| Author | Message | ||
     Gustavo Figueira de Paula (gustavo) Intermediate Member Username: gustavo Post Number: 20 Registered: 6-2008 |
Franz and Zvi, In NIR spectroscopy, 8 nm is enough resolution to perform good calibrations in many situations. But sometimes we don't need a calibration, but peak resolution - sometimes NIR is used to chemical investigation much like MIR. I like NIR because SNR is much higher, sample preparation is much easer and data acquisition is faster, when comparing with MIR. But there's a trade-off: band overlapping, a issue even in MIR, is much worse in NIR. Better resolution means better peak resolution. So, sometimes, much better resolution is needed. If it is the case, 1, 2 nm resolution in a DA spectrometer could be achieved without pain. About diode area: when I'm talking about noise, it means thermal noise. Thermal noise in InGaAs photodetectors is a big issue, so we pay thousands of dollars for very expensive thermoelectric coolers without dropping tears. Thermal noise (larger portion of the dark noise) is a function of several parameters, but increases proportionally to detector area. More interface, more noise. Bayspec uses a diode arrays with 256 pixel, each pixel has 50 x 500 microns. StellarNet's detector has 512 pixels with 25 x 250 microns each. Easy calculations lead us to a factor of four in area; it means that Bayspec detectors can harvest four times more photons (if all other parametes are the same) than StellarNet's detector, but also, the thermal noise in dark condition is also four times higher. Since detectivity of InGaAs diodes is already pretty high, noise becomes much more important than sensibility - less sensibility with high detectivity is usually enough. Of course, 256 pixel detectors are much less expensive, and if 8-16 nm resolution is OK, we can buy a better TEC and have an overall less expensive system with enough performance. But the right balance of noise, resolution, price and cooling is much particular of each application, so only you can determine, after trying, what combination will be the best. Gustavo de Paula. | ||
| Zvi Barnea (barnea) Member Username: barnea Post Number: 14 Registered: 11-2010 |
Thank you very much Gustavo. I also did not completely follow what is your conclusion regarding 256 vs. 512 pixels size in BaySpec & StellarNet? In the beginning you say that StellarNet has 512 pixels while BaySpec has only 256. However in the following sentence you say that BaySpec has larger pixel size?(!) Might be that pixels size is not the pixels number? I need 30 � 35 seconds for each sample, will FT-NIR suit? What about the noise above 2000 nm? Where water stops absorbing? 2000 or 2200 nm? Will StellarNet build upon request a 1850 machine with 512 pixels? Any information about BWETEK as regard to 512 pixels? Zvi | ||
| Francesco Davini (franz) Member Username: franz Post Number: 14 Registered: 2-2009 |
Gustavo, I can hardly understand what you meant in: "BaySpec NIRS-02-1100-2200 has a 256 pixel detector, giving you (roughly, the grating differs) half resolution compared to StellarNet RedWave-NIRX-SR-512 (512 pixel). Also, you must note that pixel size is four times larger for BaySpec's detector - it improves sensitivity, but also improves reasonably noise. This is why StellarNet don't build spectrometers with 256 pixel detectors, as said by the chief engineer to me not much time ago. " I don not know the mentioned apparatuses at all, but as far as I know a larger pixel size should give lower noise, not higher. But I am a chemist, not a photoelectronicist... Also, another question arises from yours and other posts about Diode Arrays. Is all that resolution worth anything in normal applications? I think that most commercial NIR calibrations made with FT or moving grating NIR instruments use bandwidths in the range of 4-16 nanometer. Which the practical advantage of higher resolution vs lower light energy hitting the sensors? Thank you all for your explanations: I am here mostly to gain knowledge. | ||
| Gustavo Figueira de Paula (gustavo) Intermediate Member Username: gustavo Post Number: 19 Registered: 6-2008 |
Hi Zvi, BaySpec NIRS-02-1100-2200 has a 256 pixel detector, giving you (roughly, the grating differs) half resolution compared to StellarNet RedWave-NIRX-SR-512 (512 pixel). Also, you must note that pixel size is four times larger for BaySpec's detector - it improves sensitivity, but also improves reasonably noise. This is why StellarNet don't build spectrometers with 256 pixel detectors, as said by the chief engineer to me not much time ago. About water: the 1940 nm band is perhaps the most important, and should be taken into account when working above 1700 nm. When I bought a D.A. spectrometer I choose a 2200 model because a I need to measure water. If you want to measure amines or amides, probably you will consider a 2500 nm detector. The band you want to detect is perhaps the most important criteria to select a detector. About Arcoptix: the Arcospectro-NIR is a FT-NIR instrument. If it's a better choive over D.A. spectrometers is a question of weighting your demands. If you need much fast data acquisition (miliseconds), D.A. is THE choice. But if time in not a problem (seconds to acquire a spectra is OK), consider FT. There are discussion threads exactly about this topic, you should read a bit more to be able to compare these techniques. Gustavo de Paula. | ||
| Zvi Barnea (barnea) Member Username: barnea Post Number: 13 Registered: 11-2010 |
