Jump to content
UltravioletPhotography

Schott 2017 Filter Program - Graphs Compared


Recommended Posts

This is related to this previous topic:

https://www.ultravio...__fromsearch__1

 

Here is a comparison of all 4 combination filter graph types from the 2017 version of the Schott Filter Program .

Graph types are:

Combi Diabatic T graph (T = transmittance, including calculations for surface reflection) (Combi = combination, or combined/stacked filters)

Combi Diabatic Ti graph (Ti = internal transmittance, with no calculations for surface reflection)

Combi Linear Ti graph

Normalized Ti graph

 

Several things to point out:

 

1) The 'data input' page has a default number in the Peff field at the bottom of the page.

That default (which I think was about 0.91 to 0.94 in my first use of the program) is about the same as the average reflection factor (P) for most Schott filters.

Since gluing a stack together only leaves two sides of reflection then the Peff would be close to the average P number for most Schott filter types.

However, if you are not gluing the filters together, such as screwing several filters together as a stack, then you can calculate the Peff and enter it manually.

First, find the reflectance factor P for each type of glass in the stack, which is located in the 'data table' page.

Multiply all the P numbers together, for example for a two filter stack, P x P = Peff, or for a three filter stack, P x P x P = Peff, etc..

Then enter the Peff into the Peff field at the bottom of the 'data input' page.

The Peff is only used by the program when drawing a Combi Diabatic T graph. It is not used for Combi Diabatic Ti, Combi Linear Ti, or Combi Normalized graphs.

The Peff will be different depending on if the stack is glued or non-glued, and (in the case of non-glued) how many filters there are in the stack.

 

2) There is a typo in the 'data input' page header, I have pointed it out in the screen shot below.

"ti COMBI = Peff ( ti1 x ti2 x ti3 x ti4 x ti5 )" should read as "Ti COMBI = Peff ( ti1 x ti2 x ti3 x ti4 x ti5 )"

(I think that is just a simple cosmetic mistake in the program, which has no calculation influence)

 

Here is a shot of the 'data table' page, showing where you get the reflection P data to calculate the Peff to enter in the 'data input' page.

post-87-0-52549000-1561935710.jpg

 

Here is the 'data input' page, where you can enter the Peff calculation at the bottom (as well as the glass types in the stack, thicknesses, etc..

In this UG5 + S8612 stack, the Peff = the P of each glass type multiplied together, 0.91 x 0.913.

post-87-0-76127100-1561935721.jpg

 

This shows all 4 graph types, including T at the top which is calculated as a non-glued stack.

I am still working on lining these up cosmetically and other details such as 1E-03 vs 1E-10 graph depth. Also notice the Diabatic T and Diabatic Ti graphs have different % line spacing.

Some of these details need to be fine tuned still once I figure out how to fix them

post-87-0-49689500-1561935738.jpg

Link to comment

Here is what I believe would be the difference between the non-glued Diabatic T Combi stack and the same stack glued.

post-87-0-36842300-1561964665.jpg

Link to comment

Thank you for this explanation. Much easier to follow than the manual.

 

So glued to nonglues is 5% maximum gain which would correlate roughly to a 1/10 stop increase. Something I don't think I would notice. Good to know.

Link to comment

Keep in mind, these are just data generated graphs.

The REAL test is to compare with whatever other filter(s) you want, if you have them, and see how different the exposure time is.

I mostly use Combination graph plots to visualize and compare suppression.

Thankfully, the suppression basically stays the same at the levels I am examining with these plots.

Here again, the REAL suppression test is a UV blocked stack using a long pass filter, using the same exposure time as the optimized UV-only shot.

So I am rather happy with all these versions, Ti, T glued, and T non-glued.

The nice thing about separate filters is that they remain more versatile for other combinations and uses alone,

and they work very nicely for me when stacked, I use them all the time that way, without being glued, and I get excellent results.

Never forget, glue is available, and any stack can be glued on request.

Link to comment

There is a small totally unimportant rounding error in the calculated reflection factor for the stack. It should be 0.831, not 0.830.

Just wanted to show that I'm checking your calculations. :)

Link to comment
Yes you are correct, 0.910 x 0.913 = 0.83083. I didn't round it, I just cut it off, but it is closer to 0.831 like you say, but should not make much difference in the graph.
Link to comment

I added transmission/suppression depth down to 1E-10 on this graph, and the same diabatic graph line spacing structure to keep all T and Ti graphs lined up for overlay.

