This is a testing site only. See the live Public Lab site here »

# Public Lab Research note

The user 'dhaffnersr' has been placed in moderation and will not be able to respond to comments.

# Plab v2.5 (upgrade 2) Comparison Spectra to SpexFluoroMax Spectrometer

by dhaffnersr | 22 Jun 14:12

Abstract

This was the original spectrometer from Plab's spectrometer kit that inspired me to modify its concept and build the DH 4.2 spectrometer, so I kept their original design intact and only put the parts in an enclosure, I thought perhaps this might make a good visual comparison between the original and my modified design.

Measurements between the slit to cmos camera eye are not as important as precise alignment of camera to DVD grating and DVD grating to slit entrance, so with this in mind, the camera is now at 39 deg relative to slit entrance and DVD grating is at same (39 deg) relative to cmos camera eye, for a UV-VIS spectral bandwidth of 343nm - 700nm (this is due to the camera's limitations.)

I have reduced the ruling density of the DVD grating down to 4.7g which equates to 1,351 lines per mm. The slit width is now 0.12mm. Project box dimensions are: 190mm X 109mm X 55mm

Slit width dimensions are: 25mm in length Cmos camera from slit: 88mm Camera angle: 39.5 degrees DVD grating angle: 39 degrees Entrance coupler for slit and cuvette holder: 44mm in Diameter

New Plab v2.5 Upgrade pics below

First plot showing Plab spectra as compared to SpexFluoroMax

Second plot with more detail

As you can see by the data, I am within 9nm [525nm] of the target of 516nm for the spectral data taken by the Oregon Medical Laser center for Fluorescein in Ethanol. The new enclosure has eliminated a lot of stray light that occurs with light entering the slit, according to the University of California Santa Barbara (UCSB) Science Line, an object's color represents the wavelengths of light that the object reflects. For example, white objects reflect all wavelengths of light, red objects reflect red wavelengths, and black objects absorb all wavelengths and reflect none.

Light energy can be converted into heat energy. Because white objects reflect all light, they do not have as much energy to convert to heat as black objects; therefore, light-colored objects are typically cooler than darker objects. To demonstrate heat absorption of various colors, you can place pieces of variously colored construction paper in the sun and record their surface temperatures every few minutes. The dark colors become hotter than the lighter colors. _ref _ http://www.reference.com/science/color-absorbs-heat-b73ca519b983f629?qo=contentSimilarQuestions

Preparation for the Fluorescein sample are straight foward, dissolve 1 gram fluorescein powder in 250ml distilled water and vortex in a magnetic stirrer for 2 hrs or until no particles remain and run solution through a filter to remove remaining micro particles, now you have a 0.1M solution of Fluorescein standard to use.

Fluorescein was tranfered to cuvette via serological pump pipette at 0.1 ml per drop for a total of 0.4ml of fluorescein sample in semi-micro cuvette.

ref

http://publiclab.org/notes/dhaffnersr/05-14-2016/dh-4-2-build-4-0-1-spectrometer-vs-spexfluoromax-spectrometer

http://publiclab.org/notes/dhaffnersr/04-20-2016/plab-spectrometer-version-2-5-vs-spex-fluoromax

Great to see this, Dave - I've been working hard on SpectralWorkbench.js this week. Can you share data file sources for the two spectra you've compared here? I'd love to see them in a SWB set -- thanks!

Is this a question? Click here to post it to the Questions page.

hey Jeff, @ warren, here are the links to the data: http://spectralworkbench.org/spectrums/82191 today june 22 fluorescein in ethanol

http://omlc.org/spectra/PhotochemCAD/html/037.html fluorescein in ethanol Oregon medical laser center

http://spectralworkbench.org/spectrums/80425 CSV beta upload-Oregon medical laser center-fluorescein in ethanol

@warren, hey jeff, just a quick note, I wanted to point out that I also used gorilla 2 sided tape to secure the camera and wood block down in the enclosure, the same principle I used for the DH 4.2 spectrometer and it works beautifully.

