Jump to content
UltravioletPhotography

Good practice and setup for spectrometer measurements?


Recommended Posts

I will have access to an USB UV-VIS Flame spectrometer from OceanOptics. 200nm - 1000nm

 

Doing measurements with spectrometers is a quite new field for me.

My main test areas would be transmission measurements and emission measurements from different light sources.

I'm looking for information about how to do good measurement setups to get most performance and avoid false readings.

 

I hope to find some literature that I can learn from about this.

 

Please let me know if you know about any good sources for this kind of information.

Link to comment

Thanks Shane

 

The specifications and manual are available on the web.

https://oceanoptics....ads/FlameIO.pdf

The model is a Flame-S-XR1

Range 200nm - 1000nm

(Grating: XR1-500 lines/mm blazed at 250nm)

 

I know in general how to set up these measurements and how to do the basic calculations, but have not much practical hands on experience.

 

The reason for my post is to get an edge in the methods.

 

I want to optimise the methods and get as much performance as possible from this type of array spectrometer.

I know the dynamic range is not comparable to other types of spectrometer setups.

Looking for smart tricks.

What to do to squeeze the last performance out of it.

What to avoid to not get false readings.

Link to comment

Ulf, it seems we don't have many references available from UVP members. Please keep us posted on what you do find. If you post your experimental plan and results, perhaps someone will comment on it?

 

Have you looked for any spectrographic websites online?

Have you searched for any textbooks on the subject by looking on Amazon or elsewhere?

Link to comment
  • 2 weeks later...

Thank you for the links. They are interesting.

 

The STAGE-RLT-T must be practical. It is often vital to have a stable setup to get reliable results.

 

I have been building a structure with the same purpose based on arca-clamps and a long arca plate.

It is a work in progress, and when I have something more presentable I might post a picture.

Link to comment
  • 2 weeks later...
I'm also trying to find out more about this, and will feedback anything useful I come across. I'm changing jobs soon, and will lose access to the lovely UV Vis device I use for filter transmission measurement and reflection for the UV standards I've been making. In future I will have access to an Ocean optics spectrometer, and while I fully accept there are limitations to these, I want to pull something together which can give me some rough data on filter transmission, and reflection of my UV standards, as well as looking at lens transmission. I like to look of the RLT-T stage mentioned above, so will see whether I can pull something together based on that.
Link to comment

The stage discussed above seems to be a mechanically sturdy but simple construction, relatively easy to duplicate with a home-made solution. Basically, what you need are two SMA panel fiber sockets (found on eBay) mounted with the panel mounts facing toward each other, and a mechanism to change the distance between the two mounts (for example a slider, bellows rail or focusing rack). One possible solution is to use cheap bellows, remove the cloth/leatherette, and mount on each standard an aluminum sheet with the fiber mount at its center. Paint all surfaces of the standards facing each other flat black. Verify that the two mounts are aligned,(e.g. look into one, the opening of the other should be visible at its center, and circular. If not circular or not coaxial, there is a sideways or tilt misalignment. You might even try to pass a plastic straw or rigid wire through both socket holes.

 

As for the measurement protocol, this is how I am planning to start soon. Please comment if you have better ideas.

 

Connect one mount to a fiber leading to the light source, the other to a fiber leading to the spectrometer. Place the sample/filter/lens between the standards, and use the crank of the bellows to change the distance between the standards (as little distance as practical should be best). Eliminate bright ambient lights (e.g. wrap the test rig in black velvet) when operating the spectrometer. To compare filters, I would first take a spectrum with the filter in place (S1), then take a second spectrum after removing the filter (S0) but without changing anything else, then compute the transmission spectrum as S = S1/S0. This eliminates the bias introduced by the spectrum of the light source not being flat (BUT, read the following paragraph).

 

The light source must not have emission gaps or deep valleys in the wavelength interval of interest. If the light source has an emission gap, or a low emission, in a given wavelength interval, then you get a lot of noise (generated in the spectrometer sensor) in this interval, and the measurements in this interval are useless. If you use a 365 nm LED, for example, any measurements outside roughly 360-370 nm consist only of noise, and are therefore useless. The light source should be strong enough to give an unprocessed reading from the spectroscope of at least 50% of its maximum output value. You may get away with a 10% reading (noise will be 5 times higher, so get a stronger illumination, for example use a quartz lens as a collimator to focus the light onto the end of the fiber). A 1% reading is way too low, and noise will be 50 times higher. The spectrometer must likewise have a good sensitivity in the interval of interest (or else, garbage in = garbage out).

 

Measuring the transmission of a lens is more complicated because lens speed and focal length play a role, but it is doable, especially if you just want to visually compare transmission spectra (as opposed to measure them quantitatively). Keep the lens centered and aligned to the illumination beam. For example, mount the bellows vertically and use the lowermost standard as a shelf for the lens. Move the lens around until you get the highest reading (meaning that the lens is centered on the beam). You cannot use a baseline spectrum (= infinite transmission) like above with the filters, since with a lens it makes no sense. You cannot directly compare the transmission of lenses of different speed, FFL and/or FL, either (it requires more complex equipment including e.g. integration spheres).