(by the way, this diabatic line spacing has the same spacing structure that Schott uses for their Ti catalogue graphs)

Here is a comparison of Ti glued, T glued, and T un-glued using the same UG5 + S8612 stack.

post-87-0-88926600-1562040115.jpg

Link to comment
Andy Perrin
You cannot measure anything down to 1e-10! That is 1 in ten billion. Whatever is shown there is just meaningless. The data you started with isn’t perfect and so the errors in that Ti data are probably bigger than 1e-10.
Link to comment

I have Schott's phone number if you would like to give them a call... ;)

Andi, I can't say, I am just making the graph show the full data they have in the program, like the 2012 did. I have no idea how their data was originated.

However, their program does contain data that goes below 1E-10, and so does the data sheet for S8612.

Take a look at those. Maybe ask Schott how they come up with those numbers.

I really can't say, but I do find the data and graph plot below 1E-05 of interest to me.

I didn't make the data on their data sheets... I didn't put that data in this program, they did.

I would guess it is not exactly 'meaningless'.

 

First, here is the Schott S8612 Data Sheet:

https://www.schott.c...jun-2017-en.pdf

 

Here is a screen shot of the 800nm range of data for S8612 in the program:

post-87-0-22485000-1562055962.jpg

Link to comment
Andy Perrin
I don’t have to call them, I know the answer. It came from an equation, not a direct measurement. But equations have error. They can produce very tiny numbers like that which are not accurate. Schott depends on users to apply common sense to the numbers. It doesn’t mean they are lying or false advertising or something. They don’t need me to tell them this.
Link to comment
Andy Perrin
Yeah, if you have a look at what people like Jonathan and Ulf can measure (and how noisy those measurements are) you can get an idea of what can be done with spectroscopy. I’m sure Schott can do better by a factor of ten or more, maybe 100 even, but if you look at all the talk of stray light, no way can they do all the way out to 10 orders of magnitude. I am probably hitting a dead horse now. I will be quiet.
Link to comment

Andy, It is Schott's data. I didn't put it there.

I have no idea how Schott got their data.

 

I don't see any harm in the plot data below 1E-05, it is there if you want to look at it, if you feel it is 'meaningless', then ignore it,

it doesn't make the rest of the graph meaningless, the rest of the graph remains the same as it would be if the bottom was left at 1E-05.

All of the previous diabatic Schott graphs I have made going back 8 or more years now all go down to 1E-10, and that was the default, so this is nothing new.

Link to comment

You cannot measure anything down to 1e-10! That is 1 in ten billion.

 

With the way that Schott calls 100% = 1, I think "ten billion" is maybe slightly different.

Link to comment
Andy Perrin

I don't see any harm in the plot data below 1E-05, it is there if you want to look at it, if you feel it is 'meaningless', then ignore it,

it doesn't make the rest of the graph meaningless, the rest of the graph remains the same as it would be if the bottom was left at 1E-05.

All of the previous diabatic Schott graphs I have made going back 8 or more years now all go down to 1E-10, and that was the default, so this is nothing new.

Eh, if everyone is comfortable knowing that stuff at the more extreme ODs is extrapolation and probably irrelevant, then it's fine. I just didn't see any point in modifying the graph to include some meaningless data.

Link to comment

It isn't meaningless data if Schott has it in their program and on their data sheets.

A 2mm thick Schott NG1 filter (ND type) has an optical density of 10.

Link to comment

First of all I must praise Steve for his ability to manipulate the Schott's calculation worksheet so well.

Normally when I try, the results are not that good.

Might be due to that we have different versions of Excel program and OS, but I often screw things of loosing graphical components in graphs etc.

The only thing I object about Steve's graphs is the highlighting at the OD 3 line.

IMHO that is seldom enough to cancel unwanted leakage from VIS or IR, at least not with my UV-gear.

 

Andy is correct about the difficulty to measure a filter in high OD regions, but there are ways around that.

The problem is actually similar for getting precision in the other end too with a very low OD.

 

Schott's data is for a material at a defined thickness, but must not have been characterised only at that thickness.

To get information about the transmission in the spectrum with high OD measurements can be done with thin samples and recalculated to the reference thickness.

To not spoil the recalculation precision the thickness must be measured quite accurately.

For low OD materials the same process can be done with thick samples instead.

 

Steve can tell how thin it is possible to grind and polish a filter glass.

 

There are situations when a higher OD than 5 is important and it might be nice to present a graph illustrating that.