Cool, check it out: https://spectralworkbench.org/sets/3281

You have to equalize height to see them plotted against each other.

@warren, hey jeff, man that is totally cool!

Can the value of this dissimilar comparison be explained? Not only are the peaks different by 9nm ( a substantial discrepancy ) but the spectral curve characteristics have very little in common -- save a general "lump" in the spectrum near 520nm despite the claim of very similar sources being tested.

Is this a question? Click here to post it to the Questions page.

@stoft hey dave, the comparison is really not so dissimilar, if you mean the spectral lines do not exactly match then yes, you are right. First, spekwin32, the software I use stretches the spectrum when I am working with it, Dr. Menges has not expained why, but it snaps back when another spectrum is added, but this is inconsequential.

Two, The Oregon Medical Laser Center is using a Fluorometer, a much more sophisticated piece of equipment, Dr. Menges has this center as a reference because it is the standard in the industry, and I need those spectra's as a reference when measuring quantum yields, it is very difficult to find spectral data on the internet that one can down load for free.

Three, just because the spectral lines do not match exactly does not negate the validity of the data. I have had this data reviewed also on hackaday by others that have far more expierence than me, plus I am working with Dr.Menges on his project on hackaday as his bet tester, and he has reviewed many of my spectral work and has corrected errors on my part.

Four, I do not know, and it is not stated in the literature the chemical preparations done for the fluorescein sample tested. So I used an industry standard solution of fluorescein that I made ( I have posted several times how I did this and all the details) and diluted the sample to that of Lange's Handbook of Chemistry Section 4.14 thru 4.178 and my own research and analysis using different dilute sample.

All this beign said, I have posted research notes before on this very comparison with the spectral graphs separated, one showing the OMLC's fluorescein sample and mine, using this very spectrometer.

Five, My fluorescein may be slightly chemically different from their (OMLC's) fluorescein sample, that will make a slight difference.

Finally, I am using a DVD piece as my diffraction grating, it is 4.7g at 1,351 lines per mm, the slit is an acetate "film" which I tried to make as straight as possible but probably has it's own imperfections also, the cuvettes I use are superior, semi-micro UV plastic made from a polycyclic material and does have an absorption of less than 0.1. again though, it probably has it's own microscopic imperfections that may cause slight distortions, taking all these factors into account I think that my data is pretty remarkable in that I am using a home built kit, that's the whole point dave, it's a comparison on accuracy and precision, 9nm is really not THAT far off, especially since I am not even using any type of filter, ie,. a Notch filter or Bandpass filter for the laser.

I'd like just a little respect, that's all.

Thanks Dave H

I'm sorry you found asking questions disrespectful as asking critical questions (questions which are critical to the analysis process) are just part of the process and should be expected.

The questions about dissimilarity still stand because there are a number of significant unexplained differences between the two plots when one would expect very little difference in both shape and peak center since the substance being tested is the same chemical.

Comparisons between devices are useful for this vary reason; when one does not get the same result, it is a prompt to find out why (and possibly correct the issue) so as to make better measurements. Observing that the two curves are "similar in some ways" does not address the underlying questions as to why they are not the same (or at least really close). Not finding out why, limits future the "comparison(s) on accuracy and precision" you referenced.

@stoft, Hey dave, I'm not upset or anything like that about beign asked critical questions, I suppose I should have labeled this study as "Possible esoteric qualities inherent on theoretical comparisions."

I'm joking, I know the "why," I am trying to figure out the "way." I have two spectrometers now, 1 with a high speed cmos camera and 1 with of course the JDEPC-OV04. I will be doing this again using spekwin32's prototype hardware control program which works fine with the JDEPC-OV04, I can subtract the dark counts and control the camera's parameters (albeit, the one's that can be accessed) this one I can access the gamma correction and exposure time, which should be very helpful, maybe not so much the gamma correction as the exposure time.

I will compare the fluorescein samples again using the two different spectrometers, that should be interesting, as far as the respect, I meant I would just like to be acknowledged to a degree for progressing to this point, than when I first started doing this.

