Sciencemadness Discussion Board

Cheap, Low-Resolution, Raman Spectroscopy

 Pages:  1    3

radagast - 18-2-2013 at 18:11

Amateur chemists need a way to identify and characterize compounds, and buying analytical equipment on Ebay or Labx is an expensive gamble. Because I lacked the knowledge to fix an FTIR (as len1 did), I turned to raman spectroscopy for a possible solution, based on the article by Mohr et al. describing a < $5000 dollar Raman spectrometer constructed from an Ocean Optics spectrometer, a 532 nm Melles-Griot laser pointer, and an expensive notch filter):

http://www.sciencemadness.org/scipics/ed800081t.pdf

and here:

www.sciencemadness.org/scipics/si002.pdf

(Note to mods: there's an old related thread at http://www.sciencemadness.org/talk/viewthread.php?tid=13332 but it's lengthy and unwieldy so I decided to start a fresh one).

The Mohr et al. article is terrific, but the spectroscope would still cost > $2000, and if you willing to spend a thousand more, you could just buy a raman spec from StellerNet or another company. So I tried to drive down the cost a bit through the following modifications:

(1) For a spectrometer, I used a $250 uncalibrated DIY spectrometer, 1800 lines/mm grating, sold by Science Surplus, which I calibrated using a neon indicator lamp from Radio Shack:

http://www.science-surplus.com/products/spectrometers

These spectrometers occasionally appear on Ebay, although you have to make sure you get the grating type you want. I first got the idea of using this spectrometer from Mikko, a gemologist who used it to construct a rudimentary raman spec capable of distinguishing diamond from cubic zircon.

http://gemologyonline.com/Forum/phpBB2/viewtopic.php?f=11&am...

(2) For a 532 nm laser source, I used the following ~$25 532 nm laser pointer from Amazon, powered by a D.C. power supply set at 3.0 volts:

http://www.amazon.com/532nm-Military-Power-Green-Pointer/dp/...

http://www.amazon.com/Philmore-Multi-Voltage-Regulated-Power...

(3) I substituted an inexpensive ~$50 edge filter for the $500 notch filter used by Mohr et al. (<i>see, e.g.</i> http://www.ebay.com/itm/Optical-Filter-541-AELP-25mm-Raman-E... and

(4) Instead of using a beam-splitter made from a microscope slide, I simply shot the laser beam at a 90 degree angle from the spectrometer collection tube. (The picture below is misleading because it shows my spectrometer in a backscattering configuration).

(5) For an enclosure, I simply chucked the apparatus into a plastic toolbox from Home Depot. I think this may partly account for the interfering florescence discussed below, but it seemed to work in a pinch.

I tested the assembled “ghetto raman spectrometer” on a tablet of CVS aspirin, using three seconds of integration time for each scan, and the average of 200 scans. I then imported the data into R, cut off everything below 540 nm, and plotted the result (please ignore the axis labels, as they're inaccurate). Although the resulting spectra has some artifacts, <i>see</i> attached PDF – possibly due to fluorescence or the filler or coating in the tablet? – if you squint and use some imagination, it has the same dominant peaks as the spectra captured by a professional instrument (<i>see</i> Horiba Spectra).

Anyway, this is very much a work in progress -- I'm a beginner at spectroscopy and am slowly learning the basics – and I welcome all comments on how to improve the setup. I plan to build a beam-splitter and compare the results.

Attachment: Aspirin GKau Raman Spectra.pdf (15kB)
This file has been downloaded 1633 times

Horiba Aspirin Raman Spectra.jpg - 8kB IMG-20130218-00079_reduced.jpg - 205kB

[Edited on 19-2-2013 by radagast]

[Edited on 19-2-2013 by radagast]

watson.fawkes - 19-2-2013 at 06:53

Quote: Originally posted by radagast  
(Note to mods: there's an old related thread at http://www.sciencemadness.org/talk/viewthread.php?tid=13332 but it's lengthy and unwieldy so I decided to start a fresh one).
Good call. That thread was lots of talk and little action. This is a much better start.

Perhaps the best part of this design for most people is that all the electronics for the spectrometer came as a single unit, prepackaged. Not having to work out CCD electronics certainly removes a major component of skill required to build one.
Quote: Originally posted by radagast  
(3) I substituted an inexpensive ~$50 edge filter for the $500 notch filter used by Mohr et al.
[...]
(5) For an enclosure, I simply chucked the apparatus into a plastic toolbox from Home Depot. I think this may partly account for the interfering florescence discussed below, but it seemed to work in a pinch.
The notch filter acts as a monochromator. The same effect can be accomplished with a prism and a slit, but it takes more distance between prism and slit to accomplish the same thing. The filters are far more compact, something certainly worth paying for in a professional context, but perhaps something the amateur could live with and save money. The "slit" only needs to be single-sided to mimic the filter behavior of the edge filter. It can be easily built by stacking together a few razor blades; the shallow grind angle makes an effective beam trap with little backscatter. I mention this because while you found a cheap filter on eBay, they might not always be available to others.

As for the enclosure, I'd guess that backscatter from the "fourth" ray off the mirror might be your dominant source of the spurious signal you mentioned. Putting a mirror pointed at the ceiling should give you a quick indication if this is so, or a piece of black cloth, or whatever. If indeed that's the case, you could put a beam trap there (again, constructed of stacked blades) to remove it.

One question. You didn't mention what kind of optical table you used, nor the mounts on it.

And a request. I have found it quite useful for other people's projects to see a full bill of materials and prices paid for them. Since there's always some luck in finding inexpensive replacements for otherwise-commercial components, it's useful to see a complete bill, as a way of estimating one's own budget and effort required to save cash.

radagast - 19-2-2013 at 11:49

Thanks for the very helpful comments, Watson. To clarify re: your advice on the backscatter from the "fourth" ray off the mirror, are you referring to possible reflection from the silver plastic cover when the enclosure is closed during the scanning?

Building a monochromator out of a prism and razor blades is intriguing, and something I'd like to pursue after I get everything working.

To build the infrastructure of the spectrometer, I used a used 8"x8" Thorlab breadboard, and ordered optical posts and post-holders from Thorlabs.

I've attached a very rough draft spreadsheet of the parts required to build this spectrometer, in three configurations:

(1) a "barebones" 90 degree configuration, including the requisite lens tubes, screws, and filters (which I'd consider too much trouble to machine), but excluding the breadboard, optical posts, etc. which could be made from wood -- c.a. ~$500 USD

(2) a standard 90 degree configuration, which includes the Thorlabs breadboard, optical posts, etc. -- c.a. ~$800 USD

(3) a standard beam-splitter configuration, which includes the Thorlabs breadboard, optical posts, etc. -- c.a. ~$1000 USD

This list isn't 100% accurate, since I've assembled it partly from memory, but I think those prices should be in the right ballpark. It assumes that one has a Windows XP computer with a serial cable, and that you buy all the components new.

EDIT: I've adjusted the baseline a bit using the "baseline" package in R. Also, the price estimates above are too high by $50; the DIY spectrometer from Science Surplus was actually $200, not $250.

Attachment: Raman Parts List v.1.pdf (116kB)
This file has been downloaded 2607 times

Baseline_Corrected_Raman.JPG - 54kB

[Edited on 20-2-2013 by radagast]

radagast - 22-2-2013 at 12:06

A word of caution to anyone using a direct D.C. power supply to power the 5 mW green "laser pointer" like the one I listed above: I'd bet my life that once you connect the D.C. power supply to the laser, the laser emits substantially more power than 5 mW at 532 nm (and possibly at higher IR modes, too).

Frankly, I had my doubts that the laser pointer only emitted 5 mW at 532 nm even with the supplied lithium battery, but connecting it to a wall-powered power supply really unleashes the beast within. The green reflection from irradiating a white aspirin tablet is so bright that it's uncomfortable to view from even 10 feet away in a darkened atmosphere.

532 nm protective glasses (preferably with IR blocking as well) are required for this type of project.

Backscattering Configuration

radagast - 23-2-2013 at 16:45

A quick Saturday-evening update before I begin drinking . . .

I've found that when using the 90 degree configuration to measure solid samples, the precise location of the laser beam on the sample can affect the intensity of the raman shift. I got some better spectra after my Thorlabs OG10 safety glasses arrived, because the glasses permitted me to aim the laser better.

I got substantially better results, however, when I switched to the back-scattering configuration (as designed and described by Mohr et. al). The back-scattering configuration is simply a beam-splitter constructed from a shard of mirror glued to a microscope slide, which reflects the laser light through an Amscope microscope objective. The objective both focuses the laser light on the sample, and captures the backscattered raman radiation back through the microscope slide where it's collected, filtered, and processed.

This configuration captured raman spectra that were more intense than the previously-captured spectra by several magnitudes, resulting in a superior signal-to-noise ratio. Below, I've attached a sample spectra generated by the backscattering configuration, as well as a few pictures of the setup.

EDIT: Once (if?) I ever get a stable baseline without resorting to numerical methods, I'd really like to try to pair the raman spec with TLC to identify different eluting compounds.

Amateur_Raman_Spectra_Aspirin_Backscattered_GKau.jpeg - 34kB

Attachment: Aspirin_Backscattered_Raman Spectra (GKau).pdf (12kB)
This file has been downloaded 849 times

IMG_0070.JPG - 210kB IMG_0072.JPG - 235kB IMG_0074.JPG - 244kB

[Edited on 24-2-2013 by radagast]

IrC - 23-2-2013 at 23:08

Try more like 20 to 50 mw. I have a couple of those. 5 mw is quoted to get past customs.

I like this deal better:

http://www.ebay.com/itm/261089893420?ssPageName=STRK:MEWNX:I...

If you look at the fine print you see "Output power:< 2mw". It is actually 200 mw or very close but ebay would kick it off and customs grab them thus the lie. It is more of a code they use to get these sold and delivered. Easy to figure out when you read "Burning: can burn the black matches within 10cm after adjustable focus". Mine lights red non strike anywhere matches easily and cuts out patterns in trash bag plastic quite easily. As you can guess it is not a 2 mw laser. Neither is the one you have a 5 mw. Be sure to add a note you are in the U.S. and it will come with the sleeve and 18650 battery. However do not run it above 3.7 volts if you are going to use a supply. No reason to damage it and as I stated it will light matches at 3.7 volts so the power is plenty for your project.

At $33 including the battery, sleeve, laser, keylock, and safety glasses I think this is a great deal. Also includes a screw on front assembly which projects the beam like a disco ball with dozens of points. I just use it to keep the lens covered and dust free when I am not using it.


radagast - 23-2-2013 at 23:39

Quote: Originally posted by IrC  
Try more like 20 to 50 mw. I have a couple of those. 5 mw is quoted to get past customs.

I like this deal better:

http://www.ebay.com/itm/261089893420?ssPageName=STRK:MEWNX:I...

If you look at the fine print you see "Output power:< 2mw". It is actually 200 mw or very close but ebay would kick it off and customs grab them thus the lie . . .



Thanks for the heads-up, IrC. When I saw the beam of my laser glittering brightly, I began to get the clue that it was not in fact < 5 mW. And, when the beam felt like a burning ember after adjusting the focus a bit and shining it on my hand, that's when I started sprinting for the safety glasses. In fairness to the manufacturer, that happened after I supplied power from a D.C. adapter -- but that just makes me think that the laser pointer contains an over-spec'ed (or lacks) current regulation.

I was simultaneously so angered and pleased that I immediately had my eyes checked for retinal damage, and then ordered another one.

I'd find a 200 mW laser "pointer" terrifying, but I'll probably order it anyway to try it out. The problem is that the current less-powerful laser is already strong enough to partly overwhelm the edge filter, and I have a feeling that a 200 mW laser would be too much without adding a stronger filter. In addition, the seller says that the laser should not be operated for longer than 30 seconds due to heat buildup (another clue that this isn't your average "2 mW" laser . . .). This is a problem, because I generally collect hundreds of spectra and average them, with each spectra taking anywhere from 50 milliseconds to 10 seconds each. But at $30 bucks or so, it's basically the cost of two NYC vodkas, so why not . . .

I have to LOL at the idea of using a 200 mW laser as a pointer at a presentation at my job while trying a case. Really gives a literal dimension to the phrase "blind justice" . . .

IrC - 24-2-2013 at 00:32

Just be careful one good hit even from a piece of the power reflected, that eye is gone for life. Just had a mini ice storm 2 days ago and shining up to low clouds with some de-focusing it lit a big patch of cloud very brightly. First pointer I have seen do that. With tighter focus you can see it on very high clouds, one of the reasons I like the green. My 300 mw blues burn things even better but just not visible enough to see in a cloud. Running on the 18650 I have kept it lit a few minutes and while getting warm I don't think it cut the life down very much. Would be a different story on a power supply.

I do not need to say don't point it up when any aircraft are anywhere around. This laser is most definitely not a toy, thus the set of keys to protect kids when your not around.

I should add this is a very good thread, should probably be in technochemistry. Also I wish you would add some in depth detail in all aspects of your project here these are the subjects which make me want to go start building things. Would also be nice if someone added some good detail on the devices they are using to detect trace chemicals on things from a hundred or more feet away. Very interesting topic.


