DIY SWIR/NIR source from a halogen lamp

[DaJMasta]

This forum has seen some SWIR and NIR threads, so I assume this is the best place to put it.  I designed a printable lamp housing for a reasonable power selectable band source made from a printed box, a tungsten halogen lamp, a longpass filter, and it seems to work well.  The box is sized and designed around a 10W, 12V halogen lamp - the heat output of halogens is high, so there are lots of ventilation slots, but they angle away from the front face with the filter, so a minimal amount of visible/out of band light will show on the front face target.  I used some reflectors found on ebay, but there is probably a good variety that will work, and reasonable gauge solid core wire makes for a reliable lamp mount that can be bent to focus against the reflector in back.  The optical filter can be selected for usage (a tungsten halogen lamp is a very broadband source), and is a standard 50x50x2.5mm size (2" square) that is often available on the used market.

https://www.thingiverse.com/thing:4758234



For the pictured box, I used a 2800k halogen lamp (lower color temperature means a larger portion of the total emission below the visible band), a reflector off ebay, and a Schott RG1000 filter (1000nm longpass).  The filter will likely be the majority of the cost of the whole thing.  I screwed on the reflector with M3 machine screws and nylon spacers, then I stripped the solid core wire and bent it approximately in place, securing in two places with zip ties, so that I could solder the lamp in place to the wires.  Then I made sure to clean the lamp with isopropyl alcohol, slid the filter into the front slot, and pressed the two lower halves together.  A couple of nuts and bolts, one on either side of the bottom, help keep the press fit parts hold the wires in place better, then you press fit the top piece on it.  The bottom has a 1/4" 20 TPI thread built in for a standard camera threaded mount.


(the room is much darker than the high gain of the visible camera makes it look)

The results are good!  You get stripes of unfiltered light facing to the side and back, but looking at the face of the light you only get what comes through the filter, I've got a set of cameras setup for multiband visualization and you can see that the RG1000 passes as expected - the NIR camera can see it in its reduced sensitivity region (sensitive to about 1150nm, I figure), whereas the SWIR camera sees a bright light.  You can see a tiny bit of UV through the slots, but I think being low color temperature and probably a standard glass lamp (rather than quartz glass), there is little UV emitted by the lamp used.

http://medpants.com/ee/SWIR%20Source%20SWIR%20only.jpg

(the plastic magnifying glass passes SWIR just fine!)

[Fraser]

Nice project 

Only one safety related comment though. Halogen lamps are well known for the high temperature at which they operate. Such is necessary for their correct operation to preserve the filament. As such care is needed when using Halogen lamps near plastics. My Tiertime UP! Box 3D printer uses fire retardant 3D printed parts for its extruder head covers because Tiertime warn that normal 3D ABS burns easily and is a potential fire risk. I know you are being careful and will not leave the lamp unattended whilst operating, but do keep an eye on it as if the top casing melts down into the Halogen lamp things could get ‘interesting’ .

Lamp houses for Halogen lamps used with Microscopes use metal, ceramic and heat resistant plastics, and for good reason.

Fraser

[Ultrapurple]

That's an excellent piece of work, and a very useful design for the community.

Thank you also for posting images made at so many different wavebands. I see I'm not the only person who has LWIR-to-UV capability!

Bearing in mind what Fraser said, is it worth considering adding some simple thermal monitoring or protection in the housing (eg a resettable thermal fuse)?

[DaJMasta]

Thermal fuse would be pretty easy to put up top, I left it going for a few hours and examined it with a thermal camera too as I figured that was the risk with it, and with a 10W bulb it seems to be fine - no softness or drooping in the top plastic and no obvious hotspots that are too high externally.  The center of the filter window has a similar hotspot, and as expected, the wires coming through the base of it, but at least provided that the lamp doesn't move from its set position it appears to be fine for operation at room temperature.  I wouldn't put it in a high vibration environment or anywhere particularly hot, though.

Aside from making it larger or having a thermal cutoff, I'm not sure what could be done to make it safer (I guess underdriving the lamp would do it too).  I had thought about looking for an IR reflective paint, as it may give a bit more output from the front window as well, but while I have the equipment to test paints for their performance, I don't know of ones that really advertise it and don't really feel like taking the time  Too many other projects on the way.

[bap2703]

Lamp houses for Halogen lamps used with Microscopes use metal, ceramic and heat resistant plastics, and for good reason.

Got one just next to me:


Interestingly the lamp is encased in a metal shield.
The vents are only there to cool the casing, but no external air flow is hitting the lamp itself.
That might also help with the temperature stability of the lamp .

[mjs]

Nice one, where did you buy the RG1000 and how much did it cost ? I should do one, too!

Regarding light output, the relative IR/VIS output is indeed higher at lower temperature, but the absolute IR output is higher at higher power/temperature. Of course the poor filter needs to absorb and turn into heat all that VIS part and needs to be cooled well.

Have you measured the output spectrum ? I did some work with a slide projector and there was a clear dip at 1450nm or thereabouts - probably due to OH or some similar absorption somewhere. Exactly where I would have needed the light for water absorption measurements..

[DaJMasta]

I think both I used, an RG1000 and a Corning glass 2540 which was supposed to be equivalent, I got in boxed sets of several filters, so thankfully, less than normal!  I think you can get a 2"x2" RG1000 on the used market for somewhere in the ballpark of $30 or $40, so not cheap but not the same tier as brand new.

I haven't properly measured the spectrum, I think I have a spectrometer that can get to the near end of the likely output spectrum, but not really into the SWIR band, so the best I've got is the SWIR response, really.  I will have a look in MWIR sometime soonish, though, would be interesting if I could actually see "illumination" in that band, but I've only got the filter datasheet to go off of so far.


