Some unusual UG filters. UK supplier, H V Skan

[JMC]

Steve, I've removed it from the graph in the thread for now. Until I've spoken with the supplier, it's a 'mystery filter' hence I can't label it as anything.

[dabateman]

Steve, I've removed it from the graph in the thread for now. Until I've spoken with the supplier, it's a 'mystery filter' hence I can't label it as anything.

 

Can you post it just as that. A mystery filter. It would be interesting to see it. Maybe Cadmium or I could tell you what it looks the most like.

Maybe its really 0.5mm ug1 glass. That is not uncommon on a lot of old microscopes.

[Stefano]

I saw it and it looked like a UG1 or something similar. It blocked past 400 nm.

[JMC]

Thanks for the concern folks, I have talked with Steve about this offline, I just didn't want to have the graph on the forum until it has been sorted out. It is definitely not UG5, but may be UG11, however that is speculation at the moment until I have talked with the supplier. If it is that, then it is another interesting filter for the collection, as I hadn't got any UG11 :)

[dabateman]

That woul be a fun filter to play with.

[JMC]

I have a bit of an update on this, but to be honest it has thrown me a bit of a curve ball. I shall share what I have learned as it might be useful for others.

 

I fed back to the supplier what I had found, and even suggested that perhaps it may have been a batch of UG11 rather than UG5. The samples were from nearly 20 years ago, and it is not beyond the realms of possibility that perhaps the paperwork had gotten mixed up.

 

Anyway I got a response saying that they had spoken with their optical consultant, and the conclusion was that it was within spec for UG5 at 0.5mm. Needless to say I was hugely surprised (actually stunned was the word I used), and I thought it better to try and discuss what was going on over the phone, as I was sure there must have been some mistake.

 

This is the summary of that discussion. The specifications Schott work to to release a batch is based on spectral characteristics at certain wavelengths. They shared a spec sheet with me, and for UG5 at 1mm thickness these are as follows;

 

254nm >= 0.80

308nm >= 0.94

405nm <= 0.50

546nm <= 0.05

633nm <= 0.05

725nm <= 0.85

 

If the batch matches these specifications then it can go for sale, if not then it goes to scrap. Thing is, consider the implications of this. At 405nm transmission for UG5 at 1mm thickness needs to be equal to or less than 50%. There is no lower limit specified, it does not need to be above a certain value. So it can be 0.001% transmission at 405nm and still be in spec. Likewise at 546nm, it needs to be at or below 5% transmission at a thickness of 1mm. Again, it can be 0.001% at 546nm and be in spec. Matching the accepted shape of the curve for UG5 is desired, but is not a prerequisite for releasing the product for sale.

 

I'm still trying to get my mind around this and the full implications of it. We see all these idealised filter curves in brochures and assume these are all gospel truth, and yet how often are they?

 

I will say this for the supplier, they accept that what I have doesn't meet my needs, as my needs were based on the expected UG5 curve, and are going to work on solutions with me. They are going to send me 2 2mm polished samples of UG5 from two different batches for me to test. Assuming one of these matches the expected spectral profile at 2mm, they can then machine it down to 0.5mm thickness for me. Personally I still think this mysterious filter is UG11 0.5mm rather than UG5, but that is now a secondary concern.

 

On a slightly different note, on the Schott website for UG5 there looks to be an error. On that page, they have the guaranteed spectral value for 725nm to be <= 0.05 which doesn't even match the output from their own software (it was <= 0.85 in the document I got, which makes much more sense). Try putting UG5 1mm into the Schott software and you'll see what I mean. Link to the page in question - https://shop.schott.com/advanced_optics/Products/Optical-Filter-Glass/Matt-Filter-Plates/UV-Bandpass/UG5/c/matt-plate/glass-UG5

[Stefano]

Interesting. So if a filter curve is "inside" (under) the specifications it is still in spec. UG11 is completely "below" UG5.

[JMC]

Stefano, it's not just a case of being under or over, some wavelengths are specified at 'it must be equal to or greater than', some as 'it must be equal to or less than'. Apparently as long as it matches the criteria for each of the wavelengths it is considered to be in specification.

 

Strange, I know, and I'm still trying to get my head around it all.

 

I still believe however that my mystery filter is UG11 not UG5 and I will keep trying to get to the bottom of that. I think comparing the specification data for both filters (corrected for thickness), with my own transmission data will provide some answers there.

[Cadmium]

That woul be a fun filter to play with.

 

The problem with UG11 0.5mm thick stacked with S8612 2mm is that the UG11 0.,5mm will leak visual range. You can't see that from the data or the graph, but UG11 1mm is borderline,

similar to U-340 1mm, but just strong enough to block the 500nm range visual. However, if you reduce that thickness to half, I fear you will have visual range leaks.

You can do almost the same graph with UG1 1mm, or UG2A 2mm, and those will not have visual range leakage, UG1 1mm stack will have almost the same graph, almost the same violet, but not have any other visual.

 

I see no use for UG11 0.5mm. There are mush easier to come by and better stacks to create the same. 0.5mm thick is very fragile and hard to make.

0.5mm thick UG5 on the other hand looks very interesting.

 

I will skip the usual graphs, I did them to check it all out. Just letting you know to expect a visual range leak with UG11 0.5mm thick, probably in the 500nm range.