[Edited on 2-24-2013 by IrC]

watson.fawkes - 24-2-2013 at 07:25

Quote: Originally posted by radagast  
The problem is that the current less-powerful laser is already strong enough to partly overwhelm the edge filter, and I have a feeling that a 200 mW laser would be too much without adding a stronger filter.
Yet on the other hand, the higher power makes for more non-linearity in the target, leading to better signal. What to do?

There are two ways of lowering power of a laser diode: lower the current or lower the duty cycle. The first is obvious. The second, though, opens the possibility of turning off the spectrometer sensor while the power is on. This would avoid flooding the CCD with immediately-reflected light. This technique goes by the name "time-resolved" spectroscopy. It's not off-the-shelf, as your current system is, since it requires a time base connection between a pulsed supply and the spectrometer, not to mention that a suitable pulsed supply isn't COTS (commercial off the shelf) either (not to my knowledge). On the other hand, it might be possible to modify the spectrometer if its CCD chip has the right pin, one that, say, would prevent charge accumulation which the pin was held active. The pulsed supply itself is pretty easy; it's a variable duty-cycle oscillator with a MOSFET output to switch power to the diode.

Here's a related paper, pulled at random:
A Low Cost Time-Resolved Raman Spectroscopic Sensing System Enabling Fluorescence Rejection

unionised - 24-2-2013 at 11:46

I think I may be able to beat you all for the ultimate cheapskate Raman system.

In a fully darkened room, the light scattered from a green laser pointer (probably about 1mW) by a strong solution of sodium chromate is visible through a bandpass filter that blocks the green but lets the yellow light through.

So I have seen a Raman spectrum- albeit that there's no real wavelength discrimination in these circumstances.

I'm thinking of ways to improve this. I think the first thing I need is to take the laser out of a pointer and fit it in a box that will not roll off the table and wich doesn't need you to hold the button down with sticky tape.

I forget how much the filter cost me- it was on ebay It's not very narrow band, but it doesn't need to be since it's centred on yellow light; it blocks the green completely.

radagast - 25-2-2013 at 09:58

I'm still trying to find out what's causing the elevated baseline in the back-scattering configuration. I'm thinking about installing a beam dump (as Watson Fawkes suggested) to absorb excess laser light behind the beam splitter, or putting the apparatus in a different enclosure. In the meantime, here's an interesting youtube video by "toc1955" who used a raw CCD to capture raman spectra from a blue laser. Instead of using a filter to exclude the laser wavelength, he just covered the area of the laser corresponding to the laser wavelength with black cardboard:

http://www.youtube.com/watch?v=-lVMYa25jtU

@IrC:

I'd be happy to post additional details once I get clean spectra, with a complete write-up on calibrating the spectrometer, etc. One of my motivations for pursuing laser spectroscopy is the thought that, over time, enough people will build these to create a growing library of raman spectra, which would be tremendously useful in confirming product identity.

Your warnings on wearing glasses are well-taken, especially when aligning optics. (For instance, I took a indirect hit in my left eye from the laser while aligning the beam-splitter. Because I was wearing safety glasses, I only saw a faint flash of green instead of sustaining any damage). Re: long-distance raman spectroscopy, I had actually came across an article describing the basics and will post it once I find the link.

@Watson Fawkes:

That's a fascinating article. Frankly, circuit design (and EE projects in general) are my "blind spot" and the last time I worked with MOSFETs was almost 15 years ago, when seeking to build a tesla coil as a kid. Would love to pursue this further when (if?) I ever gain knowledge sufficient to create those timing circuits.

@Unionised:

Nice work. I actually experimented with using my 532 nm protective glasses as raman filters. It worked to the extent that the most dominant raman signals appeared on the spectra, but the raman signals were generally quite weak. I don't think those glasses were spec'd to let in a percentage of non-532 nm light.

With respect to your laser pointer, I'd recommend using a pipe clamp to hold down the button. (See pictures in previous posts). You can then use a vise to hold the laser steady.

In order to see discriminate between wavelengths, you could use a "real" spectrometer (as I did), or you can create one out of diffraction grating (a DVD would probably do the job), a high-resolution digital camera, and a few mirrors. The diffraction grating will resolve the raman light into different lines, which can then be photographed by a calibrated camera. See, e.g. "A simple spectrophotometer using common materials and a digital camera", J. Physics Education, 46(3) (claiming .5 nm resolution over the 415-660 nm spectrum using similar materials). You can then use software to convert the spectral image to a wavelength/intensity spectra. (I would have tried something similar, but getting consistent raman spectra is already difficult so I figured I focus on getting the basics right first).

[Edited on 25-2-2013 by radagast]

Toluene Spectra, and Sample Vial Test

radagast - 25-2-2013 at 21:40

One of the really neat things about raman spectroscopy is that there's virtually no sample preparation. I've heard claims that you can test samples simply by shooting the laser at a reagent bottle. I found this was true, to the extent that the bottle is made of clear glass. (YMMV -- I'm sure a different configuration would fare better with other kinds of glass).

Today, I received a tiny amber bottle of toluene, which was ideal for testing because it's a strong raman scatterer. I tested the toluene straight out of the USPS box without opening the amber bottle. Glass is a very weak raman scatterer, but I found that it blocked enough raman-shifted light to greatly weaken the signal.

After opening the bottle of toluene, I poured it in a clear glass vial, and obtained a much stronger signal. I've plotted the amber vs. glass spectra using the terrific ggplot2 R package by Hadley Wickham. I've also attached a high-resolution professional spectra of toluene. As you can see, the "ghetto raman spectrometer" did not fully resolve the ~1000 cm-1 doublet in the glass vial, but did so when scanning the amber bottle. Perhaps the dominant peak at 1000 cm-1 is obscuring its sister peak. The spectra contain other small artifact peaks, but there is otherwise a fairly good match.

Attachment: Amateur Toluene Raman Spectra (GKau).pdf (11kB)
This file has been downloaded 944 times
Amateur Raman Spectra of Toluene (GKau).jpeg - 60kB
toluene_commerical_high_res.jpg - 13kB
Laser_Toluene_Bottle.jpg - 281kB
Clear_Toluene_Vial.jpg - 251kB

[Edited on 26-2-2013 by radagast]

NHZ - 26-2-2013 at 02:54

A quick note on the laser used. Low cost 532's usually do not come with an IR filter. This allows 3
wavelengths to pass. Given the working distance of your setup, you get a ton of IR 808nm & 1064nm
spewing out at aperture.

Regarding ebay, there are China based sellers that list 405nm pointers as 5mW and have been tested
up to 80mW. They do this to circumvent customs, and keep the listing from being pulled. The insanity
to this is that common people buy these near UV lasers to play with pets and im sure even some
allow their kids to play around with them.

Always be very cautious of anything listed as 5mW and made in China. Sellers who list as 1mW
do this as some countries restrict over 1mW. Very distressing to say the least.


If you are interested, I build my own DPSS modules from the driver through to the mounting and
alignment of the pump medium. I can set desired outputs tested on a calibrated power meter.

My 532 modules are also IR filtered and include power supply. I also work with 473nm

I have some videos for reference here > https://www.youtube.com/user/TunedCavityLasers

A lot of what I do is high power stuff, but also make reliable low power as well.






[Edited on 26-2-2013 by NHZ]

IrC - 26-2-2013 at 14:44

Would one of these filters work in your project radagast? Until Friday they are having a 46 percent discount on everything they carry, mostly optical items.

http://www.surplusshed.com/pages/item/l10079.html

http://www.surplusshed.com/pages/item/l10076.html

http://www.surplusshed.com/

watson.fawkes - 26-2-2013 at 16:20

Quote: Originally posted by IrC  
Until Friday they are having a 46 percent discount on everything they carry, mostly optical items.
I couldn't find this sale on their web site anywhere. Turns out that's because it's not there. It's on their Twitter feed: https://twitter.com/SurplusShed.

IrC - 26-2-2013 at 16:27

If you have been buying from them a long time and are on the mailing list they send you a coupon code in the email. Never thought about that but your right, good you looked at their TF. While your still around do you have an opinion on the two filters I listed as relates to this thread? I have never really delved into this subject before so I am unsure until I study much more just how to choose a filter for the project. Since that would take me more than a few days and the sale ends soon I thought getting advice makes sense.

radagast - 26-2-2013 at 19:09

@IrC:

Thanks for the link! Some great stuff for sale there. Frankly, I'm not sure what they are. I'll defer to WF and other posters more knowledgable about optics, but the filter resembles a "dichroic mirror/beamsplitter" described by Thorlabs:

Quote:
A dichroic mirror/beamsplitter functions as a 50:50 beamsplitter at its design wavelength, known as the cutoff wavelength. A longpass dichroic mirror is highly reflective below the cutoff wavelength and highly transmissive above it, while a shortpass dichroic mirror is highly transmissive below the cutoff wavelength and highly reflective above it.

Thorlabs' Dichroic Mirrors/Beamsplitters are offered in eleven different cutoff wavelengths ranging from 425 - 1800 nm, and they provide >90% average transmission and >90% average reflection over their specified bands (see the graphs below). They are designed for use at a 45° angle of incidence and are available in sizes of Ø1/2", Ø1", Ø2", and 25 mm x 36 mm. Please refer to the table to the right to choose an appropriate filter for your application, and see below for representative transmission and reflection plots.

Dichroic filters feature a dichroic coating on one surface and an antireflection coating on the opposing surface. On round optics, an engraved arrow points toward the surface with the AR coating; on rectangular optics, the side with the engraving has the dichroic coating.

Applications

Dichroic mirrors/beamsplitters can be used to combine a beam that has a wavelength (or wavelength range) shorter than the design wavelength with a beam that has a wavelength (or wavelength range) longer than the design wavelength while minimizing intensity losses. Alternatively, spatially overlapping beams of different colors can be split with a single optic. This feature is commonly used in fluorescence microscopy to prevent light of the excitation wavelength from reaching the imaging detector. Please see the Applications tab for schematics of example experimental geometries.


http://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=331...

SurplusShed doesn't specify whether the filter is longpass or shortpass, and I can't really tell by the color. The filter I use (an edge filter that cuts off a high % of light under 540 nm) is dark red/orange, and this one is dark red, but I'm not sure that means too much. Assuming it is, though, I suppose you could use it as a dual beamsplitter/filter, and substitute it for both the microscope slide and the filter in my setup. The problem is that it'll still let through a lot of 532 nm light, since under our assumptions, it only cuts off at 531.6 nm. That means that the laser light will still likely drown out the raman signals. On the other hand, if this is a shortpass filter, it will block everything above 531.6 nm, including the stokes raman shifts that we need.

I might grab a few just to test it out, though. If it pans out, I'd be happy to send one of them your way. Generally, what you're looking for is a filter that will cut off around as much light < 540 nm, while letting through virtually everything > 540 nm. In a pinch, you can use an OG550 filter (basically, a piece of orange glass), but these cut off around ~580 nm (IIRC) so you lose the rich structural information between ~150 to ~1000 cm-1.

I actually have five of these OG550 filters which I'll never use, since I ordered the wrong size (they're tiny, maybe 1 cm diameter?) so if you want one, just PM me your address and I'll send one to you.

@NHZ

Frankly, I'm surprised that these laser companies haven't been litigated into the ground. I would launch a class-action suit myself but-for the fact that I love their product . . .

Do you have any recommendations for (1) a laser power meter; and (2) a high-resolution spectrometer for testing laser modes? With respect to the latter point, I'm thinking of getting one of those blue single-mode diodes that are typically used in a projector.


[Edited on 27-2-2013 by radagast]

[Edited on 27-2-2013 by radagast]

watson.fawkes - 26-2-2013 at 21:07

Quote: Originally posted by IrC  
Would one of these filters work in your project radagast?
The filter description (for one of them) is "MOUNTED 15.5MM SQUARE INTERFERENCE FILTER FOR 531.6NM". The magic search term is "interference filter". The page on Wikipedia is so-so. Much better is the page on HyperPhysics. You need to know some optics, in any case, but nothing more that first-year college physics to grasp the basics. If you look at both photos, front and back, you'll note that they show different colors, which is typical of any interference device.

The real question, though, is what kind of interference filter is it? Given the mounting and the small window, I'd guess that it's for mounting perpendicular to a beam, not originally in the middle of some more complex optical train. It might have been a laser line filter, used to pick out a narrower frequency range than the ordinary output is. I don't know enough about that class of materials to say anything about what the beam centered at 532 nm looks like in frequency space, but it's fairly safe to say that it will be narrower if you put in a laser line filter. It's also safe to say that those cheap high-power lasers are built for output power, not monochromaticity. If (a big if) it's actually a narrow pass filter, then you'd put it right after the laser emitter, before the rest of the optical train. It might cut down a lot of the spurious noise in system as a whole.

watson.fawkes - 27-2-2013 at 13:28

I got curious about why there should be commercial paired filters near 532 nm. For the easy-to-see version, see the article Neodymium-doped Gain Media. The &nbsp;4F3/2 level for the Nd ion is split into a doublet (R1-R2) by the Stark effect that arises from electric fields within the crystal lattice. The &nbsp;4I11/2 level is split, as it turns out, into a sextuplet (Y1-Y6). I found more complete data in this paper (full title formatted, just for you all): Stark components of lower-lying manifolds and emission cross-sections of intermanifold and inter-stark transitions of Nd3+(4f3) in polycrystalline ceramic garnet Y3Al5O12. Figure 2 of that paper has the intensity data for the &nbsp;4F3/2 --> &nbsp;4I11/2 transition, the strongest two of which are both near 1064 nm, which after frequency doubling gives the 532 nm green. I'd have to guess these filters were designed to separate this line pair.

radagast - 28-2-2013 at 22:54

@Watson

You are a fountain of knowledge, sir. I was wondering about the strange range of the filter too, but I can't claim to have ever even heard of the Stark effect.