I've got a monochromator with an output issue (I think PMT driver electronics), but maybe I can get around to fixing that up and seeing if it can reach that far down at some point.

[Fraser]

DajMasta,

Please be careful if you intend to expose your MWIR camera to the output of a Halogen lamp. I have been advised that such a light source can easily damage the MWIR detector array

Fraser

[DaJMasta]

The intention was to see if I could illuminate a scene with it, but noted I will be sure to be careful, haven't done much straight on illumination except for low power check sources at a distance with other cams.

[_Wim_]

I haven't properly measured the spectrum, I think I have a spectrometer that can get to the near end of the likely output spectrum, but not really into the SWIR band, so the best I've got is the SWIR response, really.  I will have a look in MWIR sometime soonish, though, would be interesting if I could actually see "illumination" in that band, but I've only got the filter datasheet to go off of so far.

Nice project! That quite a big range to measure, and I thinks performance will be pretty close to the theoretical performance by subtracting the filter response from the halogen bulb response. Attached the response of 2 typical long pass filters, and the response of a typical halogen bulb. So short wavelength behavior will be dominated by the filter, and the long wavelengths by the bulb type and partly by the filter also.

Did the ebay seller have more of those reflectors? They are a nice format to play around with by my Ebay search came empty handed.

[DaJMasta]

Not really cheap, but these are the ones I used and they do seem to be pretty well made: https://www.ebay.com/itm/271451806335

[_Wim_]

Not really cheap, but these are the ones I used and they do seem to be pretty well made: https://www.ebay.com/itm/271451806335

Thanks, that one did not show up as he is not shipping to Belgium (stupid Ebay search  ), but I sent the seller a message to see what is possible. 

[Ultrapurple]

...  (stupid Ebay search  ), ....

Have you ever tried searching for something on Amazon?

[DaJMasta]

Ugh, parametric style searching on amazon is as worthless as it gets.  Sellers can and do offer their items in whatever category and specification they like, even when it's wrong or when it overlaps hugely with a category in an entirely different section, but it doesn't matter either because amazon will suggest whatever's popular or promoted that's slightly related to your search terms.  You can search for something hyperspecific and if amazon understands at all what you may be looking for, most of the stuff that comes up is specifically not what you were looking for.

At least with ebay it doesn't add items with similar terms in the title to your search without any way to disable it, so if you pick the right term, you can filter things pretty well.

[_Wim_]

Have you ever tried searching for something on Amazon?

Yes, I do not do that any more. Ali-express is also not the best. I fail to understand how it is possible big companies cannot make such an important feature for their business better 

[_Wim_]

Not really cheap, but these are the ones I used and they do seem to be pretty well made: https://www.ebay.com/itm/271451806335

Just bought a couple! Seller changed his listing to worldwide shipping after my request. Indeed not the cheapest, but the tabs on the reflector make installation and alignment so much easier compared to the typical reflectors with only rounded surfaces. 

[Ultrapurple]

I fail to understand how it is possible big companies cannot make such an important feature for their business better 

Simples: if you search for "10k 1/4W 1% metal film resistor" they hope you'll also buy, on impulse, a tacky garden ornament, a book on perpetual motion techniques and a monthly subscription for tinned unicorn food.

The problem is, it does actually work (as my somewhat rotund pet unicorn will testify). 

[Ultrapurple]

Not to be confused with this:



(I bought one of these and gave it to my then-teenage daughter, who was very into unicorns - much hilarity ensued)

[lukier]

Nice project!

Could you tell a bit more about the cameras, especially SWIR and UV ones? These seem extremely rare.

[DaJMasta]

The SWIR camera has a couple threads about it floating around this section - an Electrophysics 7290 (though the older non-A version).  It was designed in the late 80s if I remember right, but uses a Vidicon tube, so it's all analog adjustments and you run into a bit of image retention on the tube and quite a bit of ghosting from the slow decay response of the phosphor used (lead sulfide).

The UV camera in this example is actually a standard camera - a Watec WAT-902H Ultimate - with the exception that it has a very high gain, low noise mode to see in very low light conditions.  The UV response of a standard CCD usually falls off pretty fast, but thanks to like 50dB of gain, you can get a usable image down to 300nm or so provided you filter the input sufficiently.  I'm using a Hoya U340 and Midwest Optical BP550 in tandem to get out all the visible and IR components in this case.

In proper UV cameras that can sense into the deep UV with reasonable sensitivity, there are some EMCCD and other specialty processes that can directly sense the short wavelengths, but most less expensive (still not cheap) options involve a standard back illuminated sensor with a coating of lumigen, so those deep UV wavelengths make the coating fluoresce and show up as an image.  Something like the ICX407BLA sensor is an example of that technique.

[Lambda]

Hello.

Just a short passage for saying thank you, DaJMasta, for this very interesting and instructive thread!

And just for illustrating your last paragraph, some interesting links on DIY UV viewer using the principle of down conversion UV -> Vis.

these DIY projects are using image intensifiers or conventional cameras (Mintron, by the way, close to your Watec, in a family "point of view"), both equipped with converting front opticals elements and UV pass/Vis blocking filter...

https://hackaday.com/2016/07/06/seeing-the-truly-invisible-with-diy-shortwave-uv-imaging/

http://uvirimaging.com/2016/07/03/diy-shortwave-uv-image-converters-for-solar-blind-and-ruvis-imaging/

Best regards.

Stéphane

[lukier]

The SWIR camera has a couple threads about it floating around this section - an Electrophysics 7290 (though the older non-A version).  It was designed in the late 80s if I remember right, but uses a Vidicon tube, so it's all analog adjustments and you run into a bit of image retention on the tube and quite a bit of ghosting from the slow decay response of the phosphor used (lead sulfide).