 

One could instead assume that the lenses your are comparing transmit light equally well at 500 nm (which is a reasonable assumption for camera lenses, probably within 10-20% after adjusting for lens speed etc.) and use the value at this wavelength as a reference to normalize the spectrum of each lens, then plot the spectra in the same diagram. This normalization forces the two spectra to coincide at 500 nm, regardless of their aperture etc, making them easy to visually compare.

Link to comment
Thanks for the info Enrico. Like you I was thinking black velvet to eliminate stray light, or I may make a box and line it with velvet, or paint with matt black paint. The comments on the light source make perfect sense, so try and find a good coherent light source which has no obvious gaps. I was thinking using a collimating lens for the input fibre, but haven't thought about diameter yet. I must admit, personally, I was thinking of including an integrating sphere into mine.
Link to comment

An integrating sphere is primarily used to homogenize light, e.g. by iluminating a sample with a narrow beam and measuring the total reflectivity of its surface, including both specular and diffused reflectivity across an angle as wide as possible (close to 180 deg in practice). With an integrating sphere, this requires just one measurement (per incident angle).

 

To measure the transmission spectra of imaging filters we only need a beam of roughly parallel rays (from one fiber exit to the other fiber entrance), perpendicular to the filter surface (we can orient the filter obliquely if we want to model the effect of using the filter on a wideangle lens). Since we can compare the spectra with and without filter, we do have a way to measure the absolute transmission value.

 

I know that an integrating sphere is sometimes used to measure lens transmission, but if we are mainly interested in on-axis transmission (as opposed to transmission across a wider angle), I would prefer to keep things simple and just use a fiber-to-fiber setup - unless trying will show that this is not the way to go.

 

I should expect that an integrating sphere also introduces losses, which might require a stronger light source with all the problems that this involves. I have a small second-hand spectralon integrating sphere somewhere in my drawers (I actually bough two, but one of them was slightly yellowish and I gave it away), so I might experiment with it if the simplest setup is not satisfactory.

 

Edit: Spectralon is known to be sensitive to air pollutants, which alter its reflectivity especially in the UV. This is another good reason to try and avoid the use of expensive and delicate integrating spheres if possible.

Link to comment
  • 2 months later...
A bit of an update on this one. I recently got an Ocean Optics RTL-T stage (yes, I could probably have made something myself, and yes it is quite a simple thing, but it seems to be fairly well made, and will keep everything aligned for the measurements), along with an OO DH-2000-BAL balanced deuterium halogen light source. The plan is to use these with the OO FX spectrometer I have, initially to see if it can be used reliably to measure filter transmission, and eventually to combine with an integrating sphere and measure lens transmission (and lens filter combinations, as well as using it for diffuse reflection measurements). As with all new kit, I'm sure there'll be a degree of learning to get the best out of it, but I'll keep the forum posted as to how it works out.
Link to comment
  • 2 months later...

Well my previous post was accurate, there was a learning curve for this (given I posted that 2 months ago).... I was having issues with signal intensity, and it was down to having a cosine corrector on the fibre connected to the spectrometer. Basically robbing loads of light. I deliberated about removing it for a while, and today I took it off. It made a huge difference, and I have managed to get some transmission spectra of my Baader U and Baader UV/IR cut filters. Both of these filters I also ran on the big commercial UV/Vis spectrometer (dual beam, monochromator, scanning, integrating sphere etc etc) I had access to in my previous job. Below are initial scans from my Ocean Optics based setup (OO), and the Commercial UV/Vis spectrometer (Comm), for the Baader U and UV/IR cut.

 

post-148-0-12375000-1517172185.jpg

post-148-0-62545700-1517172192.jpg

post-148-0-62100700-1517172200.jpg

 

Ok, these are fresh out of the box as it were, and I have done little to optimise the spectrometer settings (I used 500us scans, with 20 averages, so 10ms total acquisition time).

 

They seem to be pretty good at first look, but what concerns me is the signal at short wavelengths, just below where they transmit the most light. Where it should be dropping to zero, according to the Commercial spectrometer, it's 'non zero' for the Ocean Optics system. I understand there is something called 'stray light effects' with these types of spectrometers, and I guess that's what I'm seeing here? If anyone can confirm that please let me know, especially if you have any way of minimising it through through different acquisition settings on the spectrometer.

 

It's a shame as apart from that, I was amazed by how well they matched the big commercial system. Especially given the acquisition times - about 2-3 mins for the commercial system, and 10ms for the Ocean Optics one. Also it takes up about 1m on a desk in my workshop, including the computer, as opposed to a whole bench in a lab for the commercial system.

Link to comment
For a dichroic filter, the angle(s) that the light hits the filter will matter a great deal. Could this simply be alignment differences?
Link to comment

Andy, the setup includes a collimating lens on the input fiber, which I have set to make the incoming light into as parallel a beam as I can. So in theory it should be coming in at 90 degrees to the surface of the filter. I've got a load of absorptive filters I can try too to see though.