@stoft Hey dave, I just wanted to shoot ya a quick reference on a comparison I did do using data from the Oregon Medical Laser Center, the sample was Rhodamine B in Ethanol, I superimposed my spectral plot of rhodamine b ( I used the high speed cmos camera for this one,) on the OMLC's plot so spekwin32 program wouldn't stretch the data.

http://publiclab.org/notes/dhaffnersr/05-14-2016/dh-4-2-build-4-0-1-spectrometer-vs-spexfluoromax-spectrometer

Ok, additional measurements might help to identify why: 1) the peak center is different, 2) the slopes of the curve "sides" are much steeper (while the entire spectral BW of the curves is well within the webcam's BW, 3) there is a secondary notch and peak in yours which is not there in the reference and 4) the longer wavelength "tail" is essentially 'missing'. However, I suspect none of of these differences are related to minor factors like grating material, chemical concentration variance or even one webcam vs another. I'd suggest looking at more fundamental differences like the illuminant's wavelength and spectral bandwidth --- and I'm also guessing you have not employed any form of amplitude vs wavelength correction ......... reading up on the reference data's test methods might give some additional clues. It is generally best to assume that official reference data is exact and all user collected data is in error to some degree and so always requires a search for the contributing errors and then correction techniques before any form of validation can be attempted.

@stoft, Hey dave, it might make a difference that their excitation wavelength was 470nm and mine was 532nm. That's why i am doing this again but with spekwin's hardware control, I can adjust for intensity and wavelength, I couldn't use it before because it only showed the Y axis, until Dr. Menges updated it, now it should work fine.

here are the two sample spectra separated on different graphs so the plab sample isn't crunched up on the same plot:

Also I cross sectioned a bigger part of the red shift of the fluorescent spectrum when I captured the sample, so that "tail" and it's "lump" could very well be that red shifted wavelength.

Well, the 470nm vs 532nm is likely the most important -- the peak in your plot is probably dominated by the source while they reference at 470nm is at a wavelength which is shorter than the plotted data (and probably filtered out by their system). Given the huge "in-band" 532nm source, plus the added peak at ~570nm, using a 532nm laser to make such fluorescence measurements become much more complicated at best. Unfortunately, without more sophisticated processing, amplitude correction and proof such processing is valid, 532nm plots should be considered suspect; their data is much too obscured to use for comparison.

@stoft Hey dave, yes you are absolutely correct, that is why I need to purchase either a notch filter or longpass filter for the 532nm laser, both filters are very expensive and there are a few more steps I would have to incorporate in order to get the spectrometer to operate as a raman spectrometer.

I do have a high powered 470nm Blue LED that I can use, its just that the FWHM is so wide I'm not sure the data will be any better?

Is this a question? Click here to post it to the Questions page.

Well..... not really ..... The Raman technique requires 1) a very stable (very narrow BW) laser source and 2) a 'matched' very narrow notch filter (or pair of precision LP and HP filters) because the Raman signal is contained within a few nm of a ~470nm excitation signal's center wavelength. Raman is NOT a technique of just filtering out the excitation source's BW to look at the rest of a broad spectrum. Raman configurations require very good optics, lasers, filters, etc and are not cheap or easy to construct.

In the case of the OTK, the 405nm laser (vs UV LEDs) the laser is sufficiently narrow and is at the end of the webcam detection range such that observing the fluorescence signature past about 420nm is actually possible. Using a UV LED is problematic because its much much wider BW means the excitation signal actually becomes "in-band" for the shorter wavelengths; masking a significant part of the signal being detected. Both a 470nm LED source and a 532nm source are definitely "in-band" making them effectively useless for full-band measurements.

The argument for just "subtracting" an "in-band" source (measure just the source and performing a mathematical subtraction) isn't really valid because the degree of error 1) is dependent on the ratio of source to measured signal and 2) can be extremely large in the case of smaller signal levels (the nature of which you have no way of knowing in advance ). Even the Raman technique does not take data from the band of wavelengths comprising the excitation signal.