Ah very nice camera, pity these are so rare and expensive on ebay.

The UV camera in this example is actually a standard camera - a Watec WAT-902H Ultimate - with the exception that it has a very high gain, low noise mode to see in very low light conditions.  The UV response of a standard CCD usually falls off pretty fast, but thanks to like 50dB of gain, you can get a usable image down to 300nm or so provided you filter the input sufficiently.  I'm using a Hoya U340 and Midwest Optical BP550 in tandem to get out all the visible and IR components in this case.

In proper UV cameras that can sense into the deep UV with reasonable sensitivity, there are some EMCCD and other specialty processes that can directly sense the short wavelengths, but most less expensive (still not cheap) options involve a standard back illuminated sensor with a coating of lumigen, so those deep UV wavelengths make the coating fluoresce and show up as an image.  Something like the ICX407BLA sensor is an example of that technique.

Ah so just filters. I thought you have the latter approach. I like the lumigen coating idea, wonder if this could be done "at home" to some cheapo back illuminated sensor (like OmniBSI) - a bit like scintillators for x-ray.

Thanks!

[DaJMasta]

It very well may be, but it may be difficult to apply the right thickness as to prevent blocking visible light or blurring the image - I imagine the finish has to be pretty smooth to keep from scattering the UV when it hits it.  There are some other detector constructions that seem to be able to see UV, and just a back illuminated standard detector may give reasonable UVA performance.  Not like there are tons of things to image in the shorter wavelengths of UV as they're not really safe to be around the illumination sources and then you start generating ozone or just dissipating in open air as the wavelength gets shorter.

[_Wim_]

Ideal thread to show off my “new” tool I managed to get working today  . I bought this one on Ebay after seeing it listed for a couple of months with nobody buying it, I thought it was worth the gamble:

https://www.ebay.com/itm/CONTROL-DEVELOPMENT-OSC-SU256LX-1-7T1-250-9-1-7-3-13-100um-Spectrophotometer-/184614212682?hash=item2afbdd544a%3Ag%3A9-wAAOSw5SRfFcNt&nma=true&si=XDit031vmfz5Vfp3q9pQke6Mh2Y%253D&orig_cvip=true&nordt=true&rt=nc&_trksid=p2047675.l2557

I spent quite a while looking for the software & drivers, but found them in the end. Next challenge was to find a pc with an ISA slot and sufficient room for such a big card. Luckily they had some old industrial pc’s at work collecting dust,  so I brought one of these home. After installing window98 & VNC server, I can now remotely control this PC.

This card has a 256 peltier cooled ingaas array, allowing spectrum measurements from 900nm to 1700nm. My version is cooled to 0°C. Is was sold in 2001 for the “affordable” price of 9995$ (see flyer I found online in my search for the software).

As a first test, I thought I would measure some of the halogen sources I have:
-   12V 35watt MR11 halogen bulb from ali-express (https://nl.aliexpress.com/item/32891567396.html?spm=a2g0s.9042311.0.0.27424c4d7XRuq0 ) I bought these because this had no glass cover expecting less issues whir their response. I was wrong apparently… 
-   12V car halogen bulb.
-   Avalight HAL connected using a 400µm VIS/IR fiber
-   Light source that came with the control card connected using the same 400µm fiber

Black body calibration was performed using the light source that came with the card.

Very satisfied with my purchase! 

[lukier]

This card has a 256 peltier cooled ingaas array, allowing spectrum measurements from 900nm to 1700nm. My version is cooled to 0°C. Is was sold in 2001 for the “affordable” price of 9995$ (see flyer I found online in my search for the software).

Very interesting! I didn't know how it was done in the past, sounds expensive indeed. I do have some experience with more modern version:
https://www.element14.com/community/roadTestReviews/2322/l/dlp-nirscan-nano-evaluation-module-review
(this is also quite clever use of a DLP)

I do have some Ocean Optics S2000 spectrometers, but still haven't managed to find time to make USB interface for them 

-   Light source that came with the control card connected using the same 400µm fiber

Black body calibration was performed using the light source that came with the card.

Do you know what is the light source inside?

[DaJMasta]

I've used some later Ocean Optics spectrometers and you can probably get yours hooked up fairly easily.  I think SpectraSuite as well as the earlier OOIBase32, but I believe the S2000 may be able to talk directly to the right input card without the USB ADC - though the ADC units are available for sometimes inexpensive prices on ebay.  You also only need one ADC for several S2000s so long as you set one to master and stack the rest, though I don't know of any of those kinds of spectrometers that can go below 1200nm or so, so likely only the top of the SWIR band.

Interesting to see the deeper IR spectrophotometer, haven't looked specifically for them but interesting that it's basically the same construction, just with an InGaAs detector.  I actually have a NIR marked USB spectrometer lying around, but even after taking it apart to check the control board, I see no manufacturer's mark and the USB identifier is just "Spectrometer", so I haven't been able to verify what band it even measures in, let along interface with software.  I guess an option would be to open the top, put some broadband light into the input fiber, and take a look on the linear array what lights up in NIR and SWIR on camera!  Doesn't narrow it down a lot, but maybe I can find a narrower band filter to do some checking... hmmm

[_Wim_]

Very interesting! I didn't know how it was done in the past, sounds expensive indeed. I do have some experience with more modern version:
https://www.element14.com/community/roadTestReviews/2322/l/dlp-nirscan-nano-evaluation-module-review
(this is also quite clever use of a DLP)

I do have some Ocean Optics S2000 spectrometers, but still haven't managed to find time to make USB interface for them 

The DLP technology is indeed very interesting, I would expect it to perform very good, but have not come across it yet in my day job.