I would love to get a copy of the high OD-improved 2017 version of the Schott calculation sheet!!! :D

 

For "normal" UV-photography with sunlight with a stack like S8612, 2mm + U-360, 2mm OD 5 is normally good enough showing no data in the graph in the IR region.

For experiments close to UV-B where the image sensors have a much lower sensitivity compared to the high sensitivity in IR information about high OD get more important.

 

I am not aware about any phenomenon bypassing the high OD obtained by stacking except for light leakages beside the filter stack.

Link to comment
Andy Perrin

UlfW, it's almost certainly a direct calculation from an equation, rather than a measured spectrum. If you know the exact chemical composition of a glass (which Schott obviously does), it's pretty easy to estimate the spectrum from the composition using the spectra of the absorbing materials, their concentrations, and the spectra of the undoped glass. (It is the same as one does with Beer-Lambert law to find concentrations in chemical spectroscopy, except going the other direction, from known concentrations to absorption.)

 

I mean, I don't really know for sure, I wasn't there, but the total lack of noise and the suspiciously precise numbers across the whole spectrum seem like they would have come from a model (maybe a good model -- that does not mean it's inaccurate).

Link to comment
OK, if the data is from a model as the numerical data suggests, then I would expect that Schott has verified that the model is reasonably correct by local sampling and measurements, all over the presented spectra.
Link to comment
Andy Perrin

I’m sure it is reasonably correct. Put it this way, if Schott can’t measure in a certain place nobody else will be able to either. (Except Corning.)

 

I guess at the end of the day I’m just arguing against putting info on a graph that nobody can verify and isn’t useful to anyone. I literally never need OD10.

Link to comment

I have never used Excel for anything other than the Schott program and for viewing a few spread sheets people have sent to me occasionally.

So if I can do this, anyone can, but I would advise saving incremental back up versions so that you can go-back to some version you were happy with if you mess it up so bad that you can't figure out how to get it back to how you had it the hour before...

I am currently using Excel 2016. My previous old 2010 version stopped working entirely recently, so I was forced to upgrade Excel to be able to use the Schott program again.

 

I can't say enough about the Schott program. I think it is great. I could not do what I do without it. I hope they continue to make better and newer versions of the program.

I would like a spectrometer, but here is my analogy:

Testing a filter with a spectrometer is like baking a cake using some new recipe, and then tasting it.

The Schott program is like tasting the cake first, and then if you like how it tastes, you can bake it.

A really great tool.

 

As far as importing the 2012 diabatic graph which shows 1E-10, you can open the 2012 program, right click on the graph, look for "save as template", then open the 2017 program,

click on the diabatic graph you want to change, look for "change chart type" > "templates", choose the template and load it.

You will loose some info like labels, size, etc., so you will need to reconstruct some of those items, but you will get the same graph from 2012.

That doesn't import glass data, the 2017 program already has the glass data with the 1E-10 and even below in it already, just waiting there to be used, same as the 2012 did.

Link to comment

As far as the orange highlighted 1E-03 line, that is an absolute, not a suggestion, it is not what I recommend,

however suppression doesn't need to be the same for everything. Some kinds of stacks will work fine with less.

UV only stacks will need stronger suppression, because UV light and sensitivity to UV is so weak.

One could easily draw two other lines, 1E-04, and 1E-05.

No mater what lines I draw, not everyone will agree, so I leave the 1E-03 line on there as an absolute warning line, don't cross it with any stack, unless your intention is to add some Red/IR to the mix.

Stay below 1E-04, and 1E-05 is better, but for just about any situation anything under 1E-05 is not needed and will only loose UV transmission, and require more exposure time.

We all know it can depend on the lens ability to transmit UV, but I think it can also depend on the camera.

I use to shoot often with a U-360 2mm + BG40 2mm stack, it didn't show Red/IR leak for me, but it can, and it does depending.

BG40 2mm works about the same as if you are using S8612 1mm. I recommend S8612 1.5mm at least, for this U-360 2mm UV-only stack.

BG40 will work as long as it is thicker. BG40 3mm to 3.25mm will work about like S8612 1.5mm, it may loose a tiny bit of UV bandwidth depth.

However, most of the UV capable lenses are not going to acquired any more UV bandwidth depth than the BG40 3.25 stack below,

and even with a Kuribayashi 35mm (my best lens), I am not sure I you could 'see' the difference.

 

Example:

post-87-0-71799500-1562213020.jpg

Link to comment

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...