 

EDIT - I spoke with Ocean Optics earlier, and they think that 'stray light' is likely to be the issue here. One of the drawbacks of these types of spectrometers compared to the double monochromator ones. The light source I have has two bulbs - deuterium and tungsten halogen. I may try switching one off depending on where in the spectra I'm looking, to see whether that makes a difference.

Link to comment

A little experiment earlier. I had some U340 from Steve in different thicknesses - 2mm and 2 x 4mm. So I thought I'd run them from 2mm up to 10mm total thickness, but with a lot longer total acquisition time, than I would normally use for filters. The aim was to look at the IR region and try and see what I could get down to in terms of blocking assessment. I went with 500us integration time, but 10,000 scans, for a total measurement time of 5s. I did a previous experiment with 1,000 scans, but it didn't show as much IR data as I would have liked so I thought I would push it to 10,000 scans. Firstly, the full spectra range - 250nm to 800nm

post-148-0-91549900-1517231720.jpg

 

So as expected, increase the thickness and the transmission dropped. What I thought was more interested in though was the IR region. This shows that region;

post-148-0-62822900-1517231795.jpg

 

Going in further now to specifically look at the IR the thicker samples are letting through;

post-148-0-04767500-1517231848.jpg

 

Yes, I know the baseline is dropping below zero which isn't good, but I think I can still see the bump above the baseline in the IR from the 6mm filter stack. Can't see anything from 8mm or 10mm.

 

Steve, would it be possible for you share computational data for these stacks - it would be great to see how they compare?

Link to comment

This general topic has been discussed a lot since well before I joined this group.

 

Some time ago I posted some information in a couple of topics which might be worth reviewing in case you missed it.

 

60/2.8D, 60/4.0 CO & Some Transmission Spectroscopy
and and parts of a later followup

 

What is the real response of Panasonic Lumix G3?
Link to comment

Thanks for the reminder John, you're right this area has been discussed a few times before. I suppose the issue is, that with the advent of USB spectrometers for 'relatively' small amounts of money, more and more people are getting their hands on them, and coming up with the same issues and questions when they start playing. I accept there are limitations to what they can do, especially compared with dual beam dual monochromator systems, but people will still experiment with what they have or can get hold of relatively cheaply. As with all techniques, we just need to be aware of the limitations of equipment we have, and work within them.

 

Anyway, I am concerned I have hijacked Ulfs original thread. Andrea - if you think it would be better, feel free to split my filter transmission post out from this one.

Link to comment

Here is Hoya U-340 4mm +.

U-340 4mm needs about 0.75mm to 1mm of S8612 to be suppressed.

Keep in mind that sheet thickness is 4mm, unless you can find someone who has it in block form and the equipment to slice from block, not sure about the availability of that for Hoya glass.

It could be stacked or cemented from two thickness of glass cut from sheet.

This is T data (not Ti), diabatic form to show OD suppression level.

post-87-0-77018900-1517255418.jpg

 

This graph below is Ti data.

This transmission has a peak of about 340nm, and using the best of 'accidental' UV friendly lenses (Kuri 35, for example), you will be reduced to a peak of 365nm at about 40% transmission,

thus the real transmission will be fairly constrained, and you will probably experience some strong 'grain' (noise). You may get something nicer with a UV-Nikkor, perhaps, I don't have such a lens to test that.

You will need a UV-Nikkor (or the like) to use these thicknesses and stacks to their full potential.

For the usual UV friendly lenses, you should stick to about 2mm thick for the U-340 and use thicker S8612 instead ( Kuri 35mm, El-Nikkor 80mm, etc.).

post-87-0-63016800-1517255965.jpg

Link to comment
Thanks Steve. So looking at the transmittance graphs then, the 4mm U-340 is letting between 1% and 0.1% (but nearer to 0.1%) of the IR through at around 720nm, and the 6mm one is letting through about 0.01%. My data was saying about 0.28% IR peak transmission (above baseline) at around 715nm for the 4mm U-340, and 0.01% above baseline for the 6mm stack (albeit a bit noisy). I'm quite happy with that for a USB spectrometer :)
Link to comment

From my limited tests with thicker U-340 stacks, compared to more 'normal' thickness of U-340, UG11, U-360, and UG1 used in stacks,

I find that the blue channel is greatly reduced, which is I suppose due to the lack of upper UV-A in the 360nm/380nm to 340nm range,

which is more saturated with a UV-only stack using thinner U-glass.

Thinner U-glass will move the UV peak and body transmission upward toward 400nm, whereas a thicker U-glass will more the UV peak lower.

Below is an example comparison of the blue channels from 343nm peak and 359nm peak UV-only stacks, using Kuribayashi 35mm lens.

The blue channel becomes rather dark and noisy with the 343nm stack, which is U-340 4mm + S8612 0.75mm in that test.

So unless you have a UV-Nikkor, then I think you are wasting your time and money with filter that have deeper transmission or peaks, and you will not be very happy with the results.

 

Blue channels of two UV-only stacks with 343nm peak (left), and 359nm peak (right):

post-87-0-12625600-1517261686.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...