Jump to content
UltravioletPhotography

Build thread - at home measurement of camera UV spectral response


JMC

Recommended Posts

Ever since starting with UV imaging, I've been interested in the actual spectral response of the cameras to UV - is there a significant difference between cameras, what effect does the monochrome conversion have, what happens when you put filters on etc? I looked into getting spectral response of the Canon and Nikon UV cameras I have, while I was in my previous job, but the cost quoted to me was astronomical - something like £30-40k for the 2 cameras, to have their responses mapped from 300nm to 1100nm, and I wasn't able to progress with it. There is some info online, but from 400nm and above, and with varying degrees of information on how or where it has come from. I couldn't find anything on the 280nm to 400nm range. So I got to thinking, why not do it myself?

 

I'm not a complex math type of person, so the approach I'm planning is to do it the old way - light source, monochromator, integrating sphere (with a port for the detector to measure light intensity, and a port for the camera to photograph the light). They would be mounted in a box to keep excess light out, and the camera would then image the open port of the integrating sphere. Adjusting the monochromator would let me get light of different wavelengths and the spectrometer would allow me to calculate intensity.

 

I'm not really fussed about the whole visible region, and while IR would be interesting I'm mainly after the UV region and just into the blue end of visible.

 

This is going to be a long term build thread, and not a high budget one. I picked up a manual monochromator and small integrating sphere from eBay (I think I spent about £400 on the two of them). The monochromator is for use between 200nm and 800nm, and the integrating sphere a 2" spectraflect one from Labsphere.

post-148-0-40168700-1517579218.jpg

 

Other than those two I want to use stuff I already have if possible. I made a start today - looking at the monochromator. To be honest I wasn't even sure it was going to work given where I got it, but I set it up on my USB spectrometer rig and measured % transmission as a function of wavelength (% was compared to not having the monochromator in between the source and detector fibers). This was what the output looked like.

post-148-0-96851700-1517579330.jpg

 

Obviously a lot of light gets lost in the monochromator, but also, I was manually positioning it between the two fibers which was tricky. It has a 300um output slit so I can imagine I wasn't optimally set up to get the best signal. The input port needs something to attach the fiber too as well, as it was missing any form of connector when it arrived. Overall, the monochromator looks like it works though, and given the width of the peaks I should be able to look at it in 20nm intervals.

 

As mentioned above, this is going to be a long term build thread, so I wont be providing updates every week. However I will add to it as an when I can.

Link to comment
I have often wondered about this question myself--I know all cameras are not equal, and I have sometimes wondered whether Sony cameras such as mine are less sensitive or have narrower bandpass than some other marques. About 15 years ago, received wisdom had it that cameras with CCD sensors had better UV bandpass than those with CMOS sensors. I don't hear this much any more, although the exact reasons have not been explained to me. Perhaps you can shed some (invisible) light on this old question. Related questions include whether "debayered" sensors have deeper bandpass than unmodified ones, and how much difference various conversion cover glass materials make.
Link to comment

I came across my first challenge. Once the light leaves the monochromator and enters the integrating sphere, the intensity obviously drops. Thing is it drops lots, and I was struggling to get a reading with the spectrometer from within the integrating sphere. So this leads me to my first assumption. I am assuming that the Spectraflect coating within the sphere reflects the wavelengths of interest (between 280nm and 480nm) with equal efficiency. Looking in the literature then there is a little less reflection at 280nm than 480nm, but only a few percent. As I am not dealing in absolute quantum efficiency measurement, I shall assume that what I can measure coming out of the monochromator at the different wavelengths will be the same as what comes from the sphere. As the aim is to get a relative response curve as a function of wavelength, I think this assumption should be fine (although it will slightly increase errors).

 

To be honest I will be happy if there is enough light to be photographed reliably, but that will have to come later once it's all assembled. There is a fall back position which is a larger fiber from the light source - 1000 micron vs the 600 micron one I have now. That should give me about 3 times as much light, but obviously means buying a new fiber.