Last week we installed a commercial NIR system at work. It was based on a camera sensor instead of a line array, where the camera's sensor area was splitted into 16 channels, each channel having 256 pixels (do not know how many pixels in height, was not specified). Each channel had its own fiber routed from the probe head. So the system could take 16 spectrums at once from the container (pharmaceutical vials), and we could take 5 to 7 spectra per channel per vial when the machine was running at max speed. These were than combined in a complex model to measure moisture content.

I have been involved in a few NIR related projects at work in the last couple of years, but the typical suppliers are always quite vague about the actual performance of their systems, so it very hard to compare the older technology to the more recent. It does seems that 256 or 512 pixels is still the norm, I suspect this is because the higher resolution is not needed for most applications (spectral peaks and dips in the NIR region are quite wide typically).

Illumination was based on a halogen bulb light source, so even for this >100k€ system, that still seems the way to go.

Do you know what is the light source inside?

A halogen light bulb with some electronics for constant current or constant power.

[_Wim_]

I've used some later Ocean Optics spectrometers and you can probably get yours hooked up fairly easily.  I think SpectraSuite as well as the earlier OOIBase32, but I believe the S2000 may be able to talk directly to the right input card without the USB ADC - though the ADC units are available for sometimes inexpensive prices on ebay.  You also only need one ADC for several S2000s so long as you set one to master and stack the rest, though I don't know of any of those kinds of spectrometers that can go below 1200nm or so, so likely only the top of the SWIR band.

Interesting to see the deeper IR spectrophotometer, haven't looked specifically for them but interesting that it's basically the same construction, just with an InGaAs detector.  I actually have a NIR marked USB spectrometer lying around, but even after taking it apart to check the control board, I see no manufacturer's mark and the USB identifier is just "Spectrometer", so I haven't been able to verify what band it even measures in, let along interface with software.  I guess an option would be to open the top, put some broadband light into the input fiber, and take a look on the linear array what lights up in NIR and SWIR on camera!  Doesn't narrow it down a lot, but maybe I can find a narrower band filter to do some checking... hmmm

Can you post some pictures of the device, maybe somebody on this form recognizes it.

[lukier]

I've used some later Ocean Optics spectrometers and you can probably get yours hooked up fairly easily.  I think SpectraSuite as well as the earlier OOIBase32, but I believe the S2000 may be able to talk directly to the right input card without the USB ADC - though the ADC units are available for sometimes inexpensive prices on ebay.  You also only need one ADC for several S2000s so long as you set one to master and stack the rest, though I don't know of any of those kinds of spectrometers that can go below 1200nm or so, so likely only the top of the SWIR band.

Yeah these are for UV/VIS, for > 900nm I can use this NIRScanNano.

To be honest I gave up on the original stack from Ocean Optics, the USB adapters are rare, outdated and not that cheap. I have 4 x S2000 with one master (I assume, has more electronics populated) for which I've paid less than I've seen these USB-ADCs go. So my plan is to reverse engineer their electronics and make my own USB-ADC "piggyback" PCB. Still somewhere on the TODO list among other projects

A halogen light bulb with some electronics for constant current or constant power.

Ah I thought it would be more like Hg lamp or similar calibration source.

[_Wim_]

Ah I thought it would be more like Hg lamp or similar calibration source.

The lamp is not really a calibration source for wavelength like you would use a Hg source, it just used as an illumination source with a reasonably smooth response. At first I though that using the black body scale would just take into account the sensor sensitivity, but when a do a dark & white reference, it does seem to (slightly) change the response I get using the black body reference scale (flattens the peaks & dips).

I am still working my way through the manual, so maybe their is some more detailed explanation included.

[_Wim_]

The lamp is not really a calibration source for wavelength like you would use a Hg source, it just used as an illumination source with a reasonably smooth response. At first I though that using the black body scale would just take into account the sensor sensitivity, but when a do a dark & white reference, it does seem to (slightly) change the response I get using the black body reference scale (flattens the peaks & dips).

I am still working my way through the manual, so maybe their is some more detailed explanation included.

The black body calibration is done by entering the colour temperature of the light source used in the settings. Because a halogen bulb approaches an ideal black body, it is possible to linearize the sensor to measure "absolute" intensity. Currently this was set to 2900K (default of software), but after doing a comparison with my Avalight-HAL with know colour temperature (2730K in long life mode), it seems the internal light source is around 2800K.

Here is an easy to use calculator for colour temperature and peak wavelength:
http://hyperphysics.phy-astr.gsu.edu/hbase/wien.html#c3

[_Wim_]

As a first test, I thought I would measure some of the halogen sources I have:
-   12V 35watt MR11 halogen bulb from ali-express (https://nl.aliexpress.com/item/32891567396.html?spm=a2g0s.9042311.0.0.27424c4d7XRuq0 ) I bought these because this had no glass cover expecting less issues whir their response. I was wrong apparently… 

Today I retested the Ali-express halogen bulb, but this time I made a fiber optic coupler based on a plano-convex f=40mm collimating lens, followed by a bi-convex f=25mm focusing lens (focusing directly onto the SMA fiber). The response is much flatter than before (was exactly same bulb, both times powered by 12V).

I repeated the test by shining the bulb directly onto the spectrometer input, and I get the same wobbly response that remains quite consistent even when moving the bulb up and down. Not sure what is causing this effect.   

[bap2703]

Spectrometers usually requires you to input light within an acceptance cone.
It can be that you're not within that spec by shooting directly light in it. But when you use a fiber the propagation modes makes it for you.

