1 of 2
1
Spectral Analysis of IR LEDs and Filters
Posted: 23 July 2009 10:03 AM   [ Ignore ]
Avatar
RankRank
Joined  2008-04-14
Total Posts:  148
Member

Hi guys smile

We’ve recently acquired a digital spectral analyzer so I thought we could test some of the LEDs and filter materials used on the forum. For most of the LEDs, the optical characteristics are usually specified by the producer but they are not always correct. In other cases the wavelength is slightly different or for homemade filters their optical properties are practically unknown.

First of all, the spectrometer is a USB 4000 VIS-NIR from Ocean Optics that has been previously calibrated. It has less than 1nm measurement error.

Items tested:

LEDs

SFH485
HSDL-4230
Environmental lights LEDs

Filters

(Bandpass) 850DF28 from omegabob at 0 angle
(Bandpass) 850DF28 from omegabob at 30 angle (for viewing the shifting effect)
(Longpass) Edmund Optics NT45-070
Negative film 1 layer
Negative film 2 layers
Floppy 1 layer
Floppy 2 layers
PS3 hotmirror

Reference lights

Neon lamp
Halogen light bulb

Conclusions:

Environmental LEDs are not at exactly 850nm CW. It’s around 845. It could be interesting to see a 845BP10 filter in action.

SFH485 and HSDL-4230 have a much wider emission spectrum than ENv. Lights LEDS. This might be useful since the shifting effect moves the central wavelength of a good filter by more than 30nm. For Env. Lights LEDS, a narrower emission might cause problems with wide lenses.

The hot mirror of the PS3 (and most other cameras) is very efficient at what it does- removing IR radiation. I managed to salvage one intact from the lens assembly and after 700nm it’s almost dark.

Negative films either in 1 or 2 layers are longpass filters rather than bandpass. 2 layers are better than one since they cut most of the visible radiation but the intensity of the transmitted radiation decreases.

Floppy disks are extremely poor in visible filtering applications. Not only they decrease the transmission efficiency but they do a lousy job at cutting visible radiation. I included the spectral analysis of 1 and 2 layers but 2 layers are almost opaque.

850DF28 bandpass filter works great as any other professional filter. I also tested the shifting effect by lighting the filter at a 30 degree angle. It’s easily seen how much the center wavelength is modified.

Neon lamps emit very little in infrared. I didn’t have an energy efficient light bulb to test it but I will after the well-deserved leave rasberry. I also included a normal tungsten lamp to test the filters and to prove how much infrared radiation is emitted by those lamps.

I tested just what I had lying around. In time I’ll test more and if anyone is unsure about what they have they can send me a sample (either a LED, filter, any other material) and I’ll perform the spectral analysis.

P.S. Since version 2 of the community book will include more info, perhaps it would be a good thing to include some of the data from the spectrometer for home-made filters or LEDs so that the readers can understand more about the physical part of what is happening and for choosing the right solution.

Print screens of the spectral data to follow:

A normal tungsten light bulb. It gives a lot of infrared radiation

xlightbulb.jpg

A neon light for reference. Almost zero IR output

xneon.jpg

A LED from Environmental Lights. Not quite 850nm but close enough

xenvironmental.jpg

SFH485, a common solution for FTIR
xsfh485.jpg

HSDL-4230. More expensive than SFH485 but with better spectrum emission and similar optical power (good for a wide filter, less good for a narrow one - wavlength wise)

xhdsl-4230.jpg

The hot-mirror filter of the PS3 webcam. Excellent IR cutting

xhotmirror.jpg

A long-pass filter for reference (Edmund optics NT45-070). It cuts everything below 650 but lets all the NIR sub-spectrum in.

x700longpass.jpg

Negative film 1 layer. More of a longpass than a bandpass, still pretty good results.

x1negative.jpg

Negative film 2 layers. Better filtering but less transmission.

x2negative.jpg

Floppy disk 1 layer. Low transmission and bad filtering characteristics. It does nothing to prevent visible light

x1floppy.jpg

Floppy disk 2 layers. Almost opaque, not a good option

x2floppy.jpg

850DF28. Great filter, a real bandpass solution

x850df28.jpg

850DF28 at 30 degrees angle. The shifting effect so many people have encountered with LLP is clearly seen here

x850df28-30.jpg

Image Attachments
x1floppy.jpgx1negative.jpgx2floppy.jpgx2negative.jpgx700longpass.jpgx850df28-30.jpgx850df28.jpgxenvironmental.jpgxhdsl-4230.jpgxhotmirror.jpgxlightbulb.jpgxneon.jpgxsfh485.jpg
 Signature 