So, my suggestion is that ALL data having been taken with either 400-470nm type LEDs or with 532nm or similar sources be excluded as essentially invalid because the signal being measured is heavily contaminated by the "in-band" excitation source energy. Claiming anything vague like "the effect of the source isn't that bad" is neither a valid scientific disclaimer nor a valid basis for accepting the measurement data. This is why the 405nm laser is the ONLY available option (for PLab devices) discovered so far. While it might be convenient if cheap LEDs provided an easy solution, they don't. These types of measurements are not trivial and overlooking the details can easily invalidate every measurement.

@stoft Hey dave, ok I did both spectral captures again with a 470nm Blue LED using a quartz cuvette for both, one using Plab's spectral workbench data capture program and Spekwin32's capture program, on the spekwin data I was able to adjust the gamma correction and lowered the default by half to increase the luminance and I was able to subtract the dark counts also, it would be nice if I could do that with Plab's program.

You are right, the shorter wavelength made a big difference, I should have known better.

Here is the plot showing the Plab data, with a FWHM of 27

Now here is the one using Spekwin32, with a FWHM of 21

Certainly the spectral lines are very different but the wavelength is correct for Fluorescein at an excitation wavelength of 470nm. Do you think it makes any difference that the LED has a 30 degree focal lens attached? I had to put it attach it because the viewing angle for the LED is 140 degrees.

Is this a question? Click here to post it to the Questions page.

First, ALL displayed plots used for comparison MUST have the same displayed 'X' axes wavelength range .... anything else is just too visually deceptive. "Shape" means nothing when the X-spans are different by a large percentage.

Next, your plots still show a very narrow BW measured response (relative to the given reference) and so discovering that root cause should be first priority. Anything else is just distraction at this stage. FWHM is mostly useful for narrow-band signal; not broad-band responses because narrow-band signals are generally quite symmetrical within the region of interest whereas wide-band signals often hold no such characteristics so FWHM becomes meaningless.

your quote: "So, my suggestion is that ALL data having been taken with either 400-470nm type LEDs or with 532nm or similar sources be excluded as essentially invalid because the signal being measured is heavily contaminated by the "in-band" excitation source energy."

Hey dave, I just wanted to make sure I was clear reading this point your making here, so are you saying that ALL LED data that is used in UV-VIS spectroscopy is invalid?

And if so, can you provide a little more of your data to support that claim other than your word?

I know you cannot use LEDs in raman spectroscopy because the energy of these vibrations is often tiny. So incoming and outgoing light has almost the same energy and you need instruments with a high energy resolution to separate the two light beams. I know The light of an LED is very broad in energy, It is so broad that in my opinion you wouldn't be able to see the tiny energy difference from the vibrations, and I know that's why usually continuous lasers (and their extremely narrow energy distribution) are used for Raman spectroscopy.

I figured these concepts out a while ago, but they still have validity in some applications, and they don't always have to be diode based technology, in the commercial arena, certain types of white light LEDs are begin used in the bio sciences because of their very low cost and reliability and longer life spans.

A good example is in molecular biology, DNA quantification at micro volumes and prep HPLC, ( Prep HPLC is commonly used for purifying proteins, isolating natural extracts and other routine measurements, and relies on absorption or fluorescence spectroscopy in the UV wavelength range for controlling fraction collection.) This can only be done in the ranges of UVC 200 to 280nm, UVB: 280 to 315nm.

I'm just saying that to make a blanket statement like ALL data is to be "thrown out" because you think it should be invalid because an LEDs bandwidth is much too wide, is a little much, at least there are individuals here at plab that are trying, sure they may make mistakes along the way but like everything else in life we learn from them.

Also I have never seen anyone say this: "the effect of the source isn't that bad" but you seemed to have implied it somehow?

And if you are going to make a statement like this: "This is why the 405nm laser is the ONLY available option (for PLab devices) discovered so far."

Where then, is YOUR data to support such an absolute? ( I say this because you capitalized ONLY.)