[_Wim_]

Spectrometers usually requires you to input light within an acceptance cone.
It can be that you're not within that spec by shooting directly light in it. But when you use a fiber the propagation modes makes it for you.

Good point, but in this case I think this in not what is happening. Bulb can be held at a distance of 10cm giving still the same wobbly response (taking into account that an SMA feedthrough connector is fitted at the input, light from a bulb at 10 cm should be will within the NA of the input connector (I think).

Also I did not get the same result with the car bulb held at similar distances, so there has to be something related to this specific bulb (most probably related to the reflector).

[bap2703]

It could also be higher order diffraction.
Your fiber would filter out the low wavelengths.

[_Wim_]

It could also be higher order diffraction.
Your fiber would filter out the low wavelengths.

That could indeed be an option. I will try to do some more controlled testing this weekend by slowly varying the incident angle to see what is happening in detail.

[Ben321]

This forum has seen some SWIR and NIR threads, so I assume this is the best place to put it.  I designed a printable lamp housing for a reasonable power selectable band source made from a printed box, a tungsten halogen lamp, a longpass filter, and it seems to work well.  The box is sized and designed around a 10W, 12V halogen lamp - the heat output of halogens is high, so there are lots of ventilation slots, but they angle away from the front face with the filter, so a minimal amount of visible/out of band light will show on the front face target.  I used some reflectors found on ebay, but there is probably a good variety that will work, and reasonable gauge solid core wire makes for a reliable lamp mount that can be bent to focus against the reflector in back.  The optical filter can be selected for usage (a tungsten halogen lamp is a very broadband source), and is a standard 50x50x2.5mm size (2" square) that is often available on the used market.

https://www.thingiverse.com/thing:4758234



For the pictured box, I used a 2800k halogen lamp (lower color temperature means a larger portion of the total emission below the visible band), a reflector off ebay, and a Schott RG1000 filter (1000nm longpass).  The filter will likely be the majority of the cost of the whole thing.  I screwed on the reflector with M3 machine screws and nylon spacers, then I stripped the solid core wire and bent it approximately in place, securing in two places with zip ties, so that I could solder the lamp in place to the wires.  Then I made sure to clean the lamp with isopropyl alcohol, slid the filter into the front slot, and pressed the two lower halves together.  A couple of nuts and bolts, one on either side of the bottom, help keep the press fit parts hold the wires in place better, then you press fit the top piece on it.  The bottom has a 1/4" 20 TPI thread built in for a standard camera threaded mount.


(the room is much darker than the high gain of the visible camera makes it look)

The results are good!  You get stripes of unfiltered light facing to the side and back, but looking at the face of the light you only get what comes through the filter, I've got a set of cameras setup for multiband visualization and you can see that the RG1000 passes as expected - the NIR camera can see it in its reduced sensitivity region (sensitive to about 1150nm, I figure), whereas the SWIR camera sees a bright light.  You can see a tiny bit of UV through the slots, but I think being low color temperature and probably a standard glass lamp (rather than quartz glass), there is little UV emitted by the lamp used.

http://medpants.com/ee/SWIR%20Source%20SWIR%20only.jpg

(the plastic magnifying glass passes SWIR just fine!)

What kind of camera are you using for SWIR? I notice that your SWIR pic has a glow around the edge of the picture, making me think you somehow modified a normal NIR camera for SWIR use (with some much stronger NIR leaking around your SWIR-passing filter, and giving the edge that glow).

[DaJMasta]

That's an Electrophysics 7290 (non-A), some of the vignetting is from contrast enhancement done in the capture software, some of it is difficulty in chasing it out with the adjustment pots.  Since getting a 7290A working and comparing them, I think there's just some tube or component aging on the camera used in that pic, as it's not consistent with the other camera (that camera also has a lot more persistent image retention even though both have similar ghosting).

[Ben321]

That's an Electrophysics 7290 (non-A), some of the vignetting is from contrast enhancement done in the capture software, some of it is difficulty in chasing it out with the adjustment pots.  Since getting a 7290A working and comparing them, I think there's just some tube or component aging on the camera used in that pic, as it's not consistent with the other camera (that camera also has a lot more persistent image retention even though both have similar ghosting).

Where did you get your Micronviewer 7290? I can't find even ONE of them on ebay. I type in Electrophysics Micronviewer in the ebay product search, and it has 0 search results. Also what do you mean by adjustment pots? You mean potentiometers? Do you really need to adjust anything in the camera to make it work? Isn't it just a simple PIVO (power in, video out) camera?

Also, why is the SWIR image yellow in color? Isn't the output of that camera a B&W video signal, not color?

[DaJMasta]

I got mine on ebay in unknown condition and refurbished it, took a couple of new parts and chasing down some issues, but managed to get it going again.  Was some months ago, they don't come up that frequently.  Worth mentioning: I've never searched for them with the word "micronviewer", the term "electrophysics" usually is only 100-200 listings at a time, and for people who don't necessarily know much about what they're selling, sometimes only the brand and part number actually makes it into the listing.

If you look them up on the forums you'll see some picture of the adjustment potentiometers - there are a lot of them for various aspects of the video signal and tube driving and while yes, you power it up and there are minimal external settings, there are many internal ones that are subject to the usual drift and aging seen in older analog systems.  It took me some time to get a good looking picture, but I'm starting to get the feel for it.

I colorize the SWIR image to distinguish them when streaming them all together - I'm up to 6 cameras now and most of them are not in visible color bands, so most start as monochrome and then I use a post processing filter (the same one that is boosting the contrast in this image) to basically change the white color to a bright other color so they can be easily distinguished at a glance.  I chose violet for UV and red for NIR, an my preferred thermal palate is sort of blue/purple/red/orange, so yellow gave me a good difference.  The plan is to do green for MWIR and the depth camera I'm using has its own color scheme but should be arbitrarily configurable.  Long story short, I just took a picture from a layout used for other purposes, so they are colorized for that.