Just a random collection of updates --

(1) The DIY spectrometers are back up on Ebay again. See, e.g. http://www.ebay.com/itm/Compact-Fiber-Coupled-CCD-Spectromet...

(2) I accidentally left the laser on for about 24 hours. It was warm to the touch, but seemed to be perfectly fine. So, I'm hopeful that this configuration will be relatively stable.

(3) There's a terrific tutorial on how to make your own optical breadboard at youtube by "plenum88". See, e.g. http://www.youtube.com/watch?v=sr22VxNpk8U

(4) The spectra I've posted are definitely not optimized, because the spectrometer is not highly calibrated and the laser probably has fatter spectral width than we'd like. By way of example, a dedicated laser fanatic took the same spectrometer, shot a 532 nm laser into it, and took spectra where the base of the laser spectra spanned 530 to 534 nm. See http://redlum.xohp.pagesperso-orange.fr/electronics/spectrom...

By comparison, a similar test on my spectrometer yields a substantially broader base (~528 to ~536 nm).



[Edited on 1-3-2013 by radagast]

watson.fawkes - 1-3-2013 at 09:14

I was curious as to whether the spectrometer used in this project could be hacked to do any time-resolved functions. The short answer is no. Its CCD is a Sony ILX511, a 2048 pixel linear B/W sensor. It's big brother is the Sony ILX526A, a 3000 pixel linear B/W sensor. The ILX526A has an electronic shutter function; the ILX511 does not. See this new product announcement for a summary. This announcement isn't dated, but a little digging with the Wayback Machine first shows this page in December 2003 referencing the ILX526A.

From what I can tell, the most recent linear CCD from Sony with shutter is the ILX718K, a 5363 pixel linear RGB sensor.

radagast - 1-3-2013 at 11:04

@Watson

The Sony ILX511 is also the chip used in the Ocean Optics 2000 series, so at least in terms of CCD capability, these B&WTek/Science Surplus seem to offer similar hardware as substantially more expensive spectrometers.

Do you have any thoughts on how difficult it would be for someone without electrical-engineering experience to pair a CCD with a circuit driven by commands from a computer? I think it'd be neat to see if I could replicate what's going on inside the little blue spectrometer box on an optical breadboard, and learn more about Czerny-Turner designs. I'd also be interested in developing a command-line interface to the CCD so that I could tell it to record spectra using a range of integration values to find the optimum integration time.

Plus, the upgraded CCDs you mentioned could be scavenged from scanners, and are readily available . . .

[Edited on 1-3-2013 by radagast]

watson.fawkes - 2-3-2013 at 07:03

Quote: Originally posted by radagast  
The Sony ILX511 is also the chip used in the Ocean Optics 2000 series, so at least in terms of CCD capability, these B&WTek/Science Surplus seem to offer similar hardware as substantially more expensive spectrometers.

Do you have any thoughts on how difficult it would be for someone without electrical-engineering experience to pair a CCD with a circuit driven by commands from a computer?
For reference, I've attached a copy of the data sheet for the ILX511. In the block diagram on page two, you'll notice that the block next to pin 1, VOUT, is labelled in part "S/H Circuit". That stands for "Sample and Hold", and pin 1 is an analog output. In order to get data, it needs to be fed into an A/D converter. So the short answer to the question is no, you can't drive the chip directly from a computer.

The longer answer is that it doesn't take a huge amount of electronics knowledge to interface the chip with a microcontroller and to build something that would interface nicely with a host computer. The saturation voltage is listed at 800 mV and the dark voltage is listed at 3 - 6 mV with excursions twice that. There might be some issues with matching these voltages to the those of a built-in microcontroller. These could be solved with a level shifter, a simple op-amp circuit. The dynamic range of this chip is listed at only 267 (= 800 mV / 3 mV ), and if you take the denominator as the dark current variance, it means there's not much more that 8 bits of good data coming out of the chip in the first place. So if you were going to do this with an Arduino, which uses an AVR chip with 10-bit A/D converters, you'll be fine. The max sampling rate of the chip is 2 MHz and the Arduino has a 16 MHz clock rate, so there's 8 clock cycles of instruction available for your central sampling rate (or twice that if you drop it down by a factor of 2), which means you need to hand code it in assembler. All in all, not a beginner's project, but a decent medium-difficulty one.

This analysis is more-or-less generic for all these sensor chips. They've all need an external A/D to function, and there's always a sampling loop in the microcontroller code. I did a little hunting for other chips yesterday, and they all seem to have the same style of interface, even those from other vendors. The other change, and this is more significant, is that CCD seems to have passed out of service entirely for new linear sensor designs, having been replaced with CMOS sensors.

A note on chip availability. The ILX526A seems to be available as NOS (new old-stock) for around $20. If anybody's going to go to the effort to build a linear interface board, I'd highly recommend using a sensor with shutter control. Given the amount of time invested in such a project, it seems silly not to have the option of using it later.

Lastly. If you want to experiment with spectrometer optics, the very easiest way of getting a linear sensor like the one you're already using is to buy a second box like the one you've got and extract its sensor board for your own use.

Attachment: ILX511 2048-pixel CCD Linear Image Sensor B-W, Sony.pdf (129kB)
This file has been downloaded 848 times

radagast - 2-3-2013 at 20:23

@Watson

Thanks for the detailed advice, Watson -- that's very helpful! It'll probably take me at least several months to get a prototype working, but I've been eager to learn some basic circuit-building so this will be a great way to do that, and hopefully work my way toward using a ILX526A chip. Re: the easiest way of getting a linear sensor, I'll snag another one of those DIY Science Surplus / B&WTek spectrometers, take it apart, and replace the built-in breadboard with a custom setup. Even if it doesn't work, I could always put the circuit and CCD back into the built-in breadboard, and calibrate that spectrometer for use with a laser of a different length.

@all

(1) I built a few more beam-splitters to determine the effect of the mirror size on the spectra, and found that mirrors can introduce artifact peaks. For instance, one of the annoying artifact peaks at around ~2600 cm-1 in the aspirin spectra disappeared after I installed a new beam-splitter.

(2) The DIY Science Surplus / B&WTek spectrometer comes with a help file containing codes for communicating with the spectrometer, which opens up the possibility of writing your own serial interface to it. I'm going to poke around and see whether I can write a simple Python program to control the spectrometer.

[Edited on 3-3-2013 by radagast]

watson.fawkes - 4-3-2013 at 09:02

I spent some time over the weekend looking into linear image sensors. The main application seems to be document scanning, so a number of these chip are "x 3" with RGB outputs. There seems to have been much more competition a decade ago for these. What seems to have happened is that the business has gone primarily wholesale, with large sales to major manufacturers.

Here's the upshot: TSL1412S - Linear Sensor Array with Hold. Available through ordinary channels like Digi-Key and Mouser. Cost is USD 40 - 45. This is the chip I'd use in any new design. Widely and easily available, has a shutter, adequate linear density, good dynamic range.
One thing I learned that I was a bit surprised to learn is that a few of the vendors don't sell through to retail channels; you have to call them to know anything at all about buying from them. These, not so fortunately for the amateur, also seem to be the most scientifically oriented ones. I didn't talk to any of them. I've listed them for completeness, but I wouldn't use them in any design for amateurs, because of the hassle of getting them as well as the uncertainty about availability in the first place. Also, I'm not sure that Sony and Toshiba shouldn't be on this list.No longer producing product. Best I can tell, they've left the business. Plenty of data sheets for previous products out there, and some NOS (new old-stock) inventory.

IrC - 5-3-2013 at 11:18

radagast "I actually have five of these OG550 filters which I'll never use, since I ordered the wrong size (they're tiny, maybe 1 cm diameter?) so if you want one, just PM me your address and I'll send one to you."

Thanks I just received it today in the mail. Have a bit of studying and building to do. I am thinking about looking into Watsons flatbed scanner sensor idea. Now on to searching ebay for a good new or used CCD, one I can find in depth data on before
I buy it. Have an Arduino 2560 Mega which might be a good project to work on.

After doing some searching I find a lot of the information to be quite vague. Example in the image. Is the Rayleigh filter set at a specific angle such as a Brewster angle? Do you want the beam hitting the target sample as tightly focused as possible? Would blue work better than green? I know different filters but I am curious. Also if we have a liquid sample in a Curvette, would the point of beam focus be in the center of the Curvette? Most of the pages I have looked at are geared towards describing the setup in broad, general terms. Do not really find a specific example with everything shown with precision, leaving many questions.


RamanSpectroscopy.jpg - 53kB

http://www.ebay.com/itm/Mini-Projector-DJ-Disco-stage-light-...

I hacked one of these. Two lasers in nice holders, one shining through a flat square splitter at 90 degrees, other at Brewster angle, resulting in a single beam of both. Circuit board regulates each and has a fet condenser mike for activation of both lasers from room audio. Floating differential input but an audio transformer is easy isolation (you cannot reference either side of the mike to ground or it kills the laser supply). This could be used to switch the beam assuming you remove one laser for the project, say keeping the green for example. The green is around 80 mw and the red is closer to 120 mw. At $28.99 free shipping it is a great source for lasers of decent power plus the other components. Not to mention it is a neat light show even if somewhat boring after a while. Instructions say do not leave on past a couple hours. I left it on for two weeks nonstop with no observable reduction in either laser, the unit has a very efficient yet quiet fan. You can see the fan output in the third image below. Runs on 12 volts and comes with everything in the pictures. A goldmine of hackable parts for this project and very easy to disassemble, nicely built. I am sure the motorized grating assembly could be used for something else and you can use the beamsplitter as in the first image above.


Lshow8.JPG - 40kB Lshow82.JPG - 65kB Lshow83.JPG - 41kB


[Edited on 3-6-2013 by IrC]

smaerd - 6-3-2013 at 13:12

Incredible work. Gets me thinking about making an IR spec. Wish I had anything to add on hehehe.

IrC - 6-3-2013 at 14:04

I should add the 'beam turning optic' in the top pic is what I was referencing concerning using the really high quality splitter that comes in the laser light show. The green laser will shine at 90 degrees to the splitter face, and the signal will go through it at roughly the Brewster angle. Does anyone see any reason this will not work?

bfesser - 6-3-2013 at 16:49

Every time I see the title of this thread, I read it as 'Cheap, Low-Resolution, Ramen Spectroscopy':
top-ramen.jpg - 90kB
[Edited on 3/7/13 by bfesser]

Fixed broken HTML -Texium

[Edited on 11-19-2018 by Texium (zts16)]

radagast - 6-3-2013 at 17:54

@all

Not directly related to raman, but once you get into lasers, you might as well go all the way into laser-induced atomic emission spectroscopy. Not the most useful technique for organic chem analysis, but very impressive:

http://www.viewsfromscience.com/documents/webpages/methods_p...

@Watson

As always, thanks for the expert advice -- I'll look further into TSL1412S. I have a feeling that those other suppliers may require ordering in bulk, but I may give them a dial just to check. In speaking with other scientific suppliers (e.g. Gilson, Shimazu, B&WTek) I've found that their tech support are often willing to go the extra mile for interested amateurs, so you never know.

@IrC

With respect to the positioning of the laser, the ideal spatial resolution, and the angle of the beam -- those are good questions, and largely empirical as far as I can tell. I've been tied up with my day job for a while, but I'm hoping that I can test the spectrometer in a bunch of different configurations this weekend and post the resulting spectra and higher-resolution photos of the my current setup so that other people don't have to reinvent the wheel. (I dream of high-res raman but my highest res camera is the one that came with my Blackberry . . . go figure).

You can use a blue laser, and the wavenumber shifts should still be the same as those from a green laser. As your excitation wavelength decreases, the raman signal increases exponentially. However, the tradeoff is that fluorescence also increases, and my view is that flourescence is a much bigger problem than detecting the faint raman signal. (Hence my interest in Watson's article re: time-resolved spectroscopy).

Thanks for the details on the laser projector . . . looks promising and I'll check it out.

@smaerd

I wish someone would build an IR spec . . . I searched for a DIY IR spec, but it seemed like nobody had done this and published the results. Raman and IR are complementary techniques, so it would really be something if we could have access to both.

What would be really cool is if there were something like a handheld Heath-kit raman set available that you could put together yourself, which had software for the iPhone or Android-based phones which would match up to a library of spectra taken by other users . . . I feel like that is what's needed to gain a critical mass of freely-available raman spectra. Probably a pipe dream, but it'd be great to be able to analyze your tap water with your phone and a little accessory, or your soil, or whatever you want.