Look, I have read all your postings here at plab, you are one smart cookie, no doubt, but your interpersonal approach leaves much to be desired.

Is this a question? Click here to post it to the Questions page.

Last; first. I do take issue with claims of a negative "impersonal approach" when it intonates that I have not included an emotional response within a scientific discussion. While individuals do become emotional about science, science is the pursuit of truth and is not innately an emotional argument. One of the elements I try to contribute to PLab is clarity and critical thinking because scientific progress is not achieved by emotional reactions or arguments. I often dig through many layers of hypothesis and measurement anomalies before finding supportable data to publish; it is simply the process of the scientific method.

I did contribute a research note comparing LEDs and Lasers as excitation sources for use in broad-band fluorescence -- meaning the acceptance of spectral data over the BW of the PLab spectrometer. [ See: https://publiclab.org/notes/stoft/02-23-2016/405nm-led-vs-laser ] So, I am not just talking out the top of my hat.

My generalized paraphrasing of imprecise references about measurement data is just that. Authors of notes do sometimes make vague disclaimers (for instance, your word choice was: "...9nm is really not THAT far off...") without logical (clear reasoning described) support for those conclusions. In the world of measured spectra, 9nm could either be of little significance all the way to extremely significant -- it depends on what is being measured. For peak fluorescence of a test to duplicate a known reference, 9nm is significant because one would expect the peak to match the reference quite well. When it didn't, that should automatically trigger further investigation; not a dismissal of the offset. If it's a laser, 9nm is even more important; so my point still stands. When making a claim, it is necessary to prove (show basis) why it is justified -- otherwise, it should not be a claim, it should only be a hypothesis (conjecture) with more questions.

My statements about the validity of using "in-band" excitation sources for broad-band spectral data collection (the majority of the fluorescence work and discussion within PLab -- which, obviously, describes the context of PLab notes) is not dependent on specific chemistry; it simply concerns the measurement concepts. What I stated was that it is not possible to make a valid (accurate) broad-spectrum measurement of an unknown fluorescence spectral signature when using an "in-band" excitation source because the source signal obscures the data being measured. And, I pointed out that just performing a mathematical subtraction of the source spectra was invalid (inaccurate in the general case) because the resultant errors can amplify dramatically (worse as signal level drops) and raise the measurement uncertainty (inaccuracy) beyond the measurement value itself. These effects only get worse as the BW of the source becomes wider (i.e. LEDs).

This is why using a 405nm laser, which has an extremely narrow BW (<1nm) and is close to being "out-of-band" (for a detection bandwidth of 400-650nm for webcams) allows 420-650nm broadband fluorescence measurements while a 470nm LED, with an ~30nm BW, and a 532nm source do not -- because they are "in-band". Just the fact that a fluorescence signature can be measured is insufficient. Therefore, I still draw the conclusion (from the numerical values of the measurement technique) that yes, the 405nm laser is (presently) the only viable option for PLab devices to measure broadband fluorescence. [...and this even ignores the evidence that the fluorescence efficiency (source power vs fluorescence signal level) is much much lower with LEDs than lasers which also has other practical implications. ] One could argue to simple ignore all spectral data below the upper end of the excitation source's BW. However, in many cases (other than with the 405nm laser) this would cut-off a significant range of wavelength information which is critical to the measurement -- effectively making the measurement data invalid.

So, relative to the spectral fluorescence data presented and the types of fluorescence applications as described within PLab, I still believe the present broad-band measurement techniques are valid only with out-of-band excitation sources and that the 405nm laser is (presently) the only viable option.

It's in the details of the numbers ......

@stoft hey Dave, well thank you for replying, I appreciate that. I said before, I have read ALL your work here, and I have learned from it also, I do agree with a lot of your points you have raised, I know the terminology and the linguistics of research and science, I am a enormous proponent of the critical thinking process, I wish more people would utilize it.