Thank you all again and I guess I will say it every time I communicate. Howard, That is very interesting what you point, especially than when you look on the characteristics of detectors with upper limit of 1900 nm, you see that they do not need cooling system and as mentioned their price is only a bit more than those of 1700 nm. So I am certainly going to pursue this issue. As regard to your notes: 1. I thought water starts to interfere with 1850 up to ~ 2200. Above is as I understand a big problem. 2. You 2nd note- " StellarNet's spectrometers are the cheapest I�ve found for 1900-2200 range with 512px detectors" I am not sure about it, I did not tried to figure out but this is what Karl found: Stellar Net Inc. offers the DWARF STAR spectrometer for the 900-1700 nm region at $13,125.00, and the RED-WAVE-NIRX-SR for the 900-2300 nm region for $23,995.00" I did not finish my search I found BWTEK, which should tell me about their product, BaySpec is also in the frame and this they wrote: "The price of the NIRS-0900-1700-SMA is $8,970. Lead-time 3-4 weeks. We do offer options for extended range out to 2200nm and 2500nm, but there are some trade offs on cooling that you may need to consider. The 2200nm cut off detectors are 10x more noisy than the 1700nm type, and the 2500nm are 100x the 1700nm. It is often recommended to incorporate a deep-cooled detector option (-55dC). 1. NIRS-02-0900-1700, $8,970 (<4 nm) 2. NIRS-02-1100-2200, $15,970 (4-5 nm) 3. NIRS-DC-02-1100-2200, $24,970 (4-5nm) 4. NIRS-DC-02-1250-2500, $29,970 (4-5nm) We regularly work on custom options, so if you would like to discuss further your application and requirements we can suggest the best price/performance trade offs" Recently, I found an interesting Swiss company that offers FT-NIR. Below the link you can find their presentation and further below what they wrote in a letter (slight differences which I think can be settled). The main question remains, which is best for the application and which is the most reliable tool http://www.arcoptix.com/arcspectro-nir.htm PRICING STARTING AT 11'990 EUR!!! (ANIR 0.9-2.6 m with software) ● High stability in wavelength position and intensity (no baseline drift) ● High wavelength precision (ideal for peak detection) ● No baseline drift (due to device principle) ● High resolution (1-9 nm) ● USB powered ● Fibered system (NIR multimode fibres) ● Large wavelength range (only limited by the detector) ● No second order spectral contributions (FT spectrometer) ● Cost effective (single NIR photo-diode) Their Letter "The new NIR spectrometers will have enhanced sensitivity because they will use completely new interferometer, with enhanced light throughput. This will particularly be beneficial for diffuse reflection applications like yours. The price for the NIR systems with uncooled detectors will be around EUR 13�000 (this slight increase in price is partly due to the new electronics with embedded spectral processing), while cooled models will be around EUR 15�000. The new NIR spectrometer will be released in approximately 5-6 weeks from now". Any experience ? thank you, Zvi | ||
| Gustavo Figueira de Paula (gustavo) Intermediate Member Username: gustavo Post Number: 18 Registered: 6-2008 |
Zvi, StellarNet's spectrometers are the cheapest I�ve found for 1900-2200 range with 512px detectors. They work well, but the noise level is somewhat high, even with double stage TEC - but it's a inherent problem from extended-range InGaAs detectors, not from StellarNet's design. --- A second comment is about why the 900-1900 range is missing. I believe that the water band at 1940 is the main reason. If you want to see water, you need a a 2200 detector; if water doesn't matter, 1700 is enough. Why a 1900 detector, with higher noise and lower resolution, will be an advantage? In fact, theoretically speaking, is possible to design detectors with higher cut-off wavelength anywhere between 1700 and 2500, but must be a very good (commercial) reason to choose these or that range. From Wikipedia (InGaAs): "The optical and mechanical properties of InGaAs can be varied by changing the ratio of In and Ga, InxGa1-xAs.[1] The InGaAs device is normally grown on an indium phosphide (InP) substrate. In order to match the lattice constant of InP and avoid mechanical strain, In0.53Ga0.47As, this composition has a cut-off wavelength of 1.68 μm." "By increasing the ratio of In further compared to Ga it is possible to extend the cut-off wavelength up to about 2.6 �m. In that case special measures have to be taken to avoid mechanical strain from differences in lattice constants." --- Venkynir, in DA-spectrometers, what I observed both in silicon and IngaAs photodiode arrays is a change in "noise structure" when timing is changed. It's an electronic noise which changes the relative intensity at each wavelength due to variations in several electrical parameters affected by the integration time. What happens is an overall change in noise intensity, but also a change in the noise profile over wavelength. So, if you change the integration time, you *MUST* take a new dark/white reference. Note: increasing integration time increases noise intensity, but the signal increases hundreds of times more, so you keep the signal-to-noise ratio in acceptable levels. Gustavo de Paula. | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 404 Registered: 9-2001 |
Zvi - of course it's always hard to read the minds of someone, and try to figure out why they did what they did. But I can make a guess, in this case: an instrument that goes to 1900 nm for the same price would compete with the one with 1700 nm capability that they already have and are selling, while requiring extra engineering effort to bring it to market. Once a company has decided to provide a longer-wavelength capability, they might as well go "all the way" and provide the 2200 nm capability, since that gives them a justification to increase the price more than they could for a 1900 nm instrument. And trust me, this would also be more than the price increment for the 2200 nm detector! Of course, I may be wrong about the details, but the important point to keep in mind is that decisions to develop a new instrument are not based solely on technical merit, or even necessarily price alone. The decisions also include considerations of development cost, size of the market expected for different capabilities and the price they expect to be able to charge for those extra capabilities, as well. I'm sure there's more, too, but this should give you an idea. \o/ /_\ | ||