@bfesser

Pretty sure you just won the thread, almost no point in continuing now! By the way, would like to hear your progress on the CCD controller, as we're working toward similar ideas.

[Edited on 7-3-2013 by radagast]

watson.fawkes - 6-3-2013 at 18:40

Quote: Originally posted by radagast  
In speaking with other scientific suppliers (e.g. Gilson, Shimazu, B&WTek) I've found that their tech support are often willing to go the extra mile for interested amateurs, so you never know.
That is indeed true. On the other hand, one of my reasons for recommending that chip was that it's appropriate for an open source design, so that the supplier who isn't set up to handle the volume of interest from less-than-expert amateurs is not swamped by such calls.

IrC - 6-3-2013 at 18:53

When you have the time to operate your setup I would be interested in an analysis of the beef noodles, with and without MSG. I never liked the chicken and their spicy shrimp always burned my guts out so no need to do them.

bfesser one of us has been up for too many hours. I just realized your spectral lineup of noodles.

On the link for Laser-induced breakdown spectroscopy I noticed mention of a very old book, "How to work with the spectroscope", 1878, John Browning. I looked it up on Archive, they have it. Old but interesting, I D/L'ed the PDF version. Often I find these old texts very informative. Working with crude technology they built some very high quality instruments. Not to mention often these older books do better jobs of teaching some subjects than many newer texts.

http://archive.org/details/howtoworkwithspe00browrich




[Edited on 3-7-2013 by IrC]

Marvin - 7-3-2013 at 11:31

Really cool project. I have some of the bits, I'm playing catchup.

The TSL1412S looks to be CMOS technology and the datasheet indicates it does not integrate well beyond 100ms. Doesn't sound suited to Raman.

The ILX511 is sensitive, 2k CCD and has a low read noise. The Toshiba TCD1304AP is almost a 4k CCD. It is gated, has a better well depth and has a slightly greater read noise. Both of these are inexpensive and available.

If the B&W based spectrometer 50um slit could be replaced with a 25um slit, this would significantly boost the resolution. Did this come up in any of your conversations radagast?

radagast - 7-3-2013 at 20:45

@Marvin

Thanks, I think it's terrific that you and IrC are working on this project, too! I haven't thought about changing the spectrometer parameters too much (other than mulling over the CCD stuff and ordering The Art of Electronics so I can get a better handle on circuit basics), because I have a feeling that the spectrometer I'm using right now actually has decent resolution if used correctly (1 nm or so). The main thing reducing the quality of the spectra, I believe, is my poor calibration of that spectrometer, and my lack of technique as far as where to shine the laser beam, etc. -- e.g. what IrC was asking about above.

Reducing the slit is a good idea -- the only issue is that it would reduce the amount of light coming in, and the raman scattering is already so weak. But once we get the other parameters fixed, it's certainly worth a look.

@all

I passed by FAO Schwartz today, and inspiration struck: I thought I might be able to create a Lego 90-degree prototype for those people who might be interested in this project, but don't want to spend barrels of cash for Thorlabs posts, a breadboard, accessories, etc. (You can pry my Thorlabs parts out of my cold, dead, hands because I love'em, but they are pretty pricey). Much to my disappointment, all the Lego kits there were custom Lego sets, and there were no simple Lego bricks there.

Long story short, I raided the Lego store at the Rockefeller center, and after 20 bucks, 15 minutes of shopping, and about 15 minutes of building, I'm proud to present the LEGO RAMAN SPECTROMETER, which is even shabbier than the first version of the raman spec I built.

Interestingly, the 90 degree Lego Raman spec took pretty good spectra of aspirin -- in fact, just as good as the backscattering configuration with Thorlabs parts. After some experimentation, I found that the sample has to be close to the collection filter as possible -- otherwise, the raman signal will be too weak. I've attached the professional aspirin raman spectra, too, for comparison.

I haven't found an effective way to take liquid spectra at the 90 degree angle yet, but will experiment a little this weekend, as well as finally taking some pictures with a higher-res camera of the Thorlabs backscattering configuration.



IMG-20130307-00089.jpg - 356kB
GKau Raman Spectra Lego 90 Degrees.jpeg - 54kB
Professional Raman Spectra (Aspirin).jpg - 8kB
Attachment: GKau (Lego 90 Degree) Aspirin Raman Spectra.pdf (17kB)
This file has been downloaded 595 times
IMG-20130307-00095.jpg - 341kB
IMG-20130307-00096.jpg - 352kB
IMG-20130307-00097.jpg - 327kB
IMG-20130307-00098.jpg - 365kB

[Edited on 8-3-2013 by radagast]

watson.fawkes - 8-3-2013 at 06:34

Quote: Originally posted by Marvin  
The TSL1412S looks to be CMOS technology and the datasheet indicates it does not integrate well beyond 100ms. Doesn't sound suited to Raman.

The ILX511 is sensitive, 2k CCD and has a low read noise. The Toshiba TCD1304AP is almost a 4k CCD. It is gated, has a better well depth and has a slightly greater read noise. Both of these are inexpensive and available.
There are higher-quality chips available, that's certain. If you don't plan on publishing a design for other amateurs to learn from and build with, use anything you'd like. General availability is not a concern in such a case. When I was looking, however, the Toshiba and Sony chips were only available in various end-of-life channels, which include eBay and the various surplus-stock IC vendors; I couldn't find any of the major distributors that were stocking them.

IrC - 8-3-2013 at 11:51

Watson you sound like the CCD expert here, what do you think about the one in the DSC-F707 or 717 Digital Still Cameras? Don't have the part number easily available right now. I have 4 of the 707's and 3 of the 717's. Half of them have had the CCD fail which I replaced by ordering the CCD's from a seller on ebay. Very hard job as they had to be soldered in after removing the original parts. While I do not have a clean room and the optical assembly had to be disassembled, they all take fantastic pictures. Meaning I do not think they collected very much microscopic dust during the work. I will have to go through the junk box for one of the old units and see if a number is printed on it. They are 5 megapixel, the camera is 12 mp including the digital processing. One problem is these things seem to fail far too often, I have read Sony had troubles with some kind of manufacturing defects in this line of CCD's. This makes me worry if the work is worth being done based on this part. I only bring it up as somewhere in a box I still have a new replacement spare, but no data sheet. Are you familiar with the CCD in this camera line and know where a handy PDF link for the CCD is?

Also would you start this project using one knowing the failure of the CCD was the most common failure for the model series?

I just thought of this: Since I have dozens of the 707's and 717's as spare parts I have plenty of components. One part was an IR filter in a holder controlled by a solenoid. In nightshot mode the solenoid moves the filter out of the way and alters the focus to compensate for the shift. It is 1.24 mm thick, 12.95 mm X 10.95 mm in size. Since the green lasers usually also have significant IR output would it make sense to add one of these after the laser so that only the green light made it into the rest of the assembly?



[Edited on 3-8-2013 by IrC]

radagast - 9-3-2013 at 11:11

(1) BTW, while browsing W's laser blog, I came across his designs for a CCD driver, which uses the ILX553B CCD:

http://redlum.xohp.pagesperso-orange.fr/laser/gratingOSA.htm...

(2) I was puzzled as to why I kept on getting that strange background hump in the aspirin spectra, when there was no mention of any baseline correction in the Mohr et al. article. Then it dawned on me (facepalm) that I was using a laser that was at least several times more powerful than the 5 mw laser used there. Assuming that fluorescence doesn't increase equally across the wavelength range while cranking up the laser's power (and I don't see why it would), this might account for the background hump.

The logical solution would be to decrease the laser power (maybe with a variable resistor?) but I first tried lowering the integration time from 3 seconds (as described in the paper) to .7 seconds. The result was a substantially flatter baseline. Not quite there yet, but getting there . . .



Attachment: GKau (Reduced Integration Time) Raman Spectra.pdf (12kB)
This file has been downloaded 528 times
GKau (Reduced Integration Time) Raman Spectra.jpeg - 57kB

Professional Raman Spectra (Aspirin).jpg - 8kB

[Edited on 9-3-2013 by radagast]

IrC - 10-3-2013 at 03:08

How much improvement do you think there would be to have the laser diode heatsinked with a thermoelectric cooler to reduce mode hopping and other effects which I believe could be considered as adding noise to the desired signal?

watson.fawkes - 10-3-2013 at 07:29

Quote: Originally posted by IrC  
Watson you sound like the CCD expert here, what do you think about the one in the DSC-F707 or 717 Digital Still Cameras? Don't have the part number easily available right now. [...] Are you familiar with the CCD in this camera line and know where a handy PDF link for the CCD is?
[...]
Also would you start this project using one knowing the failure of the CCD was the most common failure for the model series?
A bit of internet searching yielded the part number: Sony ICX282. Here's its new product announcement. Datasheet is attached. The biggest problem you'll have with using a camera chip is that the drive circuitry is significantly more complicated. While looking this up, I also found the Sony CXD2498R, a 48-pin chip specifically design to drive the CCD. You'll need much of the capacity of such a drive chip, so you can either use one or implement what you need of one with a microcontroller. Frankly, I can't see the advantage unless someone else had already done the engineering work. It's a good bit of work to fiddle with all that signal generation. For personal use, almost everybody's time-vs.-money tradeoff would seem to indicate just buying a linear sensor chip. They are much, much easier to interface.

As to the issue of reliability, if you've got lots of them, then there's your reliability solution.
Quote: Originally posted by IrC  
Since the green lasers usually also have significant IR output would it make sense to add one of these after the laser so that only the green light made it into the rest of the assembly?
I doubt it. The IR band in a spectrometer is typically not shining on the CCD, so it won't introduce much noise if its internal optical table is moderately-well built.

Attachment: ICX282AQ, Diagonal 11 mm Frame Readout CCD Image Sensor, Sony.pdf (522kB)
This file has been downloaded 633 times

watson.fawkes - 10-3-2013 at 08:17

Quote: Originally posted by radagast  
Assuming that fluorescence doesn't increase equally across the wavelength range while cranking up the laser's power (and I don't see why it would), this might account for the background hump.
Fluorescence is the absorption and re-emission of photons, so its emission spectrum is generally independent of the pumping energy. Emission lines are doppler-broadened for each transition, so with (a) a room-temperature sample and (b) a spectrometer that doesn't discriminate with really fine resolution and (c) a laser with a broad emission spectrum and (d) and an organic sample with lots of transitions, you'll end up with a big hump like you're seeing. (a), (b), and (c) each contribute their own convolution width, and (d) means you're summing more than one Gaussian together.

What to do about it is another thing entirely. In the time-resolution paper I posted up-thread, they're using extraordinary sharp pulses and using a photomultiplier tube in single-photon mode to count. They're not doing simultaneous measurements at different frequencies; they're scanning. So that's not the easiest way to improve fluorescence rejection.

Other than improving the quality of all the components, the easiest thing to me seems to take advantage of the different ways that Raman scattering and fluorescence behave with respect to frequency. The Raman spectrum will be shifted when you change pump frequency (by rotating the prism); the fluorescence spectrum won't be. Some signal processing allows nulling out the bulk of the fluorescence. You can take spectra at different frequencies by monochromating the laser. Run the beam through a prism to disperse the green beam. Use a lens to put the rays parallel again after you've got enough separation. Then use a slit to select a pump frequency.

To estimate how well this might work, compute the wavenumber shift for ( 532 +/- 3 ) nm. That's what you can reasonable get out of a laser with an 8 nm emission width. That number is around 212 cm-1. At +/- 2 nm, it's 141 cm-1. At +/- 1nm, it's 71 cm-1. (Note that this is roughly 35 cm-1 per nm of pump wavelength difference.) The broadest peaks are aroung 50 cm-1 (roughly), so even a fairly small pump wavelength difference would suffice to get good peak separation.

watson.fawkes - 10-3-2013 at 08:31

Quote: Originally posted by IrC  
How much improvement do you think there would be to have the laser diode heatsinked with a thermoelectric cooler to reduce mode hopping and other effects which I believe could be considered as adding noise to the desired signal?
There are lots of sources of laser width broadening. This one would do something, but frankly I just don't know how much difference this one technique would make when compared to all the others that are possible. We're not relying on any of the coherence properties of a laser as an illumination source here, only monochromaticity. It would seem that building an optical train to monochromate the laser would be the cheapest way to narrow the wavelength width of the illumination.

The longer answer is that these 532 nm lasers are a three-stage system. They use (1) an IR laser diode to pump (2) an Nd-doped crystal whose output is then (3) run through a frequency-doubler. It's component (2), with all of the different transition lines, where the bulk of the frequency noise comes from. The issue is that it's the partition of energy in the Nd-excited states that matters. The thermal broadening of the individual Nd peaks is far less important as a source of noise than is the multiplet structure of the Nd transitions (see my post about with some of the spectroscopic data). Selectively pumping only a single transition in the Nd would seem to be the first thing to do, but that requires re-building the laser emitter system.