"Wishing" is not a very good plan though, and neither I suppose is making vague statements when it comes to accuracy and precision, in that, I acquiesce. I should be more precise and clearer in my statements, like when I said that "9nm wasn't so bad." I need to put statements like that in context. Math, you are certainly right, it's always in math, key to the universe, it doesn't satisfy curiosity though, it's a means to a logical end, the fun part is the journey to get there.

Figure 1: Deuterium lamps offer significant light output across the UVC wavelengths while xenon flash lamps offer light at various wavelengths across a broad spectrum. LEDs are tunable and offer high light output at a specific wavelength required for measurement.

UV LED Vs. Mercury Spectral Distribution

I whole hardly agree with you that the LEDs that I am using have a wide bandwith and thus pollute the data, I know, but the point is in the general research of the possibility of using a more cost effective and reliable light source, that can be tuned to the wavelength of interest. its the discovery that makes the scientific process exciting, it's not only the mathematics and the cold calculations that define me ( I'm just going to speak for myself,) it's in the "what if" its possible? What if using LEDs for certain uv-vis spectroscopy applications can work? I think its worth exploring the possibility.

Jargon, does NOT a scientist make! Passion for one's work does, failure and success, synonymous with each other.

So I really don't think I should just discount all the work I've done here at Plab, others may believe so, I do not, and I'm not going to just wait until I reach that "eureka" moment that may have taken me 365 days to attain, before posting my findings, its kind of like building a data base of knowledge and progression, just because everyone expects things to remain within protocol doesn't mean one cannot deviate once in a while. The data is still valid if the numbers line up? right? The method can be altered or adapted in other ways, no one is breaking the laws of physics or any universal principles.

To bring this back to your original issue about me comparing my spectra against that of spectra gathered by a $25,000.00 piece of equipment, I lose, it IS in the getting closer that matters. I suppose then that, Thomas Edison should have just thrown in the towel after test # 2000 no? Of course not, I have a bigger picture in mind here, a spectrometer with a built in LED source that can be tuned to the appropriate wavelength that an individual wants, but I have to work within the parameters and monetary budget that I have and equipment that I can construct. That's what makes it fun, and if it becomes work, then I can't love it and I will soon lose interest, to me there is never a point in doing that which you hate. Is this a question? Click here to post it to the Questions page. Reply to this comment... I am the kits fulfillment coordinator at PL, I don't have very much (any, really) scientific background but I do have a background in economics/business. My opinion should be taken lightly, but @dhaffnersr if youre hoping to find a low cost method but still want it to be accurate then I think you can choose either an LED or a laser, 405 nm lasers seem to be selling (and I don't know if they're the right ones) at between$10 and $50 (I'm sure they go up from there but those are the common market prices); and while that would make the user purchase price of the desktop spectrometer double, it would still only double to$100. For accurate, reproducible data I'm very, very, very confident that the PL spectrometer would be as popular or more popular than it currently is. I base this confidence off of the average ticket price of the other kits at the PL store, the volume of DSK 3.0's sold, and the expectations of our customers as derived from our brand image.

So feel free to use either source, whatever works. Don't let the cost be a determining factor in the search for good data if the laser might make your quest easier.

@abdul I do have the 405nm UV laser pointer, but I find it's usefulness limited for certain applications. I originally I bought the Plab 2.5 spectrometer from Edmunds scientifics magazine, and I have modified it since then, the problem is in the kits portability. The 3.0 still has to pass some kind of spectral "muster," no one that I know of, has done any kind of analysis with it yet and posted any results here at Plab.

I think a kit with a built in universal light source would be an extremely marketable item.

Light sources are both a challenge and the key for many aspects of spectrometer applications. I'd suggest that a "universal light source" isn't an applicable concept. Instead, I'd suggest narrow-band or wide-band for most applications suggested for the PLab devices.

Specific excitation for fluorescence: LEDs are interesting, cheap, have a variety of wavelengths but are less stable than lasers (close-in phase noise), are generally not coherent and are NOT wavelength tunable. Lasers are much more stable (wavelength), coherent and provide much higher spot-intensities. There are some special tunable lasers (a stable laser passes through a crystal which tunes the shifted-wavelength output as I recall -- very expensive stuff). In general, these need to be "out-of-band" for fluorescence.