| Zvi Barnea (barnea) Member Username: barnea Post Number: 12 Registered: 11-2010 |
Thank you all, I found several machines with 900 to 1700 nm and some that go to 2200 or 2500 nm. The noise is increased with 2200 nm and the resolution is decreased. Cooling system improves it and increases the price. However, I did not find (in an ordinary list) an upper limit of 1850 or 1900 nm. This puzzles me since you can find InGaAs detectors in this upper limit with almost similar price as the 1700 nm detectors price, which are much cheaper than InGaAs detectors with 2200 nm upper limit. I wonder why such jump from 1700 to 2200 nm naturally developed skipping the mid level of ~1900 nm, despite the commercial availability of the corresponding detectors. Zvi | ||
| Art Springsteen (artspring) Senior Member Username: artspring Post Number: 35 Registered: 2-2003 |
Hello again, Karl is right on Stellarnet but there are also less expensive alternatives from BW Tek, Ocean Optics, Zeiss (Tek5) and Avantes, just to name a few. Whether they are as 'good;'as Stellarnet is up to interpretation but they do exist, in the $4-8K range. Of course, what no one has mentioned(the elephant in the room) is the cost of all the peripherals- light sources and stabilized power supplies, sample handling, standards to make sure you are linear and your wavelength scale is correct, software that matches your application. In the end, the spectrometer may be one of the lower cost components. <a > | ||
| Karl Norris (knnirs) Senior Member Username: knnirs Post Number: 48 Registered: 8-2009 |
Zvi, Stellar Net Inc. offers the DWARF STAR spectrometer for the 900-1700 nm region at $13,125.00, and the RED-WAVE-NIRX-SR for the 900-2300 nm region for $23,995.00 I doubt that you can find or build the required spectrometer at a lower cost without resorting to discrete filters as suggested by Art Springsteen and Howard Mark. Karl | ||
| Zvi Barnea (barnea) Member Username: barnea Post Number: 11 Registered: 11-2010 |
Howard is right. I started tis discussion and eventually remained silent as I had to enter some waiting period. The spectra I need are 800 � 2000 nm and the application is for biological fluids and tissues. My target price is < 2200$ and if necessary < 2500$. The issue is that with my device there would be couple of small add on accessories (estimated to be < 300$). What I learned from my InGaAs wondering is that you can get a manageable price, once you can deal with upper limit of 1700 nm. Above this, you may need a cooling system which increases the price substantially. We have very good feasibility results with an expensive device and now we wish to repeat it as well as validate it with a cheaper device (~ 12000$) and wavelength upper limit of 1850 nm (not 1700 nm) that can interface with a friendly chemometrics program (that by itself is here about 4000$). We are in a waiting period as bureaucratic funding issues take longer than I thought (not always but definitely in this matter). Any advice regarding the all related issues is welcomed. Zvi | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 403 Registered: 9-2001 |
We seem to have two separate discussions going on in this thread. Since Zvi began it, and most of the postings relate to Zvi's questions, I have to recommend to Venky that, if he wants to continue his discussion, he start a new thread so that things don't get too confused. That said: Karl, I agree with you, but only to an extent. "Low cost" is relative; if competing instruments sell for, say, $100,000 then a price in the range $1,000 to $10,000 would be "low cost". It's also important to know what the market is, and how large the market is. Consider, say, hi-fi's. I've long felt that there's little reason why an NIR spectrometer has to cost much more than, say, a good hi-fi system. The reason they don't is because instruments are manufactured basically one at a time "by hand", while hi-fis are manufactured by the tens of thousands using automated equipment. Similarly with detectors: when you buy then singly or in (small) lots of 100 or so, then they're expensive, but if you can come to a detector manufacturer with a (legitimate) order for 100,000 - 1,000,000 units, I think you'd be surprised at how low the price can become. Even Bob Rosenthal has managed to get the manufacturing cost of his body-fat meter down to where it's selling price is affordable for individuals to buy (couple of hundred dollars). Here again, it's a device sold to a broad market, where he must be making them by the tens of thousands, and undoubtedly there are automated manufacturing steps involved. Similarly with the pulse oximeter. Zvi never said what price point he is aiming for, only that it should be "low", I don't think it needs to be as low as $25. Zvi also didn't say whether the market he is looking at is large enough to support the sort of low-cost manufacturing needed. First, he has to develop a device that will work, for whatever application he has in mind. Then the manufacturing engineers can worry about finding ways to get the price down to what the market needs. \o/ /_\ | ||
| Karl Norris (knnirs) Senior Member Username: knnirs Post Number: 47 Registered: 8-2009 |