IrC - 10-3-2013 at 11:44

If you are designing a new highly expensive instrument with deep pockets building your laser from scratch sounds like a good idea. But if you are a poor bastard like me working with surplus the prefabricated diode is all that makes sense. Would in these circumstances it make more sense to build a filter as rigidly as feasible in poor bastard dollars such as below. Also what is the proper width of the two slits and could a grating with proper line spacing work? This includes the question what should the spacing be (in microns or nanometers, or even lines per inch if a grating?). Seems if frequency is varying and the laser power was high enough (so that enough light of proper frequency always reached the target), hopefully only the frequency desired would reach the target. Also I still don't know if adding the IR filter from a surplus DSC-F707 right after the diode (as we know usually there is much IR from the pump diode IIRC around 900 nm) would improve things. Seems to me reducing any IR would help in reducing signals caused by excitation of the target by undesired frequencies. Then again if the filter shown was added one would think the filter would stop the IR. Another thought is at what rate would optical power at the desired frequency vary? Slower than sampling times thereby removing it from concern? Seems like mode hopping would be much quicker than the window of sampling time.



Czerny-turner.jpg - 29kB


[Edited on 3-10-2013 by IrC]

watson.fawkes - 10-3-2013 at 15:56

Quote: Originally posted by IrC  
Would in these circumstances it make more sense to build a filter as rigidly as feasible in poor bastard dollars such as below. Also what is the proper width of the two slits and could a grating with proper line spacing work?
A monochromator made from an optical train would eliminate the need for an IR filter, because the IR wavelengths would be rejected along with everything less. There's a tradeoff between power, frequency selectivity, and sampling time, certainly, but it's not like amateurs need to run highly-accurate spectrum in less than a minute.

The monochromator chain in the illustration would work fine. It's built entirely from reflective elements (mirrors), but it would also be possible to use transmissive elements (lenses).

As for the exact parameters for designing a system, it's impossible to say categorically. The system actually needs to be designed. There are just too many variables to say anything, such as the focal lengths of the mirrors, the distances between elements, etc. I'd recommend learning the paraxial-ray approximation and the matrix method of analyzing them. It's a way using 2-by-2 matrices to model optical trains. It works very well for first-cut analysis of most optical systems. If you need more, for example to deal with chromatic aberration, you'll need frequency-specific ray tracing or its ilk. This method of analysis should be present in pretty much every introductory physics textbook on optics. An abbreviated treatment (adequate at least to get started) is in Building Scientific Apparatus. It's now in 4th edition, but previous editions are adequate for many purposes (except electronics, which always moves too fast).


IrC - 10-3-2013 at 19:37

Thanks Watson I'll do some study on that. Do you have a diagram laying around of the filter using lenses instead of mirrors. That one I have not seen. I picture a series or somewhat offset inline set of lenses (and slits?) where each following lens is set to only catch the light spread out unequally by frequency (chromatic aberration) of the desired wavelength. I had figured the IR blocker would not be needed with the filter as it would be outside of the slit. Never hurts to be sure though. I was assuming some of my blues really put out a lot of IR. Even with my glasses for the blue laser on I have noticed after cutting things for a while my eyes feel fatigued and a little painful similar to too much time up north in the snow on a bright day. Very similar. This led me to believe there is excessive light I cannot see which the glasses are not blocking. Whether IR (most logical) or UV I do not know. I can say the eye pain is similar to too much UV I have encountered in Arctic hiking. However I have in the past felt this with high powered IR before so it seemed most logical considering the pump laser in the blue diode assemblies.

Thanks for the PDF's. I had the service manual for the camera line but when I checked Sony a few years ago I could not find the PDF's you just found. I admit it would be much work but I had been thinking one does not need to address the entire CCD. Using the correct center scan mode and only looking at a portion of the vertical would simplify things. I was also thinking ignoring the green pixel data would be another level of blocking the laser from the desired signal. Of course if the material being studied had lines right in that area this would be a bad idea as information would be lost. Then again using a green source would seem to do this anyway, I assume this is why some common commercial units use IR, to give a wide spectral range for the desired information. Another thought was using the input for substrate bias like an electronic shutter. You are right though it would be a hell of a long and complex project.

The reason I was bringing up the IR filter was from study of the pic in my first post above. I could not see what else they meant by 'laser cleaning optic'. A bandpass filter would be expensive. Your idea is simply build a cleaner laser but I think this is outside the money and technical abilities (or at least equipment) of the average amateur. I assumed the filter radagast sent me is for blocking the green just ahead of the spectrograph. Since the pic was produced by an entity which builds these devices I could not fathom why the cleaning optic would be there if not needed. What is annoying is the lack of information and precise data nearly everywhere I search the subject. Nearly is too weak, pretty much every explanation out there is vague and generic.


[Edited on 3-12-2013 by IrC]

Marvin - 13-3-2013 at 16:52

The things that make this project so feasible are the spectrometer (200 USD is cheap), the cheap edge filter (45USD for the dichroic or the Schott OG550 for a little less), the wide availability of cheap DPSS laser pointers and reports from two people that it actually works. It may be best if further speculation on general methods are put with the rest in the big Raman thread.

IrC,
If you can afford the spectrometer I would recommend going with it, the default grating is even right for Raman. They are sparsely on ebay and should the supply dry up there is nothing close. It is possible with tweaking this setup will be good enough for useful chemistry. Otherwise you'll be working hard and spending money just to get a signal.

IR will go right through the low pass. The Stokes shifted light we want may be 4 or more orders of magnitude weaker than the returned laser line. IR Reflections, scratches on the grating etc may swamp the detector. Spectrometers aren't designed to provide this level of out of band rejection. IR may even produce it's own Raman spectrum with enough intensity to appear over the top of the signal we want (reflected by the grating at higher n unless we have an order sort filter).

We don't need an expensive bandpass filter for the laser, just something to block the 808 and 1064 lines. A regular IR filter may do the job. Trial and error will tells us what we can get away with.

Briefly about building a spectrometer with lenses, aside from chromatic aberration, every optic element introduces a new set of reflections. It can be done well with SLR camera lenses which avoid many problems but this is not a cheap option.

IrC - 13-3-2013 at 21:46

Only ones I've seen so far range from a grand to three. In my searching I did find some useful info and a few patents which might be worth reading.

http://www.oceanoptics.com/Applications.asp

Patents Title
7,149,033 UV visual light beam combiner
6,839,176 Composition and method of making high-reflection silver mirrors or thin-film optical filters
6,700,690 Tunable variable bandpass optical filter
5,381,505 Optical fibers with a light absorbing coating
6,638,668 Patterned filters for spectrographic imaging
5,711,889 Methodology to make optical filter arrays
4,884,697 Surface profiling interferometer
20090189513 LED using thin film dichroic filters
20090028756 Patches for non-intrusive monitoring of oxygen in packages
20080199360 Method and composition for a platinum embedded sol gel optical chemical sensor with improved sensitivity and chemical stability
20070122311 High performance materials for optical sensors for hydrocarbons environment
20070052961 Method for extending the color gamut for dichroic color mixing systems and colored gobos
20060092520 UV visual light beam combiner
Ag+ Coatings for improved reflectivity in VIS region Tunable variable bandpass filter

Forgot to ask but you didn't post a link. I did some searching here and I believe you mean "Laser-Diode based Raman Spectroscopy"?

From the prices I have seen searching thus far, I think I will try building my own spectrometer from the ground up. I have single lens reflex cameras and parts from years of rebuilding them as a hobby. Wonder if one of my 58 mm Zeis lenses would be useful in this. As for gratings 600 mm-1 and 1,000 mm-1 are two common choices. I have 531.49 mm-1 and 1,000 mm-1 so that is a start. What I absolutely cannot find are low cost slits, .25 um, 50, 100, and a few wider. I Cannot think of a low cost hacker way to build them with precision.


[Edited on 3-14-2013 by IrC]

D'oh

radagast - 2-4-2013 at 19:11

Welp, the good news is that I'm going to get more practice calibrating these DIY spectrometer units, and will document my next experience carefully as soon as I crack open my new spectrometer.

The bad news is: development on the raman spectrometer came to a screeching halt after I accidentally plugged my netbook power supply into my (prior) spectrometer unit and likely destroyed it. Let this be a lesson to y'all to not keep a cobweb of power cords under your workbench, which is exactly what I did.

BTW, before destroying the unit, I got a nearly-flat baseline spectra for aspirin by using a Radioshack potentiometer to reduce the current and thus the laser power.

@IrC: I believe Marvin has one of these DIY spectrometers already, so if you'd like to take a crack at rescuing this old spectrometer and using it yourself, I'd be happy to mail it to you sometime this month for free. I looked inside but couldn't find any fuse, and have a feeling that any repair is simply beyond my circuit trouble-shooting abilities, but I bet you can do better since you know more about circuits.

IrC - 8-4-2013 at 17:17

This is a question for Watson. I found this Color Light Sensor for 5 bucks for the Arduino. 10 bit resolution and real easy programming. I wondered if it would not be too insane to consider maybe gluing the sensor off center to a small dynamic mike element. If the other side was held rigid, a small microvolt DC voltage (or a sawtooth?) could make it scan a line across the grating. What do you think?

https://www.sparkfun.com/products/10656

No doubt other ideas to make it scan exist, I was just thinking about the complexity of writing software to address a CCD as opposed to trying something like this. They even provide sample code and other useful info for free which make it very easy to get data from this sensor.

So have I gone off the hacker deep end or do you see this as something useful?

watson.fawkes - 8-4-2013 at 17:38

Quote: Originally posted by IrC  
I wondered if it would not be too insane to consider maybe gluing the sensor off center to a small dynamic mike element.
The photodiode on that chip is 420 microns square, per its data sheet. The width of the pixel on the ILX511 is 14 microns, a factor of 30 smaller. It's conceivable, barely, that you could deconvolve the signal from a moving fat pixel and get reasonable data out. I don't recall off the top of my head how the error analysis goes, but I'm not optimistic. I don't recommend it. A $45 linear CCD chip is a much better investment than a $5 sensor and a bunch of electromechanical bother. (Plus it's a BGA device; I hope you can get it mounted.)

IrC - 8-4-2013 at 19:50

I have 3 Sony CCD's for the DSC-F707/717 which was my plan. Nice to have a couple backups since I have a dozen of the still cameras, and the CCD likes to go black far too often. Only cure is a new CCD. Never know when one will be needed, I have had to replace 3 of them so far as it is. I saw this simple $5 sensor and how easy the code was so I had to wonder about it. Still stuck trying to find any kind of optical slits. Damn impossible to find and harder to make with precision. Would be nice if one existed with a micrometer adjustment at low cost. Honestly I cannot find any reasonable source whether fixed or adjustable for the hardware hacker, even at unreasonable cost. I wish I could think of a good surplus cheap source for optical slits.

CCD-717.JPG - 50kB

watson.fawkes - 9-4-2013 at 04:40

Quote: Originally posted by IrC  
Still stuck trying to find any kind of optical slits. Damn impossible to find and harder to make with precision.
I have heard that disposable razor blades word well, though I haven't tried it. They've adequately straight, which is the harder part. Getting them aligned parallel is much easier. The human eye is exquisitely sensitive to parallelism through a thin slit. I read a fascinating story years ago (don't recall of in print or online) about Starrett's tool room where they manufacture squares. It was a dim room, lit only with the monochromatic test lights on the benches, used to illuminate the slit between an inspection square (the reference) and the workpiece. As I recall, they could easily get to a micron or two deviation at the end of a 6" square.

IrC - 9-4-2013 at 10:58

Believe it or not two single edge razor blades was my very first thought. To make it adjustable would require a screw pitch finer than anything I have yet seen. Just fixing them in some mounting structure which could be adjusted to the right parallel gap then holding them unmoving while clamping in place seems difficult. I had been thinking about a bias voltage on a piezo element with very precise op amp error control with possibly using a light sensor to give feedback by monitoring light through the slit. Obviously more is wider and less is closer. Yet even keeping a light flux that steady looks difficult at best when your talking about the need to maintain a gap in a micron range. I have thought of many different approaches but I keep coming back to the lack of the proper items on a hobbyist budget. Lately I have been thinking of ways to scan an ion beam in my deposition experiments. In effect painting a mirror on a glass or quartz plate with a slit in the middle. What I run into here is again my lack of money for the kind of high tech equipment required, as right after this thought came the question. How can I avoid statistical deviation of metal ions in a scanned beam which would land in the micron range in places other than where I wished. All this brought me back to uselessly searching for a low cost factory made pair of slits.

Honestly the best hope so far is the misfortune of radagast and His offer to send me the apparatus He damaged. Combined with my great experience of repairing electronics with no hope of service data or parts availability who knows I may yet prevail.

I still wish someday to build from the ground up a working Raman Spectroscope apparatus. Either that or settle for a dinner made from bfesser's Raman noodles.

unionised - 9-4-2013 at 11:04

A pair of razor blades on a ring shaped magnet is a "classic" solution to the problem

IrC - 9-4-2013 at 13:07

Quote: Originally posted by unionised  
A pair of razor blades on a ring shaped magnet is a "classic" solution to the problem


Thank you that is a very good idea. Avoids many problems. I can see I need to spend a lot more time reading 'classics'. Oddly in many searches online using terms such as "optical slit construction" or "design" I never found that mentioned. Which makes me wonder. So many websites with theory on so many topics. What would be nice is more which focus on actually building things along these subject lines. I have found several patents which typically (as all patents do) tell enough to cover an idea legally but provide little to aid in actually building a thing. I did find a page at Edmund, have not been to their site in a long time. At $98 each still fairly expensive.

http://www.edmundoptics.com/optomechanics/apertures/pinholes...

