you used in your stack?

[ulf]

4 minutes ago, Andy Perrin said:

Ulf, if you look at his outdoor photo as a whole, nearly everything but the snow went dark, including car bodies (not just windows), sides of buildings, etc. That is not typical of UV-A results. So I’m leaning towards this being UV-B but it would be nice to have a UV-A of the same scenes as a benchmark. 

I am rather sure that there are no upper UV-A in the pictures, but also that the spectra recorded is mostly not much or possibly not at all from the range close to 308nm.

 

This is just theoretical speculations and as long as we have no good idea about the spectral content of the illumination or the sensitivity of the sensor it will be just speculations.

[Andy Perrin]

Ulf, I am not sure I followed all of your comments above (to Lukas, not the one to me). Are you thinking his U-340 is cutting the UVB? There is no BG glass in his stack so why are you bringing that up?

[ulf]

7 minutes ago, Andy Perrin said:

Ulf, I am not sure I followed all of your comments above (to Lukas, not the one to me). Are you thinking his U-340 is cutting the UVB?

No that is mainly for cutting IR if it is thick enough, but it too have a rising left slope at the peak of 310nm, especially if it is thick.

All rising slopes at 310 combined will shift the 310-peak towards longe wavelengths. The question is just how much.

[Andy Perrin]

Ulf, the 310 region is near the very wide U-340 peak. I just checked. 

 

this is for 2.5mm but it’s near the peak, not much downslope:

https://www.phillips-safety.com/shop/optical-filters/optical-filter-glass/u340-hoya-optical-filter/

[ulf]

5 minutes ago, Andy Perrin said:

Ulf, the 310 region is near the U-340 peak. I just checked. 

It is rather near, but not in top off and the slope depends on the thickness of the filter.

Already a 4mm filter give a significant slope

Here I borrowed some graphs from Jonathan:

UV-B is AFAIK defined up to 315nm.

Will there be any meaningful amounts of that left or is the images?

I am not saying it is not, only that it might not be.

[Andy Perrin]

I’m sure he is getting mainly the longer end of the band because the sun falls off quickly there also, but the 2.5mm transmission is 73% at 310nm, so 4mm would be (0.73)^(4/2.5) = 60%, then the 308nm filter passes 52% there, so perhaps 30% gets through. Everything else is much smaller. I do think the curves line up well enough that it’s probably UV-B and some shorter UV-A since both the filters pass short UV-A. 

[lukaszgryglicki]

- There is no BG glass.

- Hoya U340 is *ONLY* to block IR, it is *4mm thick*. Only U340 is not cutting much UV-B, otther U360 or S8612 are cutting UV-B - they are NOT used here.

- 308nm transmission spectrum is here: https://docs.google.com/spreadsheets/d/1y2XAs5Y12vt_tVL7wj4EZregpfbQ8yrxny7_OY7qyrY/edit?usp=sharing.

 

I did calculations for 320 from charts I have and it seems to be very close to 0, a lot more is passed in 310 area and then even around 300.

It of course have some UV-A (short wave edge) but I'm thinking that image is really mostly UV-B. My calculations for IR blocking are suggesting that IR leaks would be very very very small, I can't see anything that looks like IR and every glass I have is almost black (excluding UV-Nikkor quartz/MgCl glass).

 

 

[lukaszgryglicki]

Quote

 

Ulf, the 310 region is near the very wide U-340 peak. I just checked. 
 

 

This is exactly what I found and this is exactly why I use this stack IV 308 and U340 (4 mm) seem to be a very good companion.

I calculated transmittance around 306-314 and it is about 30%, peak was about 33%

 

EDIT: also I assumed UV-B is until 320nm, if you define it to 315 then I'm almost sure there is UV-A there, even quite significiant because of dropping sun output in shortwave and dropping sensor sensitivity.

Still I think I've recorded mostly 300-320 nm... (mostly doesn't mean there is no leak in 320-360 at all or IR leak around 710-720 - I just don't see this).

 

[Andy Perrin]

We should do a full calculation of the actual curve accounting for all of this and see were the real peak is. If I have time today I will investigate. 

[lukaszgryglicki]

I'll do in my IV308 sheet and LYK.

[ulf]

4 minutes ago, Andy Perrin said:

We should do a full calculation of the actual curve accounting for all of this and see were the real peak is. If I have time today I will investigate. 