Zvi, You continue to emphasize low cost for your device. The best example of a low cost spectroscopic device is the pulse oximeter to sense the blood oxygen level by a non-invasive measurement on the finger. This device uses two LED sources for the measurement and is available for about $25.00. If we assign a value of $10.00 for each wavelength we can then estimate a minimum cost for a device that might require 20 to 100 wavelengths. Unfortunately this cost requires the use of a low-cost detector such as the silicon detector which does not detect wavelengths above 1100 nm. If the measurement you wish to make can be done in the 600-1100 nm region, you have a chance to develop a low-cost instrument using LED sources and a silicon detector. | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 402 Registered: 9-2001 |
Venky - if you use the formula, then any change or drift in the dark value will be subtracted out, so that after the arithmetic is done the effect of the dark is zero: A = -Log (sample-0)/(Ref-0) = -Log (Sample / Ref) and then the reading is independent of the value of the dark reading. Ideally, an instrument should do that for every spectrum, just as it should measure the reference energy for every spectrum. As I said before, the minimum you have to do that is often enough so that it doesn't change between the time you measure it and the time you measure the sample and reference energies. How often those will be is something you have to determine empirically for your instrument and your application. Similarly with determining the reference with an equation or using a segment of the spectrum: IT DEPENDS. Some of those will be amenable to letting you do what you want to do and some won't. There are numerous possible causes for changes in the reference energy as well as for the changes in dark readings. There are books on the market that discuss spectrometer design and operation, causes of "normal" changes and also changes due to fault conditions. If you're going to involve yourself with these inner workings of the instrumentation, you should read one or more of them. They will give you much more information, in a more coherent form, than can be done on this discussion forum. \o/ /_\ | ||
| venkatarman (venkynir) Senior Member Username: venkynir Post Number: 129 Registered: 3-2004 |
Hi Thanks Howard Mark !. I accept the for making mistake in typing A calculation . A = -Log (sample-dark)/(Ref-dark) Is correct . I do accept that blacking of light DARK . When it drift in counts what to be done . More over the Ref ., how frequently we have to take. Can I drive ref using mathematical equations .? Can I use segment of the spectra not attenuated as Ref . | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 401 Registered: 9-2001 |
Venky - I presume that's an InGaAs array. Not quite sure what you mean by dark current varying with "timing". In semiconductor detectors, however, true dark current will vary with temperature. However, "dark current" may not always actually be dark current. If there's any light leakage at all into the optical path that gets to the detector, then the response to that light, when it's supposed to be completely dark, will be included in the "dark current" measurement. Then, if that light varys it will also appear as a variation in the "dark" current. To some extent, the detector sensitivity will also vary with temperature. It is these external sources of variation that require the use of the more complicated formula you describe: A = Log (sample-dark)/(Ref-dark) Incidentally, your formula is not quite correct, it should be A = -Log (sample-dark)/(Ref-dark) The reason for the formula is so that: 1) First, subtract out the dark reading, so that all measurements are corrected to a "zero dark" value. 2) Second, divide the sample energy reading by the reference energy reading, so that changes in the instrument (lamp intensity, detector sensitivity, etc.) are corrected to a constant reference value. Presuambly, the changes between sample and reference readings should be slow compared to the time between the readings. If the instrument can change between sample and reference readings, then the formula won't work. How to implement all this will depend on the instrument, on the nature of the "on-line" measurement and how you can make the necessary measurements for the process involved. It shoudln't matter whether there's a sample in place or not when you measure the dark reading, since you should be blocking all light through the optics! You can sssume anything you want about the dark reading, but if you guess wrong you're not going to be able to make any sense out of your data. This is my own inimical way to say that if you want to make any assumptions, you'd better first check whether those assumptions are valid. And in the case of something like a dark reading, how long the last dark reading remains valid!! \o/ /_\ | ||
| venkatarman (venkynir) Senior Member Username: venkynir Post Number: 128 Registered: 3-2004 |
Dear All; In Diode array spectrometer( 1000 nm to 2200nm with 512 pixles ) is the dark current vary with timing ? What about refereence spectra ?. Is it necessary to calulate A=Log (sample-dark)/(Ref-dark ) . How do to that in on -line application. Can I take Dark at once and pursume that it remains same ? Dark with sample is advisable or Dark with out sample is prefered . How frequently we have update Dark& ref in on-line . Please suggest. | ||
| Gabi Levin (gabiruth) Senior Member Username: gabiruth Post Number: 47 Registered: 5-2009 |
Hi Zvi, I am in Israel, and through my association with NIR company, I have some knowledge about detectors, manufacturers, as well as knowledge about applications, instrument design etc. Please contact me by mail first, then I will contact you by phone if you provide your phone number. Gabi Levin [email protected] | ||
| Zvi Barnea (barnea) Junior Member Username: barnea Post Number: 9 Registered: 11-2010 |
