X-Y CCD array availability?

spatially-resolved - difficult

***In this TE (thought experiment), I was thinking of non-spatial, or XY in a plane. ***In this TE, just thinking of a one-off instrument. ***I have just been reading, "Handbook of NIR Analysis" - Burns. There is a short chapter on ftir, but other monographs on ae mentioned. The resultant spectra seem to have much better resolution than "dispersive" measurements.

Getting back to the Watec cam you mentioned for a moment, if one is doing reflectomery, there is a choice to change the illuminating wavelength rather than the capture bandpass...

I'll read more in this book, which goes into measurements in the petrochemical, agricultural, etc. industries. Much of that must be reflectometry.

Does the "one-off" rather than multiple instruments make a difference? It's an intriguing instrument. My interest is in finding needles in haystacks, if you follow...

Depends how much you want to spend on the optics ie how near IR but if you actually want kit that works something like the back end of the P-E OES kit that uses a low dispersion prism spectrum one way and a high dispersion blazed grating the other maps a spectrum onto a CCD with some redundancy. The technology was introduced c/a 1993.

If you insist on going too far into the IR you will have to use expensive refractive components and costs will rocket up.

Here is an example of a solar spectrum done with the same technology:

Your problem will be wanting to do it 30fp since you are not going to get many photons in a high dispersion spectrum in that short time slice.

--
Regards, 
Martin Brown

There's probably a good reason that you only see FTIR used in the IR and Far-IR. George H. (Who spent long nights with a Bomem FTIR spectrometer in grad school... It was painful to realign.. I heard the Bruker's were better, but never used one.)

They can. The two basic advantages of FTIR vs gratings are (a) throughput and (b) wavelength multplexing.

Advantage (a) (Jacquinot) A grating spectrometer has a 1:1 tradeoff of resolution vs. efficiency, because you have to cut down the slit width to get higher resolution.

The slit length causes a quadratic wavelength shift, so it's less of a problem.

In a FTIR, both the length and the width of the aperture enter quadratically in the same way, so essentially you can widen the slit to be as wide as it is tall.

Advantage (b) (Fellgett) A grating spectrometer with a single-element detector can look at only a single wavelength channel at a time, so if your resolution delta-lambda/lambda is 0.001, you lose 99.9% of your light all the time.

A FTIR is looking at all wavelengths simultaneously, and sorting out the frequencies afterwards via FFT.

The product of these factors can give a FTIR a signal increase of the order of 10**5 over a grating monochromator.

In the visible, advantage (b) more or less goes away because you can get multielement detectors cheaply. In addition, the motion has to be that much more precise, which is more expensive and takes more maintenance. Visible-light sources are also usually brighter, which makes the throughput advantage less important.

That sort of optomechanical precision emphatically does not happen by accident.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

that

recognition.

illuminated and TE cooled.

thought experiment, how would you make a camera that is also a spectrophotometer and goes, say, the range of In-GaAs. Ever think about that?

bandpasses. Or a diffraction grating with high efficiency that allows capture of a range. Or a series of sources, like leds, that allow wavelength-switching.?

next layer is the signal capture electronics and software. This layer does things like lock-in, signal averaging, image stacking, and data searching and management.

algorithms like Darwinian searching, ANN's, k-NN, etc. This layer is responsible for managing the camera behavior and exploring the data. KISS best principle here.

powerful. The machine tries say 100,000 combinations of wavelengths against some criterion we won't go into now. It notices a correlation in say 50 of those 100k. So it expands the 50 to another 100k and so on. It is managing what the camera is looking at PLUS what is done in the lower level signal processing.

Sounds like a PhD thesis to me.

?-)

ha ha. The Darwinian algorithm is actually quite practical. I programmed a version that predicted winding carbon fiber tape around an air frame shape, like a wing or nose cone. The tapes are laid down at 45 deg angles, and th e idea is to avoid "holidays" right above each other, in multi-layer wrappi ngs. Care to solve this analytically? So I just programmed the tape wrapping as a converging series of distance c alculations, maybe 1000 for each little advance. And the entire tape wrap w as one of 100,000 wraps, Then the 500 best were culled, and spun into a "se lected" pool of another 100,000 - etc. This program took one week to write from start to finish. Some were amazed, saying, "How did you figure that out?" I surprised even myself. No analysi s, no math beyond Pythagoras. jb

And to fill you in, the device is called "MSI," Multi-Spectral Imaging. Also Hyper-Spectral Imaging. Generally also "SI." It is made possible by "Big Gulp" image sensor arrays and electrically tunable filters such as FTIR, and has many variants. It has been applied t biomedical area, but also applications like remote sensing and astronomy. jb

Also Hyper-Spectral Imaging. Generally also "SI."

tunable filters such as FTIR, and has many variants. It has been applied t biomedical area, but also applications like remote sensing and astronomy.

Sounds like some interesting stuff. Can you recommend any books?

?-)