 

I started building the housing for all this, see below;

post-148-0-76576900-1518014123.jpg

 

The monochromator and integrating sphere will sit in the enclosed section at the top left of the housing, the light comes out of the integrating sphere and travels down the enclosed section from left to right, and the camera with lens will be at the hole at the right side of the housing. I've made it to fit the Rayfact 105mm, as I can use this on both my Canon and Nikon cameras. To block any external light there will be a lid on it obviously and I'll wrap around the lens with black velvet. I've made it to be just longer than the minimum focus of the lens. I theory it means I can also use 52mm filters on the lens as well to see the effects. The inside of the whole thing will be painted with Semple 2.0 Black paint (https://culturehustl...st-art-material). I've worked with this paint before and it offers excellent light absorption even down in to the UV. Lots of sanding and gluing before it gets to that though.

Link to comment

Looking forward to the results!

 

and I'll wrap around the lens with black velvet.

Are you sure that black velvet is black in UV?

Link to comment

Are you sure that black velvet is black in UV?

To be honest Andy, no. But I'm going to need a flexible cloth which I can use to try and plug any potential light leaks between the lens and the apparatus. Black velvet is where I'll start, and if it doesn't work I'll find something else....

Link to comment

I came across my first challenge. Once the light leaves the monochromator and enters the integrating sphere, the intensity obviously drops. Thing is it drops lots, and I was struggling to get a reading with the spectrometer from within the integrating sphere.....

 

This is typical, the throughput of an integrating sphere is usually ~10%. Is it possible to widen the bandpass on your monochromator?

Link to comment

You might consider using a diffuser rather than a sphere. While the simultaneous spectrometer measurement while photographing a uniformly illuminated port is an elegant approach you are still only getting relative spectral data.

 

I suggest you consider that you can connect the monochromator directly to your camera via a suitable fiber. You likely already have a suitable fiber optic diffuser tip which could be placed in close proximity to the sensor, perhaps with an extension tube.

Link to comment

Thanks John. I can get other slits for the monochromator, but they aren't adjustable. So more money if I go that route, and to be honest I don't want to widen the peaks more if I don't need to.

 

I hadn't thought about connecting the monochromator directly to the camera (I'm assuming without a lens). I'll keep that in mind, but need to test out this approach first. Fingers crossed I'll get enough light to get good, reproducible images.

Link to comment

You might also consider attaching the monochromator directly to the light source.

 

I don't recall you saying what your light source you have.

Link to comment

The light source is an Ocean Optics DH-2000-BAL John.

 

Yes, I seem to recall now you had a dual source, but I do not think you can mount your monochromator directly to that output.

Link to comment

Yesterday was a quiet day at home, so I cracked on with building this. I got the final pieces cut and dry assembled, and then painted it all up using the Semple Black 2.0. The monochromator and integrating sphere were installed, and I did a mock up with the camera showing how it will all fit together.

post-148-0-84823600-1518168715.jpg

 

post-148-0-33868000-1518168733.jpg

 

This shows the exit port of the integrating sphere and what the camera is imaging.

post-148-0-14660500-1518168724.jpg

 

And finally, with the last cover on.

post-148-0-57759500-1518168742.jpg

 

This has been a really fun build. I used off-cuts of wood I had laying around the mancave, which have come from old pallets and packing crates. These were the bits too nice to put on the log burner. Lovely high quality building materials, all wonderfully straight and true.... The one thing I did have to think about was how to attached the fibre optic collimator. The monochromator i had is an old model and the screw thread is non standard. I contacted the UK distributor for information on it (and to see whether they did an adapter). They couldn't provide me with the thread size, but did kindly offer to sell me a whole new monochromator, which was nice of them. I decided to fix a mount within the wood of the case, but just in front of the opening to the monochromator. Rather than buy an expensive custom made or 'optical grade' mounting point I realised the screw thread was the same as some fixings used for fuel systems in cars when people decide to 'bling up' their engine bay, so I bought an nut of the right thread pitch on eBay for £1 and JB Welded it into a hole in the wood I chiselled out. Job done.