VISION DECK, FTIR MULTI-TOUCH SYSTEM
Spectral analysis of Multi-touch IR LEDs and FILTERS

Profile
 
 
Posted: 23 July 2009 11:24 AM   [ Ignore ]   [ # 1 ]
Avatar
RankRankRankRank
Joined  2008-02-12
Total Posts:  1120
Member

Great work… This is going to be very helpful thanks smile

 Signature 

Company Website - Knode Research and Development
My Youtube

Profile
 
 
Posted: 23 July 2009 11:35 AM   [ Ignore ]   [ # 2 ]
Avatar
RankRankRankRankRankRank
Joined  2008-09-12
Total Posts:  2225
Dedicated

Awesome work man.  Can you explain and possibly link the omegabob filter you are referring to in the 850DF28.  From you graphs it seems like that filter would or would not be a good match up with the 850nm LEDs from environmentallights?  Also, this “shifting effect”, I know you said at 30 degree, can you explain a little about what you mean by that?  And you said you’d see it in an LLP setup, but what about an FTIR setup, etc?

 Signature 

MT Website - A Collection of the Most Relevant Multitouch Info From These Forums
Multitouch Blog
Multitouch Parts Store - Buy Infrared Touch Frames (No Camera Needed!), Complete Modified Cameras and DIY Upgrade Kits, m12/CS Mounts, m12/CS Lenses, Infrared Lasers, LEDs, etc

Profile
 
 
Posted: 23 July 2009 12:00 PM   [ Ignore ]   [ # 3 ]
Avatar
RankRankRankRank
Joined  2008-02-12
Total Posts:  1120
Member

The shifting effect is pretty simple, since most BP filters that are bought are interference filters, they shift their CWL based on the angle of incoming light. Traditionally people have installed there filter as such
?ACT=28&fid=34&aid=1245_PZz29EgfbWlLVpqvKT69
with the filter in front of the lens, what happens is the light entering the edges of the lens comes at a very steep angle. As a result the center CWL can shift.

The problem is with cutoff frequencies if they are too narrow, interference filters that are placed above the lens undergo the a similar phenomenon such as the blue shift in our case IR Shift as the angle of the light gets farther from the zero angle of instance.  Essentially what happens is that the the center of what you are focusing on gets excellent pass but becomes poor as you move toward the edges. 

The first thing you need to do is calculate the FOV angle of what your are focusing to determine how much blue shift you are going to get off center.  Then you can select a CWL that is higher than the wavelength of your LED’s or lasers.  The red shift will compensate for the blue shift and provide a good pas,s near the edges of your table.  However, as a result of increasing the CWL have to increase the FWHM to compensate for the shift that occurs in the CWL.  Otherwise the center of your table will have poor pass.

E.g. You use 850nm lasers and a webcam with a wide angle lens.  Assuming a 40nm blue shift at the edges of the screen, you might select a bandpass filter with an 870nm CWL because at the edges of the screen the CWL of the filter effectively 830nm.  As such you would then need to select an FWHM of around 50 to give an effective range at the center of the table between 845 and 895 nm and 805 and 855 nm at the edges.

Some concerns are that there might be inconsistent intensity at different angles. This is due to the nature of the interference filter, this could addressed by increasing the luminosity. Other steps to improve the blobs is to decrease the FOV of the camera lens although this requires you to have the camera farther from the display.

In any case these problems will not be faced if the Interference filter is placed behind the lens and in front of the CCD.

Image Attachments
ACT=28&fid=34&aid=1245_PZz29EgfbWlLVpqvKT69.gif
 Signature 

Company Website - Knode Research and Development
My Youtube

Profile
 
 
Posted: 23 July 2009 12:12 PM   [ Ignore ]   [ # 4 ]
Avatar
RankRankRankRankRankRank
Joined  2008-09-12
Total Posts:  2225
Dedicated

Hmm, ok, thanks for the explanation Taha, so then why does a 2.1mm wide angle (fisheye) lens on a camera setup with the filter between the lens and the sensor produce exactly the shift that you are talking about: light passes good in the center, but the light at the edges isn’t as bright and thus in these MT setups, blobs don’t show up as well.