Here are some patents I have found, have not had time yet to study them all.

2705440 Slit mechanism for monochromator
2964998 Precision light aperture arrangement
2914987 Slit mechanisms
2755707 Slit control mechanism for optical instrument or the like
3211056 Parallelogram slit structure for a monochromator
3537777 MICRODENSITOMETER BILATERAL ADJUSTABLE FIELD SLIT SCANNING APERTURE
3645630 MONOCHROMATOR
3860328 MEASURING ATTENUATOR FOR AN OPTICAL NULL SPECTROPHOTOMETER
3806251 METHOD OF MEASURING SMALL OBJECTS
3886331 Electronic scanning spectrophotometer system
3937580 Electro-optical method for measuring gaps and lines
4017162 Adjustable slit mechanism
4051502 Multiple element focal plane shutter for photographic cameras
4122461 Exposure apparatus and method for manufacturing a cathode ray tube display screen
4300835 Attenuator for stray light produced in monochromators
4316661 Electromagnetically operated shutter
4320950 Electromagnetically driven shutter
4325634 Slit width calibrator for monochromator
4455088 Monochromator with concave grating
4526470 Stray light measurement and compensation
4540282 Apparatus for optically analyzing a sample
4692883 Automatic digital wavelength calibration system for a spectrophotometer
4717254 Stray-light suppressor for Littrow spectroscope
4975919 Laser wavelength control apparatus
5128549 Stray radiation compensation
5206765 Precision slit of adjustable width for a spectral instrument
5384662 Precision optical slit for high heat load or ultra high vacuum
5497230 Spectroradiometer
5661589 Bilateral slit assembly, and method of use
6507398 Czerny-turner spectroscope
6956688 Variable width optical slit mechanism

I know this should all probably be in the other thread on the subject but it's not like much discussion has been going on in this one lately anyway.

I will be ordering a few N50 ring magnets and grabbing some blades as I really like your suggestion. Or maybe N48:

http://www.ebay.com/itm/10-Neodymium-Magnets-3-4-x-1-2-x-1-8...

More useful information:

http://bwtek.com/spectrometer-introduction/

http://www.spectra-magic.de/Framepage-E.htm

http://www.horiba.com/scientific/products/optics-tutorial/mo...

http://opticalea.com/Spectrometers.html

http://astronomyonline.org/exoplanets/Exoplanets.asp

http://www.thespectroscopynet.eu/?Spectrometers:Monochromato...

http://www.thespectroscopynet.eu/?Data_Acquisition:Calibrati...

Radagast you mentioned the difficulty of calibration, maybe this last link will be useful.


[Edited on 4-10-2013 by IrC]

radagast - 30-4-2013 at 19:01

Quote: Originally posted by watson.fawkes  
Quote: Originally posted by radagast  
Assuming that fluorescence doesn't increase equally across the wavelength range while cranking up the laser's power (and I don't see why it would), this might account for the background hump.
Fluorescence is the absorption and re-emission of photons, so its emission spectrum is generally independent of the pumping energy. Emission lines are doppler-broadened for each transition, so with (a) a room-temperature sample and (b) a spectrometer that doesn't discriminate with really fine resolution and (c) a laser with a broad emission spectrum and (d) and an organic sample with lots of transitions, you'll end up with a big hump like you're seeing. (a), (b), and (c) each contribute their own convolution width, and (d) means you're summing more than one Gaussian together.

What to do about it is another thing entirely. In the time-resolution paper I posted up-thread, they're using extraordinary sharp pulses and using a photomultiplier tube in single-photon mode to count. They're not doing simultaneous measurements at different frequencies; they're scanning. So that's not the easiest way to improve fluorescence rejection.

Other than improving the quality of all the components, the easiest thing to me seems to take advantage of the different ways that Raman scattering and fluorescence behave with respect to frequency. The Raman spectrum will be shifted when you change pump frequency (by rotating the prism); the fluorescence spectrum won't be. Some signal processing allows nulling out the bulk of the fluorescence. You can take spectra at different frequencies by monochromating the laser. Run the beam through a prism to disperse the green beam. Use a lens to put the rays parallel again after you've got enough separation. Then use a slit to select a pump frequency.

To estimate how well this might work, compute the wavenumber shift for ( 532 +/- 3 ) nm. That's what you can reasonable get out of a laser with an 8 nm emission width. That number is around 212 cm-1. At +/- 2 nm, it's 141 cm-1. At +/- 1nm, it's 71 cm-1. (Note that this is roughly 35 cm-1 per nm of pump wavelength difference.) The broadest peaks are aroung 50 cm-1 (roughly), so even a fairly small pump wavelength difference would suffice to get good peak separation.


Thanks for the clear explanation, Watson!

I haven't made much progress lately since I've been tied up with other matters, but have moved the apparatus to a smaller breadboard (6" x 6"), and bolted it to small black Pelican Protector 1300 case. I've also added a lab clamp to make it easier to hold a sample vial.

Although these changes aren't exactly earth-shattering, they make the unit completely portable, since you can just close the case and carry it like any other Pelican camera-case.

Time-permitting, I'm plan to calibrate the replacement DIY spectrometer with a neon light by this weekend and document/post the process.

Top View_Reduced.jpg - 360kB Long View_Reduced.jpg - 353kB

radagast - 19-5-2013 at 16:19

Because the spectrometer resolution varies across the wavelength range, the calibration process requires multiple sources of calibration lines. The most straightforward way would likely be to shine different lasers simultaneously while calibrating the spectrometer and adjusting each laser line so each line is sufficiently resolved. This process requires compromise, since sharpening the resolution on the lower wavelengths will decrease resolution on the upper wavelengths, and vice-versa.

Since I only had a 532 nm laser (a 650 nm laser is on the way), I decided to use a Radioshack neon indicator lamp (with characteristic emission lines from 585 nm to 743 nm) simultaneously with a 532 nm green laser. To mount the neon lamp on a Thorlabs post, I used a 1/8" drill bit to bore a hole into a 3/8" nut, and then tapped it with an 8-32 NC tap.

After adjusting the collimating mirror for a while with a pair of tweezers, I achieved ~2 nm on the upper end of the spectrum with higher resolution on the bottom (roughly the same resolution as my prior spectrometer before I broke it). Although I devised some general rules for reaching that resolution, the process is time-consuming and not reliably repeatable, and the resulting resolution is still 2x less than the spectrometer's capabilities. Thus, I've ordered micrometer linear and rotational stages to see if I can achieve better resolution and formalize the process.

IMG_0045.JPG - 334kB
IMG_0047.JPG - 397kB
IMG_0056.JPG - 415kB
IMG_0062.JPG - 451kB
IMG_0063.JPG - 422kB

Dr.Bob - 21-5-2013 at 05:23

I am really impressed with the instrumentation made by people here on SciMad. I have done a few simple repairs and tweaks on instruments, but the level of knowledge here is really amazing. Building a home IR spectrometer is pretty awesome. Just wanted to congratulate the people here that are working on these, that is just great. Same for the NMR and other items I have seen described on other threads.

Maya - 23-5-2013 at 11:45

Dr. Bob,

I've searched the threads and don't find anything on people building IR Specs, obviously just Raman Specs.
Where have you found the IR spec ref?

radagast - 23-5-2013 at 14:13

Quote: Originally posted by Maya  
Dr. Bob,

I've searched the threads and don't find anything on people building IR Specs, obviously just Raman Specs.
Where have you found the IR spec ref?


I remember reading that len1 resurrected an IR spectrometer by essentially reverse-engineering it over the course of several months. I don't know that anyone has built an IR spectrometer from scratch, though, which would be a truly impressive feat.

I acquired an old Biorad FTIR microscope for a nominal sum recently, for the purpose of stripping the electronics away and using it as the starting point for a DIY FTIR. The dessicant is a bright pink, so I'd bet that the beam-splitter is busted. But that's a long-term project that I don't expect to start for a few years, at least. I spoke with the very nice gentleman who runs FTIR.com, who informed me that it's just very difficult to resurrect certain old FTIR models, but that there is a small group of individuals who are interested in that sort of thing . . .

If anyone has actually constructed a dispersive IR spec from Thorlabs legos, I'd be very interested in learning more!

Minor Updates

radagast - 27-5-2013 at 20:11

(1) Ben Krasnow of Valve and Youtube fame has unveiled his DIY raman spectrometer, which uses a camera as the CCD unit -- definitely worth a watch:

http://www.youtube.com/watch?v=tRrOdKW06sk

(2) I'm still perplexed at how to measure liquid spectra in a 90-degree configuration. I used a variety of round vials and square glass cuvettes in different positions and couldn't get a viable raman signal, even when using a strong raman scatterer like toluene. I must be missing something, since I searched the literature and couldn't find any reference to a 90-degree configuration having problems measuring liquid spectra. If anyone has any thoughts on solving this issue, I'd be eager to hear them.

(3) The back-scattering configuration is now producing some very pretty spectra of toluene. Based on the inability of the (current) spectrometer to resolve the toluene doublet at arond ~1000 cm-1, however, it's clear that the spectrometer's resolution is fuzzier than 35 cm-1, which is consistent with the floor of the theoretical calculated resolution of the spectrometer (I say "floor" because I don't know the linewidth or mode of the laser, so the resolution could be substantially fuzzier). If I had to guess, I'd say that it's operating at 50-60 cm-1 or so (depending on wavelength), which is sufficient for spectral matching, crude structural insight, and distinguishing between, say, ethanol and methanol.

Ideally I'd like to drive it down to 20 cm-1, as I feel like that's where it turns from an amusing toy to a seriously useful tool. That may or may not be possible with the present spectrometer but we will see whether I can align the spectrometer more accurately with the micrometer stages to get closer to that . . .

[Edited on 28-5-2013 by radagast]

unionised - 28-5-2013 at 00:43

Could you check up on the link?
I get a "video does not exist" error message.

radagast - 28-5-2013 at 06:23

Quote: Originally posted by unionised  
Could you check up on the link?
I get a "video does not exist" error message.


Edited the link (it should be http://www.youtube.com/watch?v=tRrOdKW06sk).

One thing I really like about Ben's design, besides the use of a camera, is that he mounts the laser apparatus and beamsplitter all in one tube. This raises the possibility of making an enclosed raman probe shaped like a gun, where the grip houses a laser, the muzzle houses the microscope objective, and the backscattered light is reflected to the back of the "gun" and collected through a notch filter. Assuming that there's no problem with sampling in ambient light, this would make the apparatus considerably cheaper since you wouldn't need an enclosure or optical breadboard. Plus, having a point-and-shoot probe would really enhance the convenience of sampling a chemical.

I also forgot to mention that you can communicate with the spectrometer through minicom. I used the Mac OS X port of minicom to do so, but I'd imagine you could use linux or windows' versions, too, which opens up the option of writing a PySerial script which will automatically take a variety of integration times (this would require integration with the laser, since you need to reset and recreate the dark background every time you switch the integration times).

radagast - 1-6-2013 at 17:34

So my Dad once told me that nothing makes a physicist or physical chemist angrier than messing up their aligned hardware, and I now really know what he meant, in the way that you can only understand if you've spent hours pushing a collimating mirror back and forth with a pair of tweezers. The result is that I'll get motivated every week to work on it, and then quit in rage after a few hours of tinkering, only to get interested in it the next weekend.

Alignment is difficult because optical components of the DIY spectrometer are not mounted on any kind of translation table, which makes sense given its price. The collimating mirror mounted on a single screw like a compass. In other words, it can spin around and you can also adjust the distance between the "needle" of the compass (the mounting screw) and the other components on the optical bench. The other components in this cross-czerny design (the CCD, the diffracting grating, and the focusing mirror) either have less degrees of freedom or are mounted in a manner that makes them much less frustrating to align.

In search of an easier way to align the collimating mirror, I built a Rube-Goldberg-esque apparatus, whereby a micrometer linear translation stage stacked on a rotational stage is suspended above the spectrometer. After attaching a standard optical post onto the stages, I then cut a piece of 8-32 threaded rod to the appropriate length, and used a glue gun to temporarily attach the collimating mirror to the rod. So configured, the apparatus permits precise adjustments to the collimating mirror. I'm still working out some issues with the design, but so far, it seems to make the alignment process much easier.

GKau_Contraption.jpg - 511kB

[Edited on 2-6-2013 by radagast]

smaerd - 2-6-2013 at 04:19

Very nice. Apparently this is how a lot of rotating diffraction gratings and other such optical parts are managed. A small linear actuator pushes/pulls one side of a mirror or grating, etc.

I'm really really impressed with the work everyone has done on this. Ben Krenshaws set up is pretty nice.

Quick question how much did the optical bread board set you back(the one in the last picture)?

[Edited on 2-6-2013 by smaerd]

radagast - 2-6-2013 at 11:24

Quote: Originally posted by smaerd  
Very nice. Apparently this is how a lot of rotating diffraction gratings and other such optical parts are managed. A small linear actuator pushes/pulls one side of a mirror or grating, etc.