 

Anyway enough of the build story. The exciting bit now, does it work? I took some pictures with it last night. I shall be posting them later, once I've gone through and had a good look at them.

Link to comment

The exciting bit now, does it actually allow me to get photos. In short it does, but wow is there a lot of light loss and it is right on the edge of what the camera can see. I ended up using 30s exposure, at ISO1600, and f4.5 with my Rayfact 105mm lens with my Canon EOS 5DSR. The first set of images is using a standard camera - un modified and visible light capable. I started at 480nm and worked downwards in 20nm intervals (the images have been cropped slightly and reduced in size for here, other than that they haven't been modified).

post-148-0-44206000-1518173277.jpg

 

post-148-0-87852800-1518173286.jpg

 

post-148-0-17786400-1518173294.jpg

 

post-148-0-83785400-1518173304.jpg

 

post-148-0-36535000-1518173314.jpg

 

At 400nm and under with these settings the image then looked black down to 280nm. If however I play with brightness and contrast in the images I can see the circle at 400nm, and even something very very faint at 280nm, however it is extremely hard to see. I am not sure whether this is an artifact or whether it points towards an extremely low level transmission through the UV/IR cut filter, but I'm sure this will be only 1 of many questions as I work through it all. I also need to get a black cloth to cover everything, just in case of light leaks.

 

It looks like in essence it works - I can see the light from the integrating sphere, and between 420nm and 400nm the internal UVIR cut filter in the visible light camera kicks in and blocks the shorter wavelength light. Woo hoo....

 

I must emphasize these are jpegs just taken quickly to test out the system. I still need to figure out the correct approach for how to correct the images for the actual light intensity at the different wavelengths. I know I'm going to have to use raw files, but I haven't thought more about that yet.

 

The light is dim, very dim. I'd love to get the 1000um fiber which should give me about 3 times more light, but given the cost that will need to wait for a while. Next up, images from the monochrome 5DSR conversion.....

Link to comment

Now, the monochrome camera. Same settings, ISO1600, 30s, f4.5 with the Rayfact 105mm. Camera is a multispectral monochrome conversion from MaxMax. The window over the front of the sensor is a Schott WG280. Again I started at 480nm and went down in 20nm intervals to 280nm.

post-148-0-25959500-1518174037.jpg

 

post-148-0-27173200-1518174044.jpg

 

post-148-0-06043700-1518174052.jpg

 

post-148-0-77152300-1518174059.jpg

 

post-148-0-63094200-1518174066.jpg

 

post-148-0-18599200-1518174077.jpg

 

post-148-0-74085100-1518174088.jpg

 

post-148-0-78967900-1518174095.jpg

 

post-148-0-96899900-1518174109.jpg

 

post-148-0-86069700-1518174117.jpg

 

post-148-0-28669600-1518174129.jpg

 

Very different behaviour now to the visible light un-modified camera. Below 420nm the camera retains sensitivity, and while it is getting darker and darker, the light remains visible down to 300nm, and even faintly at 280nm. This is in keeping with the window being WG280 - I would expect transparency down to 300nm then a rapid drop off below that. Also the sensor will be less sensitive as the wavelength drops, so I'm not surprised the image gets progressively darker at shorter wavelengths.

 

As before these have not been corrected for actual intensity of the light source at the different wavelengths, but it does look like I can see something down to 300nm if there is something there to be seen.

 

The uncropped images do show some evidence of a light leak within the camera though at 30s and ISO1600, which wasn't there on the unmodified one. Not surprising I guess given the UV/IR cut filter has been removed.