 Signature 

MT Website - A Collection of the Most Relevant Multitouch Info From These Forums
Multitouch Blog
Multitouch Parts Store - Buy Infrared Touch Frames (No Camera Needed!), Complete Modified Cameras and DIY Upgrade Kits, m12/CS Mounts, m12/CS Lenses, Infrared Lasers, LEDs, etc

Profile
 
 
Posted: 23 July 2009 12:42 PM   [ Ignore ]   [ # 5 ]
Avatar
RankRankRankRank
Joined  2008-02-12
Total Posts:  1120
Member

Well basics of optics and lenses tell you when light passes through a lens it doesn’t come out the other side at 0deg it converges so there will always be some blue shift occurring but you compare this to having the filter in front of the lens this is very minimal. There is no ideal filter for MT but you can easily get by with having the in between the CCD and Lens.

 Signature 

Company Website - Knode Research and Development
My Youtube

Profile
 
 
Posted: 23 July 2009 01:22 PM   [ Ignore ]   [ # 6 ]
Avatar
RankRankRankRankRankRank
Joined  2007-04-08
Total Posts:  2539
Dedicated

Really great madian. So based on what you said above, is it better to have a slightly higher intensity and smaller width or a slightly lower intensity and much wider width (ie. SFH485) ?

 Signature 

MTmini, MTbiggie, & Audiotouch creator & Community Core Vision Co-founder

Follow on:
My Blog | Facebook | Twitter | Youtube

Profile
 
 
Posted: 23 July 2009 04:34 PM   [ Ignore ]   [ # 7 ]
Avatar
RankRankRankRank
Joined  2008-06-04
Total Posts:  565
Elite

Very cool, man i want a spectrum analyzer for Xmas!

Can you also test some lasers if you get a chance.  780 and 850 seem to be the predominant wavelengths i think.  Also can you test some different projection materials for IR transmission values.  I may be able to send you some Rosco, Frostings Etc.

Theres loads a stuff you could test very cool.

 Signature 

}---)))@

V JAY!  http://www.v-lux.net LASER tut http://nuigroup.com/forums/viewthread/6407/

Architectural Technician and Building Surveys

http://www.jpgainsfordassociates.com

Profile
 
 
Posted: 24 July 2009 01:12 AM   [ Ignore ]   [ # 8 ]
Avatar
RankRank
Joined  2008-04-14
Total Posts:  148
Member

@ peau

the link for the 850DF28 is this one http://cgi.ebay.com/Optical-Filter-850DF28-w-25mm-diameter-metal-ring_W0QQitemZ310134645920QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item48357724a0&_trksid=p3911.c0.m14&_trkparms=65:12|66:2|39:1|72:1240|293:1|294:50
It was a good match for Env. Lights LEDs but it’s not that good paired with the PS3 eye. Not that much because of the shifting effect but because it’s a 25.4 mm diameter filter and with the wide ps3 lens I get some barrel effect on the corners. However, it provided better results in terms of intensity than the 850DF10 we’re using now because as you can see in the spectral data the Env. Leds are at around 845, have less spectrum spread and the 10nm filter cuts too much. Still pretty usable though. Taha did a great job at explaining the shifting effect so I’ll not cover that. Here: http://nuigroup.com/forums/viewthread/2478/ the shifting effect and the problems associated with it are explained in detail and also why LLP users have more problems than FTIR users.

@Seth
Tricky question rasberry I would say, based on my experience that a larger spectrum spread works always better. Even if the intensity is less, you get more uniform lighting on all the parts of the table because shifting or not the LED will always have sufficient optical power to cover 40nm +/- CWL modification. I chose Env. Leds because of the simplicity of the ribbon but a larger emission spectrum would have been nicer. Plus, since they are not exactly 850nm this complicates a bit the choice of filters.

@ajlovegroove

Well, the prices have dropped alot, THe spectrometer+software+ handling and shipping was around 3000$. A few years ago you couldn’t get near a store without 10k. And it’s always better when you can buy it without using your wallet smile. Unfortunately I don’t have any IR lasers in the lab since I deal with green and blue lasers mostly. I plan to purchase some soon I hope. I have some diffusion and projection materials such as rosco and some ir foils, good point, I’ll test them too. If you have something else that you need tested I require as little as 1-2 cm round or square samples. I can;t give you exact absorption and transmission values because spectrometers only analyze this data as a comparison (percentage). I’ll set up a test chamber so I’ll always have the same radiation intensity and than absorption and filtering properties will be tested better.