[Ben321]

I got mine on ebay in unknown condition and refurbished it, took a couple of new parts and chasing down some issues, but managed to get it going again.  Was some months ago, they don't come up that frequently.  Worth mentioning: I've never searched for them with the word "micronviewer", the term "electrophysics" usually is only 100-200 listings at a time, and for people who don't necessarily know much about what they're selling, sometimes only the brand and part number actually makes it into the listing.

These literally say Miconviewer the side. Even if the seller didn't know that was significant, they should include it. The point of posting on ebay is to get somebody to buy what you are selling. The more keywords used (such as Micronviewer) that are in the title of your product listing, the better the chance somebody looking for it will find it and buy it.

[_Wim_]

These literally say Miconviewer the side. Even if the seller didn't know that was significant, they should include it. The point of posting on ebay is to get somebody to buy what you are selling. The more keywords used (such as Micronviewer) that are in the title of your product listing, the better the chance somebody looking for it will find it and buy it.

But the worse the description is, the higher your chances are to get it for a reasonable price! So I kinda like the sellers who have no clue what they are selling...

[DaJMasta]

These literally say Miconviewer the side. Even if the seller didn't know that was significant, they should include it. The point of posting on ebay is to get somebody to buy what you are selling. The more keywords used (such as Micronviewer) that are in the title of your product listing, the better the chance somebody looking for it will find it and buy it.

Sure, I don't know if I wrote it before replying to you, though!  I actually had to look at the side of it to stop myself from saying that it wasn't worth searching searching for at all, because I think my search terms have always been "electrophysics" and browsing or "electrophysics 7290".  Maybe there's a better way for people to list and maybe there's a higher standard they should aspire to, but you're going to find things most frequently when you're willing to try some things and try to get into the head of someone who's just mass listing stuff they don't know about.

In any case, that is how I found mine, at least.

[_Wim_]

I will try to do some more controlled testing this weekend by slowly varying the incident angle to see what is happening in detail.

With a little delay (I got distracted here: https://www.eevblog.com/forum/thermal-imaging/anyone-have-an-estimated-emissivity-for-the-ivac-9000-radiators/new/?topicseen#new    )

I have done some more testing with this bulb, moving from left to right in front of the SMA input (bumps remain) and forward to backward (see below).

Test01 is with the actual bulb touching the SMA connector. Test02 is a couple of mm's backward, test 03 is another couple of mm's backward...

Conclusion:
As long as the bulb is close, the response of the bulb dominates, and we get the typical response of a halogen bulb. The farther we are away, the more the light of the reflector dominates, and this reflector is causing the bumped response.

So when you want a flat IR source, better not to use one of these bulbs with glass reflector.

[_Wim_]

I have ordered a couple of "normal" halogen bulbs (without reflector) , and also received the reflector identical to the one used by DaJMasta in the first post

When the bulbs arrive I will perform a similar test to check the response of this setup also.

[DaJMasta]

That is a really interesting bit of testing.  I assume the periodicity is either the faceted nature of it or even the microscopic texture of the reflector itself.  I don't have the equipment to sweep through and measure mine, but I assume you'll be able to see similar waves if it's the texture of the reflector (though these ebay ones seem fairly high polish compared to a normal halogen bulb).

It could be interesting to polish a bit of metal up and see if you can get an interference pattern on it just from the coarseness of the grit used.  I've got some stones I use for sharpening chisels and such with a final strop in the ballpark of 10,000 grit, which should be in the SWIR wavelength ballpark.... A diffraction grating shouldn't be doable, but I wonder if I could at least get a pattern in a collimated reflection off of it.

[_Wim_]

That is a really interesting bit of testing.  I assume the periodicity is either the faceted nature of it or even the microscopic texture of the reflector itself.  I don't have the equipment to sweep through and measure mine, but I assume you'll be able to see similar waves if it's the texture of the reflector (though these ebay ones seem fairly high polish compared to a normal halogen bulb).

It could be interesting to polish a bit of metal up and see if you can get an interference pattern on it just from the coarseness of the grit used.  I've got some stones I use for sharpening chisels and such with a final strop in the ballpark of 10,000 grit, which should be in the SWIR wavelength ballpark.... A diffraction grating shouldn't be doable, but I wonder if I could at least get a pattern in a collimated reflection off of it.

Am am not sure if the same would be visible with the metallic reflector.

I am thinking this phenomenon could be related to the thickness of the reflective coating or some other film layer (like a protective coating). I think this because of the smoothness of the interference, I would expect surface roughness related issues to have a more ragged response due to their randomness.

What I find strange is that the peak and dip wavelengths slightly vary with distance, but this could maybe be explained by the changing angle of the thin film vs the spectrometer, causing the apparent film thickness to slightly vary, and hence also the interference pattern. 

As there is approximately 250nm between 2 valley's, this would mean a film layer of only 125nm (see below for correct thickness) would be the culprit. This would also mean this effect must be present in the visible spectrum, as the "film" is thin enough to cause interference there also.

I will run a test tomorrow in the visible spectrum and see if this can be confirmed.     

[_Wim_]

I am thinking this phenomenon could be related to the thickness of the reflective coating or some other film layer (like a protective coating). I think this because of the smoothness of the interference, I would expect surface roughness related issues to have a more ragged response due to their randomness.

What I find strange is that the peak and dip wavelengths slightly vary with distance, but this could maybe be explained by the changing angle of the thin film vs the spectrometer, causing the apparent film thickness to slightly vary, and hence also the interference pattern. 