[Cadmium]

Yes, that is not UG5, it is UG11.

[JMC]

Now that I've spoke with the supplier, I see no issue with sharing the graphical data for the 'mystery' filter. Please, no comments about 'that is UG11 not UG5', I am just sharing for completeness.

 

Full range graph, and zoomed in on the y-axis.

 

 

 

Interestingly, it has the leak around 500nm Steve was referring to.

[Stefano]

It is total transmission, right?

 

I don't think that you can block red-IR with a S8612 filter (unless it is crazy thick), since this Mystery filter transmits down to 650 nm (and even something below that). You can block that green leak with BG3 (I would suggest 2 mm thickness). BG3 would also block red, so maybe you can make a Mystery (0.5 mm) + BG3 (2 mm) + S8612 (2 mm) stack.

[JMC]

Stefano, the graphs are spectral transmission (straight from the spectrometer).

[Stefano]

Stefano, the graphs are spectral transmission (straight from the spectrometer).

Yes, so they are total transmission. If you want the internal one you have to add that ~8% loss (actually divide by ~0.92).

[dabateman]

Yes, so they are total transmission. If you want the internal one you have to add that ~8% loss (actually divide by ~0.92).

 

That doesn't make sense to me. What Jonathan has measured is the total transmission of light through the filter.

It will not magically decrease.

 

This is the amount of relative light a sensor will see. Relative to what I don't know. Do you zero based on air Jonathan?

If you zero based on a glass soda bottle, it will get better ;)

 

[Cadmium]

Now that I've spoke with the supplier, I see no issue with sharing the graphical data for the 'mystery' filter. Please, no comments about 'that is UG11 not UG5', I am just sharing for completeness.

Interestingly, it has the leak around 500nm Steve was referring to.

 

I don't understand. You spoke with the supplier, what did they tell you it is sup[pose to be? Why am I not suppose to say that I think it is??

Here is an overlay graph.

[JMC]

Ok, just to clarify. I'm measuring transmission with my spectrometer. I set it up in air. With the light source on and nothing in the way that's 100% transmission. With the light source off that's 0% transmission. I am not correcting for any surface reflections from the filter, that is part of the overall measurement. What I am not showing in any of my graphs is 'internal transmission'.

 

Stefano, as I understand it in the Schott program, they go from internal transmission to 'spectral transmission' by multiplying the internal transmission by a reflection correction factor. When I downloaded the program today to have a play with it, I noticed the default value for that is 0.91.

 

In other words, spectral transmission = 0.91 * internal transmission, according to the Schott program.

 

I've been playing with that software for all of 30mins today, so if that is wrong please correct me.

[Cadmium]

The overlay I did above is using T not Ti.

[JMC]

Steve, I am still working it with them. I have done a similar graph to the one you show today, as part of the report I am sending to them. I have yet to get a response, which is not surprising as most none essential businesses have shut down as of today as a result of a government announcement yesterday.

 

I have no great desire to have this as a thread of 'it's obvious that's UG11'. My spectra matches UG11 and not UG5, that is plain to me. Personally I have no doubts that it is UG11. The whole incident has cost me time, and frankly raised my stress levels more than I want, and I would like to not have the deal with more on here.

 

By the way, I am not refering to your comment at 7:25 this morning if that's what you think, my comment about not more 'UG11' related comments is in the rest of the thread moving forward.

[Stefano]

 

 

That doesn't make sense to me. What Jonathan has measured is the total transmission of light through the filter.

It will not magically decrease.

 

This is the amount of relative light a sensor will see. Relative to what I don't know. Do you zero based on air Jonathan?

If you zero based on a glass soda bottle, it will get better ;)

I think that he measured the overall transmission, without accounting for reflection losses.

 

Edit: yes, I was right.

[Stefano]

Stefano, as I understand it in the Schott program, they go from internal transmission to 'spectral transmission' by multiplying the internal transmission by a reflection correction factor. When I downloaded the program today to have a play with it, I noticed the default value for that is 0.91.

 

In other words, spectral transmission = 0.91 * internal transmission, according to the Schott program.

 

I've been playing with that software for all of 30mins today, so if that is wrong please correct me.

Yes, Ti is T*reflection loss coefficient. If you run your spectrometer it will measure how much light is getting through your filter, which is Ti.

[Cadmium]

No, you have it backwards. Ti is with no reflectance (internal transmittance). T is real life (with reflectance added). Ti data is used when calculating glued stacks, because gluing surfaces together avoids the reflectance of those two joined surfaces.

The graphs I have overlaid here are T graphs. I can provide Ti versions if you like, they will be close to the same, except transmission peaks will be slightly higher than the T version.

 

Graph shows how the 0.5mm thick Mystery filter compares with UG5 0.5mm thick.

[Stefano]

Yes, Ti is T*reflection loss coefficient. If you run your spectrometer it will measure how much light is getting through your filter, which is Ti.

Yes, I mixed them up.

[Cadmium]

I have to wonder how they test for Ti, do they glue the light bulb and spectrometer to the glass?

[Stefano]

I have to wonder how they test for Ti, do they glue the light bulb and spectrometer to the glass?

I think that they calculate it by measuring how much light is reflected first and doing the math later.