Link to comment

Carrying on with the monochrome camera, this is where things get funky. How far into the UV can I go an still see something? So I cranked up the ISO to 6400 and stayed at 30s exposure. I started at 380nm and went down to 280nm in 20nm intervals and then carried on 10nm at a time, down and down....

post-148-0-27464700-1518175649.jpg

 

post-148-0-89538800-1518175657.jpg

 

post-148-0-90461100-1518175665.jpg

 

post-148-0-59301100-1518175672.jpg

 

post-148-0-13133100-1518175680.jpg

 

post-148-0-32378700-1518175688.jpg

 

post-148-0-03149800-1518175696.jpg

 

post-148-0-64018400-1518175703.jpg

 

post-148-0-74173000-1518175710.jpg

 

post-148-0-82392900-1518175729.jpg

 

post-148-0-08704100-1518175742.jpg

 

post-148-0-56302000-1518175748.jpg

 

This really is pushing it now - the sensor sensitivity will have dropped drastically and the Schott WG280 glass over the sensor should be blocking most of the light down here, and yet there is still something there at 220nm - the circle is still visible, although hard to see. Of course ISO6400 and 30s exposure is not without problems (under statement of the month) and the IR leak at the edge of the images is now extreme, but it looks like the sensor will still see the UV, if there is UV there to see even at these short wavelengths. It would see so interesting to test a camera with a window which isn't designed to cut off below 280nm....

Link to comment

Could it be a leakage in the monochromator or that the material in the sphere has a week fluorescence?

I'm not sure, but I don't think so Ulf, as the normal camera images went properly black at short wavelengths at ISO1600. If there was any significant light leak or fluorescence (in the visible region) I would expect to have seen it there. I suppose it could potentially be UV induced IR fluorescence on the monochrome camera images, but does Spectraflect fluorescence at all? Boosting it to ISO6400 may be emphasising any tiny artefacts. If only I had more light....

Link to comment

does Spectraflect fluorescence at all? Boosting it to ISO6400 may be emphasising any tiny artefacts.

 

I think that depends on the purity of the material.

It might be enough with just a tiny bit of fluorescence in combination with the much higher sensitivity in IR.

Could be verified by adding a IR-pass filter on the lens. GG-420 or longer cutoff.

 

There was a discussion a few weeks ago when Cadmium demonstrated that PTFE has some fluorescence.

Link to comment

I think that depends on the purity of the material.

It might be enough with just a tiny bit of fluorescence in combination with the much higher sensitivity in IR.

Could be verified by adding a IR-pass filter on the lens. GG-420 or longer cutoff.

 

There was a discussion a few weeks ago when Cadmium demonstrated that PTFE has some fluorescence.

 

Cheers Ulf. I'll add it to the list - I think I may even have some of the GG420 in one of my camera bags.... To be honest it's a bit of an academic question as I wouldn't be trying to image that low down anyway, I only brought it up because I am a geek :).

Link to comment
Cheers Steve. Yes the lens is a Rayfact 105mm, so the new version of the UV Nikkor (the same internally as the original, but these ones are currently being made). I chose this one as I can use it on my Nikon and Canon cameras.
Link to comment

I'm not sure, but I don't think so Ulf, as the normal camera images went properly black at short wavelengths at ISO1600. If there was any significant light leak or fluorescence (in the visible region) I would expect to have seen it there. I suppose it could potentially be UV induced IR fluorescence on the monochrome camera images, but does Spectraflect fluorescence at all? Boosting it to ISO6400 may be emphasising any tiny artefacts. If only I had more light....

 

Do you have specs on the stray light ratio of the monochromator?

 

I appears to be a single, so it is likely on the order of 10-3 or -4

Link to comment

Hi John. Not to hand, will look when I get home tomorrow for the specs. It is a simple one, so I'm assuming is a single monochromator and you're probably right about the stray light ratio.

 

EDIT. Mines an older version and no longer available. However as I was searching around for info, there is a new version by Dynasil (which looks exactly the same as my one). My guess is they have changed the fixings and kept the inside the same. For the 200-800nm one that is currently for sale, which is the same wavelength range as mine, the stray light is quoted as being <0.003%.

Link to comment

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...