I’m going on leave tomorrow and I will return mid -August. If you have any other ideas just write here and I’ll test everything I can. Spectrometers are really nice to play with smile

 Signature 

VISION DECK, FTIR MULTI-TOUCH SYSTEM
Spectral analysis of Multi-touch IR LEDs and FILTERS

Profile
 
 
Posted: 24 July 2009 02:28 AM   [ Ignore ]   [ # 9 ]
Avatar
RankRankRankRank
Joined  2008-06-04
Total Posts:  565
Elite

@ madian

There are a few window foils, frostings and tints that would be great to test.  Once i get the samples from manufacturers i am trying to fake the Vikuiti effect although i dont have the ability to micro glass bead the foil!  Because i am building up layers of film the IR transmission may be significantly reduced?

Thanks

 Signature 

}---)))@

V JAY!  http://www.v-lux.net LASER tut http://nuigroup.com/forums/viewthread/6407/

Architectural Technician and Building Surveys

http://www.jpgainsfordassociates.com

Profile
 
 
Posted: 17 September 2009 04:49 AM   [ Ignore ]   [ # 10 ]
Rank
Joined  2009-06-18
Total Posts:  12
New Member

BPF_650A.jpg

BPF 650 filter for LED650

http://www.accuteoptical.com

 Signature 

http://www.accuteoptical.com

Profile
 
 
Posted: 02 January 2010 04:18 AM   [ Ignore ]   [ # 11 ]
Avatar
Rank
Joined  2009-12-29
Total Posts:  27
New Member

Thanks for doing all this great work.  It’s nice to see some real measurements of filters.

However and forgive me if this giant post is irrelevant.  I’ll be happy to edit or remove if anyone is bothered by it…
I don’t know what is being referred to as a neon bulb here.  I think perhaps people are misusing this term.  A NEON bulb is a gas-fluorescent bulb containing Neon gas - Neon is yellow.  Hydrogen reddish, argon purple etc.  they have uniqe colors and spectra.  They can also be predicted solely from their electron configurations (which is why Hydrogen has so few wavelengths it emits with only a few electron states compared to say Xenon with a zillion).

A compact fluorescent bulb is not a neon bulb.  It contains mercury gas (or it becomes a gas when electrified a t least).

There is no need to do a Spectral Analysis of any of these bulbs.  It is not possible to change the few wavelengths of light that they give off.  Regardless the further you go into infrared the less likely you are to detect anything at all (far infrared is so low energy it can hardly vibrate a molecule never mind break a bond or make a detector respond which is why we don’t do emission spectral analysis for infrared, but rather do absorption).  Each gas gives off very narrow ranges of light that they will emit.  Mixing gases just adds these narrow “peaks”.

These values can easily be looked up.  Basically the only light source that has a significant infrared emission is Incandescent lighting, possibly Xenon as well.  The rest of the gas fluorescent lighting (Hydrogen, Neon, Argon, sodium mercury) that you might come across are not significant emitters of IR.

Below I have included images of the spectra of neon, mercury,
They can be found here also
http://physics.nist.gov/cgi-bin/ASD/lines1.pl

Xenon (a peak at 828 and 875) possibly a decent IR emitter.  This is the bulb in a Strobe light.
xenon.gif

Mercury (compact fluorescent) - no infrared emissions (note these bulbs run rather cold - Compact fluorescent not Mercury Arc that is)
mercury.gif

Typical Fluroescent Bulb mix (largely mercury) which I found.  It looks like this particular photo is shifted a bit.
Fluorescent_lighting_spectrum_peaks_labelled.gif

Neon - longest wavelength peak is at 725 probably no infrared.
Neg.jpg
Sodium - distinctly yellow glow of new street lights
Nag.jpg
Sometimes Neon bulbs are mixed with Hydrogen (which basically gives you a single wavelength of 525 with all the other peaks being tiny see http://www.inoe.ro/JOAM/pdf6_2/Musa.pdf (musa et al Journal of Optoelectronics and Advanced Materials Vol. 6, No. 2, June 2004, p. 459 - 464)

summary figure
spectra.gif

Image Attachments
neon.jpgNe-Ar-H2.jpg
Profile
 
 
Posted: 02 January 2010 04:32 AM   [ Ignore ]   [ # 12 ]
Avatar
Rank
Joined  2009-12-29
Total Posts:  27
New Member

Also, it would be nice if these might be put on the same scale.  People would get a better idea of how to compare these side to side.  Or how narrow the range of light an LED actually puts out when put on the full scale etc.