Thank You Paolo, Zvi | ||
| Paolo Berzaghi (pberzaghi) New member Username: pberzaghi Post Number: 4 Registered: 11-2008 |
Zvi, you may want to check with Dr. Shenk (http://www.shenkanalytical.com/), he has put together a low cost portable system (detector, light source, battery and software) designed for agriculture application, but I think it can be easily adapted to you application. The software runs either Grams or Ucal prediction models. Paolo | ||
| Zvi Barnea (barnea) Junior Member Username: barnea Post Number: 8 Registered: 11-2010 |
Thank you, That was extremely helpful Zvi | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 376 Registered: 9-2001 |
Zvi - you're in luck!! I was somewhat idly scrolling down the list of companies I got from my Google search, past where I stopped before, and saw a link to Edmund Optical. They are generally known for providing hobbyist-level optics and such, and since the search was for InGaAs, I figured that they would have low-priced InGaAs detectors. Take this link and put it into your browser: http://www.edmundoptics.com/onlinecatalog/displayproduct.cfm?productid=2232 They even have prices for some of their InGaAs. \o/ /_\ | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 375 Registered: 9-2001 |
Zvi - very few manufacturers (of almost anything) post their prices on their web sites. They prefer anyone who is interested, contact their sales department or at least a sales representative. I also had a new idea: I put InGaAs as a Google search term, I got hits for a whole bunch of companies besides the few that I mentioned (Sensors Unlimited, Andor, Xenix, Fermionics, OSI Optoelectronics, Advanced Photonics - then I stopped scrolling). You should be able to get an equivalent result if you do that, although you'll still have to contact the companies to find out what price they charge. Keep trying; not all of them are so difficult to deal with. Still, glad I was able to help. And if you can make it to Pittcon next March, there will be some detector manufacturers there, and you can corner the salesmen in their booths until they give you an answer. \o/ /_\ | ||
| Zvi Barnea (barnea) Junior Member Username: barnea Post Number: 7 Registered: 11-2010 |
Thank you again Howard for your remarks. Obviously these discussions are not replacing a good classical learning; however, they did bring to a better insight. Naturally, once funded, there will be more professional people in this project like optic designer, electrical (opto) engineer etc.. Right now, I have to deal with the funding and it was important to me to better understand the milieu. My original aim, and frustratingly I am not there yet, was to get a general assessment on the prices of InGaAS sensors in the ranges of 900 � 1700 nm or 800 -1850 nm and prices of grating tools. Searching the web, including some of the companies you mentioned did not result in price list or an assessment. For some reason as they notice that the server is from Israel they either refer me to local agents who in some cases knew less than me??! Or just waited from me to mention a catalog number of an instrument (not parts) to be purchased. Other companies simply ignored my requests despite a long disclosure and identification questionnaire (e.g. Hamamutsu). So my initial target was to try to use this forum to get some clue on the prices of these specific parts. Along the way I learned some principles and terms, but unfortunately remained with the original unsolved issue. Naturally, the request for parts prices originated as a compelling factor by the funding systems. Nevertheless, it was very nice of you to compose these teaching and enlightening remarks and for that I am deeply grateful. Zvi | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 374 Registered: 9-2001 |
Zvi - the detectors in the first and second pictures are what I was trying to describe. In fact, in the second picture you can see the construction: the small dark object in the middle of the container, beneath the transparent cover, is the actual light-sensitive detector element. The wavelength range that the detector works over depends on the material that the detector is made of. Besides single-material detectors such as Si and Ge, there are numerous types that are alloys of semiconductors: InAs, InGaAs, InSb. A manufacturer can specify different materials and different alloys to tailor the detector characteristics. If you look closely, you'll notice wires that appear to be attached to the corners of it. In fact, there are probably two more wires, that can't be seen because they are hidden by the edge of the opening. Two of those wires are attached to the active detector element. The other two wires are attached to a thermoelectric (TE) cooler that is built into the assembly. Some types of detectors can operate at normal room temperature. Some types of detectors must be cooled, and even ones that can operate at ambient temperatures are often cooled to improve their performance (reduce noise, increase sensitivity, change the wavelength response). Therefore, sometimes the detector is packaged together with a cooler, as those two are. Some detectors, however, do not include any cooling capability internal to the package. When thermoelectric coolers are built into the package, they may be either single-stage or double-stage; the double-stage cooler can reduce the temperature of the detector element more than the single-stage can. The third picture shows a detector that actually contains 16 separate detector elements (i.e., 16 chips like the one in the second picture). That is the reason it is in a package with so many pins (actually, very similar to standard IC package). All those pins are needed in order to make connections to all the separate internal detector elements. Zvi, if I have to explain all this to you, then you're a long way from being able to build yourself an instrument. Instrument design is a complex undertaking, involving optics, electro-optics, electronics, mechanical design, and these days, computer interfacing and software. In the face of all that, there's only a limited amount of information that can be transferred on a discussion group like this. You need to take a course in instrument design, and probably also courses in optical design and detector technology, assuming you already know about electronic design and how to set up the mechanical parts of a device. I strongly suggest you find a good university course to take, or at least read some books, to at least learn what you're getting yourself into. \o/ /_\ | ||
