Spectrophotometer for doing my own tests - what equipment to use?

[CharlesS]

So, let’s say one actually does have at least a little bit of spectroscopy experience (I studied physical chemistry in college) and I wanted to do my own spectral measurements of lenses and filters… what equipment would I need to find?  Thirty years ago, it was straightforward to use off-the-shelf equipment to examine and plot spectra of fluid filled cuvettes, but it’s clear that different sample handling is required for testing glass.  If I wanted to try to source used lab equipment (from eBay, for instance) that would allow me to generate plots and convey the data to a computer, what equipment should I look for?   Thanks…

[Andy Perrin]

Ulf and Jonathan will know best, of course, but here is what I have picked up:

 

1) lenses are MUCH harder to test than filters. For lenses, you need to be gathering light from all directions, which means one of those integrating spheres I think?

 

2) Ocean Optics spectrometers seem to be the go-to around here.

 

3) You need light sources that cover the range that you want to test in. No one light source does it all.

 

4) There are many many pitfalls of interpretation in doing your own measurements, and if you read back over the history of this board, you can find some doozies. It's MUCH harder than it looks. You seem to be imagining that it's as simple as popping a lens or filter into a device, like those Perkins-somebodyelse spectrophotometers you find in college chem labs, but it's harder than that because of things like cross-talk, light leakage, noisy measurements, etc. A classic example being the "broken R72 filter" in this thread, which ultimately turned out to be the same spectrometry error made by BOTH Ulf and Jonathan. There is a learning curve here!

 

Ulf ultimately explained the mystery here if you want to skip to the conclusion.

[aphalo]

Filters are not too much of a problem if the light source of the spectrophotometer covers the wavelength range of interest. This is true only for filters with polished and parallel surfaces, i.e., that do not scatter light. If you measure a plastic film like polythene, scattering in the absence of an integrating sphere makes that only part of the transmitted light reaches the detector and the instrument reports lower transmittance than is true. I have used quite frequently an old Hewlett-Packard/Agilent 8453 Diode Array G1103A Spectrophotometer. Its range is 190 nm to 1100 nm and has an open optical pass that makes it possible to measure large filters such as a 1 meter square piece of acrylic. The software that we have in the lab runs only on Windows NT which we have in a dedicated ancient PC isolated from the LAN, and I still keep diskettes to get the data out of it. Lamps are expensive, specially the deuterium one needed for UV. But then even a spectrophotomer like this has the noise floor at around 3 O.D. so it does not provide good enough data to assess small IR leaks in UV-pass filters.

 

It depends much on what you want to measure. If you only want to identify filters out of a known list, you do not need a very good instrument. In contrast, if you want to test if blocking is better than 4 OD you will need a top-of-the-line lab spectrophotomer, most likely with a double monchromator.

 

The general-purpose spectrometers from Ocean Optics do not have a built-in light source. So you would need to buy a suitable one in addition. These spectrometers come in multiple configurations within each model. Most of these will not get you past 3 OD and some models suffer from stray light that is not uniform, which can cause difficulties. It is rather common that IR shows up as an "out-of-place" peak in the UV. This is most clearly demonstrated using an IR LED at 840 nm or 950 nm. The old USB2000 does not seem to have this problem with IR. However, for what it is, compared to current models' new price it is not very cheap second hand at over 600 €.

 

With lenses as they change the path of the light one needs an integrating sphere for absolute measurements of transmission. I haven't tried this approach, but transmission expressed relative to an arbitrarily selected wavelength or range of wavelengths (of high transmission by most lenses) does not require collecting all the light with an integrating sphere. Something to keep in mind is that glass in general and some ionic glass filters in particular fluoresce when exposed to wavelengths they absorb.

 

I haven't had the problems discussed under "broken R72 filter" with the HP 8453. However, even though the transmittance curve looks quite clean, the same data plotted as absorbance or OD show that stray light or noise is limitting the useful range to less than 3 OD. Looking at eBay the HP 8453 second hand is around 1300 €, and refurbished over 3000 €. It was available new for a very long time, at least during 20 years! I did not think it would cost this much until after I had written all the text above. I used one of these for the first time in 1995 or 1996.

[JCDowdy]

This question comes up on this site every few years, see here and further here.  However, ISO 8478:1996 has since been replaced by ISO 8478:2017.

A lot of work presented here is with DIY adapting less expensive equipment which is often fraught with problematic measurement artifacts as Andy stated.

Here is presentation (Powerpoint) you may also find helpful:

Transmittance Measurement. Presented by. Dr. Richard Young. VP of Marketing & Science. Optronic Laboratories, Inc.

 

 

[Andy Perrin]

I have often thought I would like to do some spectrometry, and then I add up the costs and I think, "for that much money, I could buy some pretty awesome LENSES."

[JCDowdy]

7 minutes ago, Andy Perrin said:

I add up the costs and I think, "for that much money, I could buy some pretty awesome LENSES."

Very true.

[dabateman]

Excellent responses above. 

The USB spectrometers like ocean optics can be found cheap used. That's why many of us have them.

I have a JAZ, with #1 grating for UV and separate module with #4 grating for IR, with a long passfilter.

You need to add a power light as well, but I liked that "feature" as then I could test lights I was making.

But you need to be cautious always with what it says and what does that actually means. 

If you want calibrated absolute transmission,  than that will require lots of work, and care. Controls and validation.  There are ISOs and USP documentation and some regulatory guidance to help.

If you just want to know if this plastic cuts off at 380nm or 350nm or 400nm. Than that's an easier question to answer. 