As there is approximately 250nm between 2 valley's, this would mean a film layer of only 125nm would be the culprit. This would also mean this effect must be present in the visible spectrum, as the "film" is thin enough to cause interference there also.

I will run a test tomorrow in the visible spectrum and see if this can be confirmed.     

Performed a test today. Test setup was slightly different due to the much high sensitivity of the Avaspec (was continuously saturated even with integration time at minimum).

Test setup:

- 400µm fiber with 30µm orrifice in front of the connector (to lower light intensity for the avaspec).
- distance between fiber and bulk +-8cm, fiber +-1cm to the side
- fiber connector to Avaspec and Control Devices Nir without moving the bulb
- exported data of both & scaled to similar units in Excel

Conclusion:
The bumps are also present in the visible spectrum. I was however incorrect about the expected thickness of the thin film, but correct that it is related to thin film interference. The wavelength dips match exactly with the theoretical ones.

Using this calculator (https://www.filmetrics.com/reflectance-calculator?wmin=200&wmax=1700&wstep=1&angle=0&pol=s&units=nm&mat[]=Air&d[]=0&mat[]=SiO2&d[]=1500&mat[]=Si&d[]=0&sptype=r) the film thickness should be around 1820nm

Update: a bit more precise estimation of the film thickness using the online calculator (for this distance. Estimated thickness is "apparent" thickness due to the angled sides of the reflector vs the fiber performing the measurement)

[bap2703]

Great analysis.
Although you don't really know the material of the coating so the thickness might be slightly different.

Just to continue digging: there's a very high frequency in your higher resolution curves. Any idea what that is ?

[_Wim_]

there's a very high frequency in your higher resolution curves. Any idea what that is ?

Noise I am afraid. Especially near the limits of its wavelength range the gain is way up to compensate for the insensitivity of the CCD array. Attached the spectral responce of the Hamamatsu S9840 that is used. As you can see, not much sensitivity in the NIR range.

[_Wim_]

Although you don't really know the material of the coating so the thickness might be slightly different.

I know. also due to the changing angle of the coating vs the probe absolute thickness determination is hard.

Attached another test I ran today, with a reflectance probe directly on the reflector. This setup should also be better to determine absolute thickness of the film...

[_Wim_]

Looking for info on thin film measurement I realized there is a plugin for my spectrometer to do it directly...

2040 nm according to that. But it is still quite fiddly, and normally you are required to take a reference spectrum without the film first (using the same substrate). I used a first surface mirror instead. 

Anyway, it is not like I "need" to know the thickness of the film, but it fun using the spectrometer for a different kind of application.

[Andy Perrin]

Hi, this is my first post on the forum (I didn't see anywhere to introduce myself, although maybe I missed the thread, since this forum is gigantic).

I found this post while searching for a MWIR light source for my Agema Thermovision 470 MWIR camera, but I also have a cooled germanium-on-CMOS SWIR cam, the NoblePeak TriWave camera, which works from 350-1600nm or so (give or take if you push it a bit).

On the topic of the post, I recently got an Arcadia Deep Heat Projector bulb, which puts out copious amounts of light in NIR, SWIR, and MWIR (roughly 1000nm to 2500nm or so is the range, but I don't have an exact spectrum). The bulb is for keeping reptiles warm, but it works nicely for illumination in those bands.

Greetings to the community, and if there's a better place to make an intro, do tell me!

[_Wim_]

Hi, this is my first post on the forum (I didn't see anywhere to introduce myself, although maybe I missed the thread, since this forum is gigantic).

I found this post while searching for a MWIR light source for my Agema Thermovision 470 MWIR camera, but I also have a cooled germanium-on-CMOS SWIR cam, the NoblePeak TriWave camera, which works from 350-1600nm or so (give or take if you push it a bit).

On the topic of the post, I recently got an Arcadia Deep Heat Projector bulb, which puts out copious amounts of light in NIR, SWIR, and MWIR (roughly 1000nm to 2500nm or so is the range, but I don't have an exact spectrum). The bulb is for keeping reptiles warm, but it works nicely for illumination in those bands.

Greetings to the community, and if there's a better place to make an intro, do tell me!

Welcome to the forum! That is some very nice toys you got there. I am sure you will fit right in.

Good choice of forum section to post is, from your list above, it seems you are only still missing an "LWIR thermal camera" to cover the full spectrum...

[Ben321]

I'd be more interested in a DIY solution for a SWIR camera.

[DaJMasta]

Well when you figure out an InGaAs FPA homebrew it may be feasible 


In seriousness, though, if you get a NIR enhanced standard camera, you can see just a bit into the SWIR band.  I haven't checked, but I've got a camera I bought for NIR use based on the newer IMX 462 sensor, and it's got good enough NIR performance that if I put a RG1000 filter in front of it, it would still see a little of the um range of shorter SWIR waves - InGaAs sensors often start around 900nm, and it certainly has no trouble with that.  Would take some experimentation, and will likely never see to 1.5um or so, but you could get a little in the right band.

[Andy Perrin]

Silicon can see up to 1200nm and that is often regarded as part of SWIR (nobody is quite sure where SWIR begins or ends anyhow...different people say different things). By 1200nm Silicon becomes transparent, so the photons just zip right through without being captured. It’s possible to improve the sensitivity up to slightly higher than 1200 by designing the chip to make the light bounce several times inside each pixel, improving the absorption. Conventional cameras are not designed this way, however.

[_Wim_]

Silicon can see up to 1200nm and that is often regarded as part of SWIR (nobody is quite sure where SWIR begins or ends anyhow...different people say different things).