| Zvi Barnea (barnea) Junior Member Username: barnea Post Number: 6 Registered: 11-2010 |
Thank you Indeed, I do not know what the meaning of "detector head" is but on the other side I do not know what is the difference or what is the meaning in your term "detectors come mounted in an enclosure with standard sizes and formats"? I did search in your previous list. The only partial success I got with "Hamamatsu" (they probably have different sites). I could not get any quotes from the list I saw and thought that they might fit � someone told me they do not deal with any company, they have to find out that it is not for military/defense purposes. Regardless, I did not get any reply to my 3-4 requests. As to your note: "you want a detector that's already sealed in a casing with the leads attached" do you mean to one of the following; The 1st attached figure relates to a group of sensors with 900 � 1700 nm. The 2nd figure relates to a group of 900 � 1850 nm -"long waves"(with 1, 2 or none cooling); Type No. Active Area Cooling Response Range peak G5851-203 di 0.3 mm 2 stage TE 0.9-1.85 um 1.1 100 MHz G5851-21 dia.1 mm 2 stage TE 0.9-1.85 um 1.1 A/W 40 MHz G5851-23 dia.3 mm 2 stage TE 0.9-1.85 um 1.1 A/W 3 MHz I also found : InGaAs 16-element-PIN photodiode array- 0.08x0.2mm pixel- 0.9-1.7um 3rd figure (have no idea if this means "head detector"?) Unfortunately, I do not get prices to any of those and for the other group (900 - 1700 nm). Plus except for the range, I do not the other criteria for appropriate choosing.    | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 373 Registered: 9-2001 |
Zvi - it's not clear what a "detector head" consists of, but it sounds like you're looking at the instrument manufacturer's listing, instead of detector manufacturer's listings. If you want to buy a detector, you need to discuss our needs with detector manufacturers. I listed some of them in a previous message, and that is only the "tip of the iceberg" as the saying goes. I think that in any case, you want a detector that's already sealed in a casing with the leads attached; you don't want to deal with a bare detector chip that you have to weld 0.01 mm wires to, to make contacts, so if that's what you're thinking of as a "detector" or "detector head" I suggest you not consider it. Tyically, "detectors" come mounted in an enclosure with standard sizes and formats (e.g., TO-3, TO-46, etc) and with leads attached so you can handle it and easily connect it to your circuitry and mount it in your equipment. F'rinstance, a TO-46 enclosure is 0.18" (4.6 mm) in diameter and 0.23" (5.8 mm) long, typically with 1-2" (25-50 mm) leads. Unless you're already an expert in semiconductor technology and handling (and it doesn't sound like you are), this is lowest level of detector that you want to deal with. Again, you need to talk to the detector manufacturers, e.g., Judson, Hammamatsu, Peinceton Instruments, etc. There's an extensive list of manufacturer's in the Photonics Buyer's Guide, or you can look in Spectroscopy magazine, or in Applied Spectroscopy, Applied Optics and other journals and magazines oriented toward the optical community. \o/ /_\ | ||
| Zvi Barnea (barnea) New member Username: barnea Post Number: 5 Registered: 11-2010 |
Thank you all for your input I am sorry for my delay in responding. My application that will be online determination (in medical term we call it also in-vivo) of several organic compounds in biological media, is based on validity studies that we did in-situ; meaning we took the samples to a very good spectrometer (instrument). Using the GRAMS software, we got there fantastic results Since it is not practical to bring this spectrometer to each sample and since it should be on line, we are aiming to down grade it in size and probably somewhat in quality to online & in-vivo dimensions. We realized that we should repeat all our studies in a more feasible instrument (commercial) with expectation that we will have to build the method from beginning, as the smaller and cheaper instrument will not produce the similar data matrices as the "big" instrument. Resolution will be probably different and hence the results will shift. Our GRAMS based method of determination uses most of the range between 900 and 1850. We "trade off" correlation of 99.8% to 98.9% with a decrease in the long wave range to 1700 nm. However, I am quite reluctant to do it; I still think we may get this additional 150 nm in detectors without increasing that much the price. So in relating to Howard's note, I would say that narrowing the ranges so not seem to reduce the price (at least now). Now if I try to analyze the prices. I see a complete instrument (900 � 1700 nm) in 3000 $ and InGAs detector head for $1460.00, however I could not find a price list for the detectors themselves � only "Detector head". I think that there is the secret � transducers by themselves can be cheaper. But I was unable to get quotes. Best, Zvi | ||