Thanks again for such great work!

Profile
 
 
Posted: 02 January 2010 05:35 AM   [ Ignore ]   [ # 13 ]
Avatar
RankRank
Joined  2008-04-14
Total Posts:  148
Member

@ Flyingpumpkin:

Thank you for the great information you specified here smile . No post is irrelevant and please do not change it smile. I have to clarify a few points though:

I put the neon lighting as a reference because I’ve seen lots of questions on the forum regarding “Will my device work in plain light?” and most people do not have the scientific background (at first) to distinguish between various forms of lighting and their infrared emission. And since the popular term was “neon lighting” I just included the spectrum to show our NUI friends that it is perfectly safe to use a table under this lighting condition BECAUSE the infrared output of such a tube/bulb is null or neglectable. It was meant as a comparison between incandescent lighting and typical gas lighting to show the infrared emission. You are right though and I should have specified which type of tube I was testing (TO-DO List rasberry).

Regarding your second post I plan to update my analysis to include uniform scale/domain measurement and to include some new lasers and materials I have tested but time is never on my side… I chose a narrow scale for LEDs because most people have narrow band filters and a “zoom” on the scale allowed them to understand the decrease of intensity around the peak output.

 Signature 

VISION DECK, FTIR MULTI-TOUCH SYSTEM
Spectral analysis of Multi-touch IR LEDs and FILTERS

Profile
 
 
Posted: 02 January 2010 02:15 PM   [ Ignore ]   [ # 14 ]
Avatar
Rank
Joined  2009-12-29
Total Posts:  27
New Member

Spectral Analyzer - how great.

I am considering using Rosco Gel filters (which are extremely cheap and easy to find here in New York) as my IR transmission/Visible blocking filter for my camera.  You can buy a sample sheet for about 50 cents. They also come in large sheets and rolls as well.

Woud you be able, at some point to look at some of these in the IR spectrum? [/b]I am not sure if this is at all possible.

There is good data on them from Rosco up to about 740 nm.  I am assuming most of these Gel filters are transparent to UV, but it would be nice to know 100%
I am going to try them out empirically with my camera to see which look transparent on IR.  I think this is a good option for some folks.

Spectra can be seen and gels ordered here: http://www.rosco.com/us/filters/roscolux.asp

I am planning to use #95 blue green or #384 midnight blue (these pass through infrared) and then combine with Medium red (to block out the remaining visible wavelengths.  This should give me wavelengths above 720 only and act as a longpass filter.

95.jpg384.jpg27.jpg
COMBINED transmission spectra :?ACT=28&fid=34&aid=4994_ukdVcjuLwuiL47P9ORmj

You can find these great pictures of Gel filters in visible light (where the combo is effectively black) compared to the same gels viewed by an infrared camera here:
http://softsolder.wordpress.com/2009/03/27/quick-and-easy-ir-passing-visible-blocking-optical-filter/
dsc00235-gel-filters-full-color.jpgdsc00236-pure-ir-view.jpg

and of course Lee sells an infrared filter for $26 but typical of Lee - no spectral data available.

Image Attachments
COMBO2.jpg
Profile
 
 
Posted: 04 January 2010 03:29 AM   [ Ignore ]   [ # 15 ]
Avatar
Rank
Joined  2009-12-29
Total Posts:  27
New Member

For those of you in the UK you can also get 3 samples of Perspex IR transmitting acrylite. 
It seems that all lucite/acrylite/acrylic/perspex is infrared transparent.  If you buy the black then you have a good visible light filter.  Should be pretty cheap as well.
http://www.theplasticshop.co.uk/sample-request-3360-0.html

UV transmission of special IR transmitting plexi:
media.nl?id=6279&c=621686&h=995dc7615bb5b6774b1e

Profile
 
 
   
1 of 2
1