I'm really really impressed with the work everyone has done on this. Ben Krenshaws set up is pretty nice.

Quick question how much did the optical bread board set you back(the one in the last picture)?

[Edited on 2-6-2013 by smaerd]


Thanks, Smaerd! I'm monitoring your own polarimeter project with great interest. In addition to being a terrific and useful project, some issues you're addressing there re: mechanical servos directly apply to my future plans here, which include building a motorized XY-translation translation stage to enable raman scanning of a developed TLC plate.

As far as the breadboard, there are actually two in the picture: a 8x8" black breadboard manufactured by Thorlabs (the base, which I bought for ~$60 off Ebay) and a smaller 6x6" aluminum breadboard of unknown origin (the ceiling, which I got for $40 off Ebay).

Since then, I've found that you can get substantially cheaper and smaller aluminum breadboards. Although it's unclear whether they were machined for optical use, I found the following types to be suitable (4x4 for about $20):

http://www.ebay.com/itm/Aluminum-Mini-Breadboard-4-in-x-4-in...

I bet you could also machine your own breadboard by drilling and tapping with a drill press and tap kit.

More Spectrometer Hilarity

radagast - 15-6-2013 at 13:01

So after getting quite close to my goal to optimize the spectrometer for the range from ~532 nm to ~700 nm, my second DIY spectrometer abruptly stopped working. It was still controllable through the serial port, and would appear to scan spectra, but no data would actually appear. I suspect this might be because I took pictures of the setup using a camera with a flash, which perhaps damaged the CCD, although that would be pure speculation on my part. In any event, the CCD is easily swappable, so I ordered a refurbished ILX511 CCD (~$15) off Ebay to further test this hypothesis. (BTW, IrC, sorry for the delay -- I intend to send out the first broken spectrometer shortly).

Before that, however, I hit upon the common-sense notion of using two lasers simultaneously to align the spectrometer while watching the physical CCD itself instead of focusing on the resulting spectra. Here, I used 532 nm (green) and a 650 nm (red) lasers, and tweaked the setup until the laser lines (1) both appeared on the light-sensitive strip; and (2) appeared as vertical lines that are as thin as possible.

Since a poorly-aligned laser will show up as a fuzzy dot or thick line on the CCD, it's pretty easy to tweak it until it appears to be a razor-thin vertical line. More difficult is to get two laser lines to both show up as razor-thin vertical lines, since you can usually get one line very thin, but the other one will either thicken or fall off of the CCD altogether.

Thus, narrowing both of the lines' thickness requires alternating between moving the collimating mirror forward and backward, which in turn affects the vertical placement of the laser lines and requires adjustment to the focusing mirror. Reiterating this process, as the Science Surplus notes say, yields better and better results.

If we're only interested in anti-stokes emissions, you might ask, why not align the spectrometer so the green line appears just to the left of where the CCD begins to both maximize the spectral resolution and eliminate Rayleigh interference? My answer is that the DIY Science Spectrometer manual says that the first hundred pixels are pretty much useless. Consistent with that explanation, I've found that if you try to use the first hundred pixels (~the first cm or so of the CCD), you just get a wacky spectra instead of a clean narrow laser spectra.

IMG_0080.JPG - 431kB
IMG_0081.JPG - 449kB
IMG_0082.JPG - 431kB
IMG_0083.JPG - 434kB
IMG_0084.JPG - 361kB
IMG_0085.JPG - 704kB

[Edited on 15-6-2013 by radagast]

[Edited on 15-6-2013 by radagast]

CCD Driver Circuit

radagast - 27-7-2013 at 15:35

Given that I have a special knack for destroying perfectly good spectrometer circuits, I took another stab at building the spectrometer (as opposed to just the raman probe) out of Thorlabs parts, a TCD1201D CCD, and an Arduino Uno. The electronics are quite affordable (the CC was about four dollars and the Uno cost about $15).

The good news is: I uploaded an Arduino sketch written by "Freddy_Dreddy" (a poster at an Arduino forum) to an Arduino Uno connected to a TCD1201D (the very chip suggested by Harristotle in his GC thread), and the output to my oscilloscope shows that the CCD is indeed giving off information based on the intensity and location of light on the CCD.

However, the Arduino sketch doesn't appear to communicate any information through the serial port, and much of the program is written using port manipulation (presumably for speed) and is otherwise much more advanced than your average Arduino sketch.

I'll post more as soon as I can either decipher the code, and get Freddy_Dreddy's permission to post his or her code here.

smaerd - 28-7-2013 at 12:18

I'd be interested in seeing the code as well. Theoretically for the op-amp circuit in my polarimeter project I will be passing enough information through the serial pipe in just enough time for the next interrupt. ... In reality I'll probably be coping with a mess hahaa.

Awesome work so far though. You're spectrum two posts back looks wonderful.

IrC - 28-7-2013 at 16:17

radagast "(BTW, IrC, sorry for the delay -- I intend to send out the first broken spectrometer shortly)."

No problem. Back in May I ordered some N48 ring magnets and straight razor blades to try the idea from unionised, but I must not be very coordinated it is still very hard to align them. When closer than a sheet of paper apart they keep sticking together. Probably should have used weaker magnets but my thought was to rigidly hold them. Must be the polarization direction is not perfectly axial as I thought, causing opposing blade edges to attract each other. I can see why the optical slits I have located are too expensive for a hobby project. Those things are very hard to build with great precision. I would like a method to easily adjust the gap on the fly as it were, say with a micrometer type adjuster. Sitting here looking at this DVD player which has trouble reading from a weakening laser I am getting ideas about robbing it for the laser lens focusing assembly. If I were to cut two small pieces of blade and fix them so the moving coil connects to one of them possibly with the right electronics I can drive the coil thereby adjusting the gap. As well maintaining it at all times so I can control the gap on the order of wavelengths of the laser frequency. Going to study this for a while to determine if this idea is worth working on.

radagast have you looked at this page (link below)?

http://creative-technology.net/MAKE.html

Or this, Spectruino, an Arduino Spectrometer

http://myspectral.com/




[Edited on 7-29-2013 by IrC]

smaerd - 28-7-2013 at 17:45

Not sure what width of a slit you need but I did some looking here's some things I've found:
second hand:
http://www.surplusshed.com/pages/item/m1570d.html
DIY option:
http://www.astrosurf.com/aras/bareges/bareges_en.htm

I like the DIY option it appears as if they use two finely threaded screws and push two blades together with them.

IrC - 28-7-2013 at 18:40

Quote: Originally posted by smaerd  
Not sure what width of a slit you need


Neither am I this is my first attempt at working with Raman Spectroscopy. Seems the useful signal is so low I thought the best idea was to be able to finely adjust the amount of laser light getting through the slit. Thanks for searching for the links. I search them often as their inventory rapidly changes. Found that item after they had run out of stock though. They recently had a big clearance sale which just ended. Was very annoying to see the sale go by while they had nothing of interest. Think I have spent at least a grand with them over the last few years but have yet to find a good adjustable slit in stock. To be honest when I first looked at the one you mention I thought the threads on the adjuster looked too coarse for extremely fine adjustments. However at that price I would have bought a few if I had seen it in time. Just to try rebuilding one with a real fine pitch screw. Adjustable slits must sell fast since I have never seen one come in and still be in stock by the time I notice it. The CCD Spectrograph link is great. I had not found that one before in my searching.



Harristotle - 29-7-2013 at 04:18

Hi Radagast,
one thing to check is the serial connection setup

find the line Serial.begin(.

Check to see that the speed of your serial port is set to the same.
If you are chucking image data back and forwards, programmers often crank up the serial speed past the default 9600. This means that you see diddly on the serial port on your monitor.
BTW, I have designed a pcb for the TCD1201D, and etched and built it. But not tested, I ran out of time (last solder at 10:30 on sunday night before start of term), and I will work on this after I finish my chromatograph. I was looking at mangling GastonLagaffe's code from the same forum. I also modified the reference circuit so that the you only need 1 clock signal from the arduino, and so that the "antiphase" clock signal was done with a spare pair of inverters that were on a 74hc04 chip I had used to interface the TCD1201D to the arduino.

Cheers,
H.


Quote: Originally posted by radagast  
Given that I have a special knack for destroying perfectly good spectrometer circuits, I took another stab at building the spectrometer (as opposed to just the raman probe) out of Thorlabs parts, a TCD1201D CCD, and an Arduino Uno. The electronics are quite affordable (the CC was about four dollars and the Uno cost about $15).

The good news is: I uploaded an Arduino sketch written by "Freddy_Dreddy" (a poster at an Arduino forum) to an Arduino Uno connected to a TCD1201D (the very chip suggested by Harristotle in his GC thread), and the output to my oscilloscope shows that the CCD is indeed giving off information based on the intensity and location of light on the CCD.

However, the Arduino sketch doesn't appear to communicate any information through the serial port, and much of the program is written using port manipulation (presumably for speed) and is otherwise much more advanced than your average Arduino sketch.

I'll post more as soon as I can either decipher the code, and get Freddy_Dreddy's permission to post his or her code here.

CCD Mania

radagast - 2-8-2013 at 21:31

@smaerd

Thanks, Smaerd -- looks like we're working toward similar goals! Love your polarimeter project; it looks very promising. The spectrum two posts ago were unfortunately the last spectrum that the second DIY spectrometer took; replacing the ILX CCD there didn't help so it looks like I'm SOL.

@IrC

I was really interested in the MySpectral project, but as far as I can see, they haven't released any of their code or hardware yet, and the resolution (7-9 nm) is too low to get good raman peaks. I'm checking out the other link . . . looks interesting.

@Harristotle

Sounds like you're making some serious progress on the TCD driver circuit; I'm eager to hear more about the PCB and your one-phase hack! Thanks for the heads-up on the baud-rate -- I had set the serial monitor to 115K (as specified in the code), but still didn't see anything come out in the serial monitor. Instead of using the Serial.begin wrapper, Fred wrote directly to UDR0, possibly to save clock cycles.

(Gaston's code, which uses Serial.begin, outputted data visible on the serial monitor, but it was incomplete).

@all

I got permission from Freddy_Dreddy on the Arduino Forum to post his code for the TCD1201D here (attached as "SerialCCD.rar"). The .c file compiles on the free Arduino IDE, except that I had to change the following line from:

if(i < 3) str = "\0";

to:

if(i < 3) str = '\0';

I was excited to see the code work, and created a little youtube video documenting my adventures with it here:

http://youtu.be/n26SK40u_To

Attachment: SerialCCD.rar (25kB)
This file has been downloaded 447 times

Harristotle - 2-8-2013 at 22:24

Hi Radagast,
Here are the eagle files that I have been working on.

They are based on the reference design in the datasheet, with the one modification of using a single clock. I haven't yet tried this, the board is still on my shelf. Working crazy hours, and writing papers, education programs, grants and other stuff.

Please feel free to play with this, I release it gladly to the commons! There is little enough that is my cleverness in it anyway.

By the way, looking at the timing diagrams, if this doesn't work, my next step would be to place a capacitor between the two inverter gates and ground to introduce a small phase delay to make it exactly like the reference sheet. But this one may work!

Must go, In Australia "Science Week" is coming up fast in schools. The theme is "A century of Australian Science". I am trying to work up a paper chromatography separation of vegemite as a contest (best Chromatogram wins a $20 itunes voucher). But I am getting a lot of smearing. Am trying to see if I can centrifuge out some particulates, and optimise the solvent system. We are using Berocca vitamin pills as standard. Literally my 'fuge has just stopped. Must go! ......



Take care and have fun, I really look forward to see your results.

Cheers,
H.

Attachment: spectrometer1201d.brd (17kB)
This file has been downloaded 692 times

Attachment: spectrometer1201d.sch (203kB)
This file has been downloaded 722 times


bfesser - 19-12-2013 at 06:01

Quote: Originally posted by Harristotle  
The Raman will take longer, my lowpass filter is no good, I will need to get another. I am happy to share what I have so far, ask and ye shall recieve!
I'd love to see any progress you've made on this, since your last post in August.

SM2 - 19-12-2013 at 07:25

Another great place to source a cheap but excellent beam splitter is from some of the members on http://laserpointerforums.com/. They are very helpful, and will get you what you need at a very affordable price.

aga - 17-4-2014 at 00:56

Quick thought on a variable Slit :-

Assuming that the razor blades are as straight as anything will ever be, you could mount 2 of them vertically on a disc (maybe a gear with pre-drilled centre hole) with each one being offset from centre by a small amount.

Looking from the side, the width of the apparent 'slit' will vary depending on how the disc is rotated.

Your work on this is wonderful. Truly inspiring.

[Edited on 17-4-2014 by aga]

aga - 28-4-2014 at 12:39

Quote: Originally posted by IrC  
closer than a sheet of paper apart they keep sticking together


Stick a couple (or a few) small neodymium magnets to the razor blades.
Vary the distances, and polarities.
Should be possible to set the Gap distance that way without the magnetised blades sticking to each other (untried idea).

An alternative is green rizla cigarette papers : 0.01 mm thick.
Use them as packing at the top and bottom of the razor edges.