| venkatarman (venkynir) Senior Member Username: venkynir Post Number: 123 Registered: 3-2004 |
Hi Barnea; I am happy to know that your are entering to NIRS Instrumentation. First be clear what you want to mesure and accuracy that is expected? Take samples of the specimen for enter range of measutrment study the signature . This will give you range selection and detector specificaiotn . Then with collected samples use the software tool for model building and validaiton . This will help you whether filter type or DAS. Finally if it is indutrial applicaiotn better TE cooled one ( S/n) better at that enviormental Once you finish this then think aobout vendors . Formulate you specification first .Ok But $ 1000 very difficults build NIRS Instrument .It is possible as Mark suggested filter type and your own instrmentation design | ||
| Art Springsteen (artspring) Senior Member Username: artspring Post Number: 30 Registered: 2-2003 |
Howard makes a very good point. A single, well regulated source (tungsten), a filter wheel with three bandpass or notch filters, and a single InGaAs (or even PbS- cheap!) detector and appropriate electronics and you've got a workable 'spectrometer'. That you can do for hundreds of dollars. We built our first workable spectrometer (for measuring motility of turkey semen- don't ask!) from two LED's and a Si photodiode. I doubt the parts cost was $40. Sometimes you just have to think outside of what we normally think of as a spectrometer. | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 371 Registered: 9-2001 |
Zvi - I think there may be a possibility for major cost savings, depending on the answer to these questions: why do you need a 100 nm range centered around the three wavelengths you mention (950, 1350 and 1700 nm)? Is it possible that you can do what you want to do if you measure only at a single wavelength in each of those ranges? If you can do what you need to, using only a single wavelength in each of those wavelength ranges, then you won't need to use a diffraction grating or multiple detectors. You would be able to use three interference filters, one for each of the chosen wavelengths, and a single detector, to make the measurements you need. One of the reasons commercial spectrometers are so expensive is that the least expensive ones need to use an array detector along with the diffraction grating, and the array detectors are much more expensive than a single detector is. You would also need to devise a method of changing the filters and possibly also include a dark reading, a "white light" reading and other suitable measurements, but the technology to do this is well-known (e.g., a rotating wheel that has the filters mounted on it), and is relatively inexpensive to implement. You would also, of course, need to devise and build the mechanical parts and electronics, to control the instrument and collect the data. Then you would not need to build a scanning spectrometer at all, this much simpler device might do what you need to do. And it might just only cost you "hundreds". \o/ /_\ | ||
| Zvi Barnea (barnea) New member Username: barnea Post Number: 4 Registered: 11-2010 |
As mentioned, I am, new in the field. Accordingly I am not sure I am coping with you. I will try to make myself clear with hope that it can booster some help. I did learn now that the "whole machine" is called "instrument" or tSpectrometer. We aim to buy one and for this discussion I will designate it as "the reference instrument", I saw several of those and our budget allows us to buy it for ~ 15000 $. I realized that this is feasible. Now, before I go to the second part, which is an integrating parts (light source, grating, detectors -transducers & 2 fibers optic), in a special device that we build, I would like to note that a new (for me) term were raised � "NIR engine". Is it an additional term to "instrument" or it refers to NIR spectra consecutive detectors panel that includes the long waves? Or else? Regardless the meaning, my new device can work in the range of 900 to 1700 nm (I would prefer that it will reach 1900 nm, but I understand that this increment causes the steepest price increase (correct?). Moreover, we do not need the all spectra, rather around 100 nm range in the 950 nm zone, additional 100 nm range in the 1350 nm zone and the last 100 nm range is in the 1700 nm zone. Now with hope that my aim and needs are clearer, I would like to focus my question: I need to assess the combined price of: light source (Tungsten?), grating, InGaAS transducers set for the above specified regions and 2 fibers optic. My desirable price for the integrated parts is within 1000 $ and any "real life" notes or corrections are welcomed as well as referring me to the right places. Sorry for my ignorance, but I hope I could deliver my massage appropriately enough. | ||
| Art Springsteen (artspring) Senior Member Username: artspring Post Number: 29 Registered: 2-2003 |
Howard Mark wrote: "yes, that is down considerably from what I recall, but it's still two orders of magnitude away from what Zvi said he wanted to spend, which was "hundreds"." Good luck on that. We see these types of posts fairly regularly on this user group. I think everyone needs a reality check. You can't get an NIR spectral engine for hundreds (or even low thousands) of dollars. Not to sound like an old fogey, but a simple web search could have led someone to that. | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 370 Registered: 9-2001 |
Art - yes, that is down considerably from what I recall, but it's still two orders of magnitude away from what Zvi said he wanted to spend, which was "hundreds". \o/ /_\ | ||
| Art Springsteen (artspring) Senior Member Username: artspring Post Number: 28 Registered: 2-2003 |