Even the question of "Is this quartz that can only see 230nm or a fused silica element that can see to 190nm?" Again a cut off question,  is what I look for. Fortunately my #1 grating and sensitive detector can see the 185nm Mercury line. 

I may not have been cautious enough with showing data here, as for lenses and filters I mostly care about cut offs, not absolute transmission.  So I hope I don't mislead anyone looking at data I have presented. 

Seeing how hard I can push a light through something is more of what I like to know. 

 

Best response to your question might be why? What do you need that data for or what do you want to know? 

It may just be cheaper to buy filters from reputable sources if its just for filters. 

But if you want to make stuff yourself and see what it does when pushed through something you made, than maybe a spectrometer is worth it,  or maybe not.

 

My set up was not bought for lenses or filter tests. But sadly the main purpose is difficult to still work through.  

[ulf]

Indeed measuring equipments for measuring filters and lenses are quite expensive, and you must know quite well what you are doing to identify and avoid all potential pitfalls.

Spectrometers like the ones I and Jonathan are using, (array spectrometers from OceanOptics), should be considered as building blocks or components to be used in properly setup measuring configurations needing many more components, some of them quite expensive too.

 

The specified crosstalk numbers for such instruments leading to as best OD3 can be overcome with complex setups and procedures. 

To do that you have to know and understand all sources of error and limit their impacts.

 

Even the easiest of tests, measuring a simple filter's transmission can go wrong as we found out in the case of the  "broken R72 filter" 

Measuring deep OD blocking or transmission of lenses is even more complex.

 

The measurements I have presented for lenses in the Lens Data section need very much work.

They are a result from a combination of several measurements of different sections of the spectra, each setup optimised and calibrated separately. 

 

 

[ulf]

9 minutes ago, dabateman said:

Excellent responses above. 

The USB spectrometers like ocean optics can be found cheap used. That's why many of us have them.

Cheap as just below $1000. Then you need fibers collimators, light sources....

Soon the cost of your "cheap" equipment for this has past the cost of a good used UV-nikkor 105mm 

 

Also, remember that the spectrometer has to be of a type with an useable configuration.

The one I got was by chance  very close to optimal for my needs.

[JMC]

The guys have covered the things to look out for very well above.

 

The main question to consider is 'why do you want one?'. Filter testing is relatively straightforward, although it is easy to produce artifacts, especially at the edges of where the spectrometer is reading. Lens transmission to get '% transmission as a function of wavelength' is not simple, and you'll find yourself going down the route of ever more powerful light sources, and building something with an integrating sphere to get something to do the job.

 

For me, I did it because of work - I needed data for my research, and it would not have been cost effective to get someone to run it for me every time I needed it. If I were doing this as a hobby, as Ulf said, for the price of the spectrometer and all the other bits, you could get a UV Nikkor 105mm lens, which would do pretty much any UV work you'd ever do.

[ulf]

The reason for me to do all this is that I am a measurement freak that find pleasure in optimising measurement methods and see how far I can push the ability of my equipment. 

So far I have spent much money on this, but also had a lot of fun in the process.

 

I have been very fortunate to find some of my measurement equipment used in good shape and to nice prices.

If I should have to buy again and get all parts sa new, I think I will have to spend 1.5 - 2 times what a new UV-Nikkor 105mm lens cost.

 

If I had not had this rather special interest I would have been much better off just buying a UV-Nikkor 105mm lens to start with.

I believe that is an excellent lens and it appears like the favourite for both Andrea and Birna here.

 

As I am not into UV-B imaging, or deeper UV, I have not needed the extra reach of that lens compared to my main favourites the EL-Nikkor 80mm lenses, old metal versions.

 

There is some fun exploring and finding UV-capable lenses and help others with limited budgets to have more alternatives.

However in the end there are only a limited number of lenses in a collection that actually get used.

[Andy Perrin]

I think this is key to how you should imagine the situation: spectroscopy should be treated as a second hobby in addition to photography. It has costs, learning curve, and pleasures that are specific to it, independent of what we do here. If you are okay with that, bon voyage, and we can always use more people who can measure things for the rest of us. 

[ulf]

17 hours ago, Andy Perrin said:

I think this is key to how you should imagine the situation: spectroscopy should be treated as a second hobby in addition to photography. It has costs, learning curve, and pleasures that are specific to it, independent of what we do here. If you are okay with that, bon voyage, and we can always use more people who can measure things for the rest of us. 

It is exactly how I see it.

 

The only thing I am not OK with is buying every lens someone is interested in to be able to measure it. 

 

Seriously again, if someone let me have a lens temporarily for measurements I can do that, as long as there is no time pressure and all shipment costs are covered.

That works when shipped within EU as there will not be any added VAT or customs fees.

 

Then I invest for the test in wear and tear of my equipment.

There is limited  life-time in the lamps in my light sources. A new lamp for Lens testing will cost me ca €1500, when it needs replacement and I think dhe deuterium lamp in the light source I use for filter measurements is almost as expensive. 

[Andy Perrin]

I think you misunderstood my last sentence there, Ulf — I meant, “we can always use more people who can measure (their own) things and share the results with the rest of us,” not that we should send you lenses. 

[ulf]

No what you wrote was quite clear and I fully agree with what you mean.

 

Maybe I should have a slightly different formatting and a smiley after that sentence (as edited above now), to indicate that it was me twisting what you said, a bit like joking.

 

I just mentioned that to make clear, that I am open to such proposals, if anyone want to have a lens measured.

 

[Andy Perrin]

Oh ok, I was afraid you thought I was imposing that on you! Sorry, my internet humor reading skills are subject to random failures. 

[ulf]

No problem, my fault that I was not clear enough showing what I was thinking. I often do such mistakes, expecting people to read my mind.