Then you just have to factor in the sunlight UV-B content at ground level this time of the year at 53°N and the decay of the sensor sensitivity.

If it was just the filter transmission I would have no problem accepting that it was UV-B images we see here 

 

I have seen a similar problem at 55°N

[Andy Perrin]

Yes I mean exactly that Ulf. I don’t see any way to get to the facts other than to take all issues into account and calculate the curve. 

[lukaszgryglicki]

SO here is the curve - just multiplication data for U340/4mm + IV308 - obviously missing part is sunlight downslope in UV-B and sensor sensitivity, but just see blocking in areas above 345 nm - OD4 is at ~345nm (from filters alone) while peak is 0.3257 at 308nm (by incident?) and > 10% is within: ~301/302 to ~320/321.

 

Updated my IV 308nm chart - added Hoya there both 2.5mm and 4mm thics (by copying data from Hoya website and then interpolating values between 10nm apart point lineary - I know not ideal), then added 2 charts "Stack %" and "Log stack" (calculates OD): 

- https://docs.google.com/spreadsheets/d/1y2XAs5Y12vt_tVL7wj4EZregpfbQ8yrxny7_OY7qyrY/edit#gid=975501234

- https://docs.google.com/spreadsheets/d/1y2XAs5Y12vt_tVL7wj4EZregpfbQ8yrxny7_OY7qyrY/edit#gid=2132093567

Data here: https://docs.google.com/spreadsheets/d/1y2XAs5Y12vt_tVL7wj4EZregpfbQ8yrxny7_OY7qyrY/edit#gid=0

 

Stack is blue, Hoya red, IV308 yellow or green.

 

 

 

[Andy Perrin]

Lukas, we need to account for the sunlight and the sensor.

[lukaszgryglicki]

For this I don't have data to account for, especially for the sensor, and I don't think anybody has. Sunlight data is of a lesser problem. But still, based on blocking only, I think this is mainly UV-B wityh maybe a slight of shorter UV-A.

 

[Andy Perrin]

Lukas, while we don't know for your sensor in particular, Jonathan measured for his monochrome sensor, and we can use that as a proxy.

 

The solar spectrum may be calculated for your location and time of day here, and the values we want are probably the global spectral irradiance on a horizontal plane.

https://www2.pvlighthouse.com.au/calculators/solar spectrum calculator/solar spectrum calculator.aspx

[ulf]

Lukasz, do you like base 11?

Your diagams have 11 divisions for 200nm. :-)

 

[Andy Perrin]

Ok, there is some bad news, Lukas. The solar spectrum has an very unfortunate GAP right where you need there to not be a gap... 

 

This is plotted from the data exported by the calculator here:

https://www2.pvlighthouse.com.au/calculators/solar spectrum calculator/solar spectrum calculator.aspx

Using the day December 21, and the time 12p, Lat/Long 52.1933, 21.7152

 

[lukaszgryglicki]

This is probably what google sheet did, I will check tomrrow or later, don't have more time today, I'm also able to calculate sun spectrum from the link provided and adjusted for my city/date, but same here - will update my data sheet with this.

Actually I like base10, base2 and base16 most (as a programmer).

 

EDIT: regarding gap, I didn't checke dthis data yet, but input would be:

- lon: 52.23

- lat: 21.02

- date: 12/21/22

- time: 1 pm

- no clouds

- sum almost highest on the sky which means (90-52)-23.5 = 38-23.5 = 14.5 degrees. (I'm subtracting full 23.5 degress because it's almost the shortest day in the year).

[ulf]

6 minutes ago, Andy Perrin said:

Lukas, while we don't know for your sensor in particular, Jonathan measured for his monochrome sensor, and we can use that as a proxy.

 

The solar spectrum may be calculated for your location and time of day here, and the values we want are probably the global spectral irradiance on a horizontal plane.

https://www2.pvlighthouse.com.au/calculators/solar spectrum calculator/solar spectrum calculator.aspx

Andy, how do you change the wavelength range of that simulator?

 

1 minute ago, lukaszgryglicki said:

 

Actually I like base10, base2 and base16 most 9as a programmer).

 

Yeah, octal and 4-bit nibbles are likely a bit outdated by now