Broadband illumination: These would be sources which are intended to produce uniform intensity over a broad wavelength range (i,.e. the range of the spectrometer. They would provide illumination for both intensity calibration and, after, for absorption and reflection measurements. The best I've found has been the Solux 4700K halogen.

A third, specific class, of illumination could be for single-wavelength absorption or reflection measurements. I don't know what applications fit this category for the PLab device, but LEDs might fill this specific niche if they produced sufficient intensity for that application.

Right. I've published many notes on measurement parameters of the PLab-3 class of spectrometer but PLab has not yet attempted to establish a well-defined and well-characterized 3.0+ device.

Right, experimenting is valuable; the key is to be rigorous about what's real and what is just an anomaly, what can be justified by data and keeping hypothesis separate from fact unless the link between the two can be fully explained and supported. Experimenting with a cheap alternative is fine, but a measurement technique must be proven viable before cost is a factor; i.e. something could be very cheap, but if it doesn't meet the measurement requirements, cheap doesn't matter. If something works, but costs a bit, at least it is a viable method -- and a cheaper alternative might be found -- which must then be compared against the more expensive method which was proven accurate. You could get lucky and find cheap AND accurate at the same time -- one single step -- but accurate and meeting the measurement goals cannot play second fiddle to just being cheap.

@stoft hey Dave, ok here is a snipet of my concept for a tunable LED, it is based on a bi-color led, the red and green type, instead it is sandwiched with all 3 primary colors ( red,blue and green) it can then be tuned within a given range by selecting their operating drive current in order to obtain a precise wavelength. The LED would have to operate at the 1W to 5W range for proper luminance.

These LEDs can still be driven by any PC's USB port which would give this spectrometer superior reliability and maximum portability and still remain at a very reasonable cost.

Well that's my concept, now my primary goal is to get both spectrometers I have built to operate correctly, I work off the OMLC reference because it's the only reliable one I can find, but it is an old reference from the 90's, which it shouldn't really matter, here at Plab other than you, no one that I've seen puts in the work in this area. I'd love to compare against other spectra's but there are none, now as far as the 3.0 is concerned, it's concept was already instituted back in 2009

http://www.asdlib.org/onlineArticles/elabware/Scheeline_Kelly_Spectrophotometer/index.html

at the Department of Chemistry University of Illinois, to me it's potential is really begin wasted, I haven't seen it utilized in any meaningful way. Unless it's nothing more than just a selling point in a marketing campaign.

Anyway, I will continue to experiment with both of my spectrometers until I get one of them to match the desired spectrum, then I can move forward to the next logical step.

Nope .... each LED has it's own "narrow" (~30nm) BW and so the red, grn and blue LED spectra are all independent - they don't mix and they don't influence each other and they don't combine to form any single different spectrum. Also, because they are not spatially located in the same "spot" and because they will each have a different efficiency and different BW, they are not easily "interchangeable".

LED vs laser power density is also a problem. LEDs are roughly 1% efficency devices with typically a 30-deg beam spread which means a 5V, 500mA max USB port can supply 2.5W (at max overload -- generally one should never expect to pull more than about 250mA from a USB port and some ports on some devices may not deliver that -- it is bad to risk user's hardware). So, this leaves only ~25mW at the diode die surface (of say 1mm area) which is then spread to something like 1cm-sq which is again just 1% of the LEDs output energy density; or just 0.25mW of optical power in the cuvette (in a 1mm "beam diameter". A 405nm 5mW laser, with ~1mm beam diameter is therefore inherently about 40x the energy density of the high-power LED.

While these numbers are all rough approximations, the order of magnitude of the effect easily explains why everyone has had such a difficult time getting enough intensity from LEDs as compared with the laser. My experience with testing @warren 's blue LED was no different -- detectable, but very poor SNR.