I still use a folded rizla paper to set the meshing of the pinion to the drive gear on model helicopters. Works a treat.

aga - 15-8-2014 at 16:07

Yo!

Tell us how it is going.

There should be a separate or sticky for this.
It is important.

Spectrum studio CSV file modification

Neotronic - 4-10-2014 at 22:58

Hello to everyone, this is my very first post on this forum.

I just want to share my modification of Science Surplus Spectrometer. Instead of calibration of squeezing the spectrum into wavelenght range, my approch was, Calibrate the spectrum adjusting the bottom pixel numbers as wavelenght. Basically in the graph normally you see wavelenght from 0 to 2048. I have modified the CSV. file to show from 383nm - 642nm. I have created a exel file to modify this range, and using Osram and Neon i have calibrated this table. For thouse who find this solution helpfull i can share and explain how it is possible. My another modification was to make this spectrometer possible to use internal and external CCD Linear Sensor IC SONY CDIP-22 ILX511 but that will be another post.. :)


Osram full.png - 94kB Neon zoom.png - 120kB Neon full.png - 106kB

aga - 5-10-2014 at 09:11

Welcome, and nice effort.

What do you use for a Filter ?

Neotronic - 5-10-2014 at 09:24

I don't use any filter, The light from osram fluorescent bulb, and neon bulb is comming direct via optical cable to Spectrometer. The slit is 50um.

aga - 5-10-2014 at 11:07

I meant for the actual Raman back-scatter thing.

What filter do you use for eliminating the Source/excitation light ?

Polverone - 3-12-2014 at 13:30

Looks like someone figured it all out. The ramanPi: http://hackaday.io/project/1279-ramanpi-raman-spectrometer

DistractionGrating - 3-12-2014 at 15:32

Sweet!

smaerd - 9-12-2014 at 17:14

That is hands down the most impressive open source project I've seen to date. Now I want a 3-D printer... Would make aquiring a lot of the components necessary for all these projects so much more accessible.

IrC - 10-12-2014 at 01:05

Quote: Originally posted by Polverone  
Looks like someone figured it all out. The ramanPi: http://hackaday.io/project/1279-ramanpi-raman-spectrometer


This project is so complex it is not easy to quickly locate additions, I cannot find calibration info (http://hackaday.io/project/1279/instructions) beyond this : Section: 9 Spectrometer Calibration has not been written yet.

Appears not finished yet, nor can I find information as to when it will be. Also I wonder what radagast is doing it's been well over a year since any word or posts but it appears at least he did check in about a month ago.

I should have listened to bfesser and picked up a RaspberryPi a year or two ago so I would have a head start on using one. Sticking to my Pics, Arduinos, ATMegas and whatnot I have spent zero time learning about the Pi. More work to do. I did download the 114 mb file ramanSpectrometer-master.zip. Need to unzip and look at these files. I see no reason a skilled craftsman cannot replicate important components without investing in a 3D printer. From all the pictures this looks feasible yet incredibly involved. May make alignment (http://hackaday.io/project/1279/log/9812) more difficult, not sure yet. The large zip contains .stl files but dealing with files for 3D printers is not something I have experience with yet. I assume they are in a code format for the printer, can anyone say if these .stl files can be converted into blueprints for study? Doubt I'll have the money to buy a 3D printer anytime soon.

aga - 10-12-2014 at 02:43

Wow ! That is an amazing find Polverone.

The biggest cost always seems to come down to the filters.

Has anyone experimented with making their own edge/notch filters, or is that seriously difficult ?

IrC - 11-12-2014 at 12:40

FYI on the .stl files.

http://www.hive76.org/rendering-stl-files-to-images

https://github.com/robottrouble/STLRender

http://www.robottrouble.com/projects

http://www.povray.org/

aga - 11-12-2014 at 14:44

OK. Here's an idea for removing much of the Source wavelength.

How about a beam splitter Before the laser enters the sample, then a moveable pair of mirrors taking the other 50% to be recombined with the Raman scattered vector, then adjust the mirrors to be exactly 180 out of phase with the source.

Plus some kind of simple filter to adjust the 'brightness' to match that of the original source, which is always present along with the Raman component.

By adjusting the path-length, the second beam (theoretically) could be made to destructively interfere with the original laser light, effectively cancelling it from the image, leaving Only the Raman spectra.

Yes, a Notch filter would be much easier, yet financially out of reach of most amateurs.

Marvin - 11-12-2014 at 17:48

Don't forget that cancelling out a beam violates conservation of energy. Practical methods result in reflecting dichroic filters. The Schott filter (OG550 was it?) is pretty inexpensive.

There are a lot of ghetto methods but the spectrum is only a fraction of the way to a useful instrument. It may be that a version with poorer specs made from cheap mass produced parts would be much more use than a few one off units operating independently. It may be that every tiny flaw in a home made device is going to cause merry hell getting comparisons to every other unit.

The RamanPi I think has yet to produce it's first Raman spectrum, I stopped following it closely when the pages became so large they crashed my netbook.

IrC - 11-12-2014 at 20:53

Quote: Originally posted by Marvin  
The RamanPi I think has yet to produce it's first Raman spectrum, I stopped following it closely when the pages became so large they crashed my netbook.


I have gone through pretty much everything and believe the answer is still no but 8 days ago FlatCat wrote: "I'll be re-writing the firmware for the imagingBoard in the next few weeks". The calibration is not yet written either so I doubt useful data on a real test has been produced thus far. You are right this project is massive. Right now he has a couple other projects on the main menu page I am finding even more interesting and less costly.

aga - 12-12-2014 at 00:50

Quote: Originally posted by Marvin  
The Schott filter (OG550 was it?) is pretty inexpensive.

Thanks for the filter code.

A quick ebay popped up a surprise :

http://www.ebay.co.uk/itm/550nm-Filter-for-IPL-Beauty-Treatm...

John Green - 4-2-2015 at 14:48

Quote: Originally posted by aga  
Quote: Originally posted by Marvin  
The Schott filter (OG550 was it?) is pretty inexpensive.

Thanks for the filter code.

A quick ebay popped up a surprise :

http://www.ebay.co.uk/itm/550nm-Filter-for-IPL-Beauty-Treatm...


New to discussion board.
Has anyone any experience with the above filter?

m1tanker78 - 12-2-2015 at 18:15

Quote: Originally posted by John Green  
Quote: Originally posted by aga  
Quote: Originally posted by Marvin  
The Schott filter (OG550 was it?) is pretty inexpensive.

Thanks for the filter code.

A quick ebay popped up a surprise :

http://www.ebay.co.uk/itm/550nm-Filter-for-IPL-Beauty-Treatm...


New to discussion board.
Has anyone any experience with the above filter?


No first-hand experience but the specs indicate ~70nm transition from the stopband to ~50% transmittance. It would certainly be useful for attenuating the laser line (assuming excitation wavelength is 532nm?) but would also attenuate the Stokes-shifted peaks near the laser line. This type of filter can't be fine tuned, unfortunately. They're cheap enough to buy one and try it out.

m1tanker78 - 22-2-2015 at 16:41

I'm in the process of gathering bits and pieces to construct a Raman spec which will use 650nm as its excitation wavelength. After a lot of back and forth, I decided to base the concept around measuring the anti-Stokes scattered light. I don't fully understand how Stokes scattering can be more intense yet, lower energy than anti-Stokes?!? Leaving that out of the equation for the purpose of moving forward, most CCDs have better sensitivity at <650nm than they do in the NIR anyway. That was an important sticking point for me when choosing how to model the concept.

++++++++DETECTOR:++++++++
The detector I chose and purchased is a TCD1705D made by Toshiba. It's a 7450 element linear CCD.

++++++++ELECTRONICS:++++++++
I have a variety of Altera FPGA boards and ADCs for pulse generation, signal collection and processing, and communication to/from a PC or laptop. I'm going to try to model this design phase around a relatively inexpensive Cyclone IV development board as well as an 8 bit ADC to read the analog charge levels on each pixel and forward to the PGA. I've tentatively chosen the TLC594IP as an entry point in spite of being only 8 bits. It's integrated on the dev board and should be fine at least for testing purposes.

++++++++SOFTWARE:++++++++
Haven't decided yet. I plan to establish a USB link between the FPGA and the PC/laptop to gather the data and control the CCD integration time and possibly laser on/off and power level. I don't want to get bogged down with writing drivers and fighting Microsoft all along the way so I might target Linux instead. I'm comfortable with Python and C/C++ (like riding a bycicle, I hope). Most or all DSP will be performed within the FPGA so if worst comes to worst, good old RS232 will be plan B. Graphing is pretty straight forward under Linux. Additional DSP can easily be performed if needed.

++++++++OPTICS:++++++++
I purchased a few shortpass filters from Edmund in which the technical notes say that a 0* angle of incidence is the normal mode of operation. That turns out to be wrong; the filters have the usual 45* AOI as the norm. The reasons I bought several are:

---1. Can be used as laser line cleanup filters.
---2. Are 'OD >2.0' so I may need to chain more than one for max laser line rejection.
---3. Would rather use edge filters than notch (don't care about anything at or above laser line).
---4. Claim to have a sharp cutoff transition and good characteristics in and out of band.

If I can find a relatively inexpensive spectrometer I might forgo the front face reflectors and diffraction grating. In keeping with my design model and goals, I'll still experiment first with off the shelf optics as well as CD/DVD/BR for the grating. I have some ideas for an electronically-controlled slit but don't want to get sidetracked so will stick to good ol' razor blade slit and magnets.

++++++++GENERAL DESIGN AND CONSTRUCTION FLOW:++++++++
I'll begin work on the detector side of the design. Having a working CCD/ADC/FPGA/PC chain should help tremendously when it comes time to align optics, establish baselines and characterize filter response.

Next will be slit to detector chain. This phase of the design can be tested with ordinary light sources. No need to fire up the laser and mess around with samples until bugs are worked out in this phase. Standard light sources such as a neon lamp can be used to establish known spectral line positions, make adjustments, etc.

From there, the laser source to slit phase will be constructed. After characterizing the laser line(s), a cleanup filter will be added along with sample holder and edge filter(s). I may elect to add one or two reference sources as well but generally want to avoid using beam splitter(s) if possible. Since the design will be open box or breadboarded at this point, I will first use a low power laser in spite of wearing safety glasses rated for 650nm.

The rest is tweaking and building an enclosure. This is where I'm tentatively going to free myself to get sidetracked on adjustable mounts, adjustable slit, whatever. (I have a tendency to get sidetracked on my projects-- need to keep it in check!). :D

Suggestions, discussion and testimonies are welcome. For now I'll start work on coding the CCD driver.
=====================================

Technical specs and datasheets:
DETECTOR: TCD1705D
ADC: TLC549IP
EDGE FILTERS: EDMUND #47-290
FPGA: CYCLONE IV E SERIES

John Green - 23-2-2015 at 20:10

Tanker,
The reason Stokes is often used rather than anti-stokes radiation is that the stokes lines are generally more intense and the reason for that is:

The probability of an anti-Stokes transition is much less than that of a Stokes transition, since at room temperature the majority of molecules present are in the lowest vibrational level.

From the Wikipedia article Stokes Line.

The anti-stokes shift is of course derived from the vibrational energy (temperature) of the molecule.


In fact the ratio can be used to measure the temperature of the sample.

Marvin - 24-2-2015 at 14:07

TCD1705D is awesome for resolution but about 20 times less sensitive than the ILX511 or TCD1304AP. Mainly I think due to the pixels being tiny and square.

The short pass filters are not edge filters, looking at the OD plot they are very soft. This may be due to the 45degree angle design. The only sharp 45 degree filter I could find was an uber expensive holographic notch filter. They might work stacked but as John confirmed anti-stokes will cost you much more signal. Maybe a factor of 10.

m1tanker78 - 24-2-2015 at 23:21

Looks like I'm headed down the wrong path. Those factors are a recipe for failure. It's not too late to rethink the detector. I'm putting the finishing touches on the tcd1705d's firmware so I'm going to go ahead and get it working before I scout around for a better detector.

I understand that Stokes-shifted signals carry more intensity but the detectors' response drops off in that region of the spectrum. The detector's response peaks in the anti-Stokes region in contrast.

With full control of the detector's integration time, I'm hoping to be able to capture some useful spectra at good resolution even at the expense of time.

I've been promised a holographic notch filter but it's intended for 632-ish nm and I don't have a HeNe laser source at my disposal.

The dichroic filters are what they are. I'll just have to make do with them for now.

John Green - 25-2-2015 at 10:47

I am surprised that I hadn't seen this sooner as I had been expecting and looking for a development like this for some time. As long as these things stay in laboratories they remain ridiculously expensive. Until a product that appeals to Joe the beer drinker down at the tavern it won't be cheap. I suspect spectrometry is nearing that point. Technology is getting to the point that these things can be put on a chip and information technology is nearing the point that no brains are needed to interpret the result.
There have been on-chip implementations before but they have been low resolution color measurement devices. At 1nm resolution and low f# this one just may be different.
http://www.opticsinfobase.org/view_article.cfm?gotourl=http%...

[Edited on 26-2-2015 by John Green]

 Pages:  1    3