Howard Mark wrote: "It was done by a company called Axsun, which was bought out a few years ago. The price, however was not commensurate with the size, unless you compared it to a diamond. The Axsun spectrometer sold for roughly $60,000, as I recall. " The Axsun spectral engine is still available, I'm pretty sure. A number of our customers still use it in their spectrometers and the price (I think) has really dropped from the figure Howard quoted. I'm guessing you can get them for <$10K now. | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 369 Registered: 9-2001 |
Zvi - I forgot to mention: the smallest commercially available assembled instrument was built using semiconductor processing technology and was able to fit a complete spectrometer on a silicon chip about the size of a standard-size IC; that's about 10 mm or so. It was done by a company called Axsun, which was bought out a few years ago. The price, however was not commensurate with the size, unless you compared it to a diamond. The Axsun spectrometer sold for roughly $60,000, as I recall. Most other of the "small" spectrometers fit into boxes roughly about 5" x 3" x 2". So unless you've got some secret methods in mind for fabricating your instrument, don't expect to make it much smaller than that; you'd have to do something like what Axsun did, to get it down to dimensions of millimeters. Howard \o/ /_\ | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 368 Registered: 9-2001 |
Zvi - the generic name for an assembled analyzer is an "instrument", or perhaps a "spectrometer". But you're correct: they are not cheap. The smallest and simplest ones I know of cost close to $10,000. Fancy, expensive ones, fitted out with all the bells and whistles can cost over $100,000. You can find less expensive instruments, for as lttle as around $1,000, if you can limit the spectral range to about 1100 nm; then an instrument can use a Silicon detector, which can be bought for a few dollars; these are the types used in cameras and "electric eyes", and other consumer goods, which is what brings down the price. But for longer-wavelength capability, you're going to have to pay. And for the lower-cost long-wavelength types, most of the extra cost is in the detector (the component, that I previously called the "transducer"), so you're not going to save that by assembling your own. \o/ /_\ | ||
| Zvi Barnea (barnea) New member Username: barnea Post Number: 3 Registered: 11-2010 |
Two seperate issues. 1. I need a complete device 2. I want to build an apparatus for certain applications from parts | ||
| Fernando Morgado (fmorgado) Junior Member Username: fmorgado Post Number: 7 Registered: 12-2005 |
Hello : Something in not clear for me. You have a NIR Instrument and you want install aditional detector? or you want to buy a complete NIR with range 2200 nm or similar. ? Fernando | ||
| Zvi Barnea (barnea) New member Username: barnea Post Number: 2 Registered: 11-2010 |
Thank you Howard, I guess you got my points correctly. I am interested in; 1. A complete device that works with GRAMS software and covers the mentioned spectra of 900 to 1850 nm. If it gets to 2200 nm it would be better. This would be our reference device. 2. The actual respective (to the reference device) transducers/detectors/sensors or whatever term is commonly used. These detectors are to be integrated in an apparatus (with light source and fiber optics) for a certain specific application. Here I got confused with the web search. I found only expensive prices, plus I am pretty ignorant as regard to define needed characterizations (except for the spectra and their physical size � few mm.) Our budget is in the range of few hundred $, but I found only in the range of thousands $. Regardless, I noticed that these IngaAs semi conductors are becoming even more expensive when the long-wave cutoff exceeds 1700 nm (the "extended Inga As") to meet our needs of ~ 1850 � 1900 nm for the long-wave cutoff. Zvi | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 367 Registered: 9-2001 |
Zvi - it's not clear whether by "detector" you mean the actual transducer that converts NIR energy to an electrical signal, or you mean a device that will measure a spectrum in the range of interest to you. In either case there are several types and a numebr of manufcturers. If you mean the actual transducer, however, we can narrow it down somewhat, because you probably would want a semiconductor detector, and would want to consider one whose active element is made of Indium/Gallium Arsenide (usualy abbreviated InGaAs). Depending on the formulation, the long-wave cutoff can be as low as 1700 nm, or as high as 2200 nm (commonly called "extended InGaAs"). The extended InGaAs will encompass your range of interest. Manufacturers include Hammamatsu, Nunavut, Intevac, Optronic Laboratories, Sensors Unlimited and a whole bunch more. You need to look in optical directories for more manufacturers. If by "detectors" however, you mean small spectrometers, again there are many, including BaySpec, (again) Hammamatsu, Thermo, and others. There are many instruments that are large and constitute a comlete analysis system, these might not be within what you consider a "detector". You can do a web search, in which case your most difficult job will be to weed out the instruments that are in different spectral regions, larger than you are interested in, and so forth. What you might do is attend the upcoming Pittcon; manufacturers of instruments as well as transducers will be there. Howard \o/ /_\ | ||
| Zvi Barnea (barnea) New member Username: barnea Post Number: 1 Registered: 11-2010 |
Hi, I am new in the field. I need your input regarding getting detectors in the regions of 900 to 1850 nm (could be in separated groups) for an application tool. I will appreciate any one who can direct me to such site (if possible with reasonable prices) |