The reason energy density maters, is because the spectrometer is designed to observe only a very small area of light (remember, it is a pseudo-collimated design based on a narrow slit). Therefore, the spectrometer can only "see" a very tiny window of light (i.e. fluorescence). So, even though the entire cuvette is "glowing", only the photons from a small area of the fluid, are being emitted directly inline with the spectrometer's optical path and will be detected by the camera sensor. Therefore, what is required is high energy density and a very good, stable spectrometer alignment on that small region where the laser is exciting the test fluid with maximum energy density.

Attempting to construct an accurate comparative measurement of a broad BW fluorescence reference curve requires duplicating at least the fundamental elements: 1) Excitation source (405nm laser might work -- at least it is narrow-band and "out-of-band" relative to the plots you presented), 2) test fluid (seems that might be the same chemical and concentration they used -- it needs to be the same), and 3) a gain-corrected spectrometer. Assuming the first two were resolved, gain correction (so the PLab device has a flat response) to eliminate that factor from altering the general "shape" of the fluorescence curve. All the other factors are probably of lesser importance at this stage.

@stoft hello dave, I don't know why you think it is not possible, you have such a defeatist attitude. Of course its possible, I'm not going to do all the bloody work for you, check these sites out:

http://www.photonics.com/Article.aspx?AID=57253 LED arrays http://www.mightexsystems.com/family_info.php?cPath=&categories_id=112 Universal light source

I also know about the LEDs that Warren wants to use and test, I have them, here is a pic of one I soldered and yes, it works, but it's luminance just isn't going to cut it.

This little thing was very difficult to work with, it really needs to be baked on a PCB with a solder reflow technique.

I guess I'm just more of an optimist when it comes to solving a problem that others seem to think is impossible.

Is this a question? Click here to post it to the Questions page.

The proposal (above) was for " ...concept for a tunable LED, it is based on a bi-color led, the red and green type, instead it is sandwiched with all 3 primary colors ( red,blue and green) it can then be tuned within a given range by selecting their operating drive current in order to obtain a precise wavelength...." Technically, this is simply incorrect for the reasons I described.

The first link just posted, is for quite a different device with thousands of custom-tuned, individually controlled LED chips on a single substrate to achieve overlapping individual spectra. The second link is for a range of LED products with a range of wavelengths and none of those (even the white light versions) appear to have a broad-band emission curve. This is why I still suggest that a Solux 4700K halogen lamp is one of the best options for a practical broadband source for use with PLab devices. I have never said that high-power thousand-LED array devices will not eventually be available and at an affordable price; but that might take a while ......

It is unfortunate that what I write is not always more carefully read and I do believe that my contributed notes stand as evidence that I have been pushing the envelope on many design aspects of the PLab spectrometer.

It is also unfortunate that simply observing the technical challenges from a more abstract perspective (which makes issues, limitations and challenges more obvious, and sometimes shows them to be impractical or in error) is labeled as a "defeatist attitude".

I've issued no personal attacks and I have only commented on the technical issues related to the topic and proposed ideas where I have knowledge to do so in a way which can communicate the reasons behind the comments. Personal attacks are inappropriate here and I have limited time resources.

@stoft I apologize if you thought I was using ad hominem attacks, it was not meant to be so. I can read very well dave, the same might be said about my postings also, research or other wise, case in point, my quote "As you can see by the data, I am within 9nm [525nm] of the target of 516nm for the spectral data taken by the Oregon Medical Laser center for Fluorescein in Ethanol."

You quoted me as saying "9nm is really not THAT far off...")

Show me were I ever said that? Things like this are certainly an aspect of "critical" thinking right? I would hope that my words are begin read correctly also, in this way there will be no misunderstandings. I appreciate the fact that your resources are limited and that you do take the time to respond, but I do have the right to address the issues about how your approach to my research is sometimes unwarranted in it's tone.

I call it the XYZ principle: x= tell the person what the problem is y= explain how you feel about it z= tell the person what you expect

This is the most up front and honest way to eliminate most misunderstandings.

Is this a question? Click here to post it to the Questions page.