I have done a test here with a normal industrial camera to see how far it "sees" into SWIR: https://www.eevblog.com/forum/thermal-imaging/electrophysics-micronviewer-7290a-heat-imaging-test-and-inherent-display-lag/msg3481036/#msg3481036

[_Wim_]

InGaAs sensors often start around 900nm, and it certainly has no trouble with that.  Would take some experimentation, and will likely never see to 1.5um or so, but you could get a little in the right band.

InGaAs based camera's all seem to fetch very high prices, I have not been able to find an affordable solution. InGaAs line array can sometimes be bought offordable on Ebay. With some sort of scanning mirror in front maybe a full image can be constructed from it. On my "todo" list, but too many projects currently...

[Ben321]

Well when you figure out an InGaAs FPA homebrew it may be feasible 

I was thinking more along the lines of motorized rotating hexagonal mirror in front of a pinhole and an InGaAs photo-diode. On the electronics side, a transimpedance amplifier for the photodiode, the output of which connected to my Picoscope USB oscilloscope. The computer then would be running my own custom software written with the Picoscope SDK, that actually pushes the scope to run faster than its rated maximum samplerate (up to 6.25 MSPS when it's rated for only up to 1 MSPS, and I have had some VERY GOOD results with this technique in other projects I've used it with). This data would not be displayed on the screen in the normal manner of oscilloscope software. Instead the raw samples would be saved directly to a file, for later viewing in a raw-image viewer type software (either the raw image data loader plugin for GIMP, or I could write my own raw image viewer software) to display the digitized signal from the photo-diode as an image.

Getting a pinhole isn't hard, but finding a motorized rotating mirror would be VERY hard. These are NOT off-the-shelf components. Every rotating mirror that I've EVER found in equipment I've disassembled, has been built SPECIFICALLY for the piece of equipment that I found it in. The rotating mirrors supermarket table-top barcode scanners are built for that specific piece of equipment. Same with the rotating mirrors in laser printers.

[DaJMasta]

Maybe something like a mirror galvanometer or pair of them would do?  They would be controllable (or variably oscillating), and if you're not concerned with a full-motion sort of framerate, they may be easier to deal with because you can configure the scanning yourself, though they are not so cheap with the required controllers.

Rotating mirrors may be difficult to find, but I think a lot of them could actually be usable if your form factor/placement/software is all configurable.  I thought about getting a flat mirror on a synchronized motor to make a linescan camera a full frame camera, and my trouble was trying to locate a mirror that was axially drilled for minimal image geometry processing - getting a suitable mirror you could attach to a collar for the motor would work, it would just need balancing and then some correction to the final image.  The technicalities of the optics are beyond my understanding, but you may also have trouble with lens convergence distance - unless you get a lens that gives you a fully collimated image to project on your mirror, you're going to need some careful focus adjustment or even lenses between mirror elements.

[_Wim_]

I was thinking more along the lines of motorized rotating hexagonal mirror in front of a pinhole and an InGaAs photo-diode. On the electronics side, a transimpedance amplifier for the photodiode, the output of which connected to my Picoscope USB oscilloscope.

Personally I would start from a linear array like this (https://www.benl.ebay.be/itm/284102443087), and move the array up and down itself behind a standard lens (so in essence imitating a rectangular sensor instead of a line sensor). This removes the whole complexity of the optics. These line arrays are relative easy to read out with a micro. Up down movement can be stepper controlled as it does not need to be that fast as your reading 256 or 512 pixels at a time.

[Ben321]

I was thinking more along the lines of motorized rotating hexagonal mirror in front of a pinhole and an InGaAs photo-diode. On the electronics side, a transimpedance amplifier for the photodiode, the output of which connected to my Picoscope USB oscilloscope.

Personally I would start from a linear array like this (https://www.benl.ebay.be/itm/284102443087), and move the array up and down itself behind a standard lens (so in essence imitating a rectangular sensor instead of a line sensor). This removes the whole complexity of the optics. These line arrays are relative easy to read out with a micro. Up down movement can be stepper controlled as it does not need to be that fast as your reading 256 or 512 pixels at a time.

What's the price of a linear array? A single InGaAs photodiode is already $80 from Edmund Optics. One place that was sellng a full rectangular array (suited for direct use in a camera) was selling that one InGaAs array chip (just the chip itself, and no optics or other electronics) for the huge price of $20000 (which is well over 90% of the cost of a fully functioning SWIR camera, so building one myself from such a component would not save much money). I can't imagine that a linear InGaAs array is not going to be cheap either. I'd guess it would still be somewhere around $800.

[_Wim_]

What's the price of a linear array? A single InGaAs photodiode is already $80 from Edmund Optics. One place that was sellng a full rectangular array (suited for direct use in a camera) was selling that one InGaAs array chip (just the chip itself, and no optics or other electronics) for the huge price of $20000 (which is well over 90% of the cost of a fully functioning SWIR camera, so building one myself from such a component would not save much money). I can't imagine that a linear InGaAs array is not going to be cheap either. I'd guess it would still be somewhere around $800.

New they are probably really expensive, but on Ebay they pop-up between 75 & 200$, which cannot be said for complete camera's unfortunately.

[ducatidragon916]

I know its been awhile on this topic, Have you had any luck with Converting the existing hardware on the S2000 Ocean Optics spectrum analyzers to USB etc? What if any information have you found on this? I know that they are boards that work on PCi slots which do the A/D conversions just wondering if you have had any luck reverse engineering on this?

[DaJMasta]

Not really in the scope of the original topic, but Ocean Optics made a USB adapter for the S2000, and a lot of the mainframes have it built in.  It's called the ADC1000-USB, and I think the original version was an ADC1000 with a second module but they eventually got integrated into a card.