| Author | Message | ||
     David von Boisman (david_von_boisman) Senior Member Username: david_von_boisman Post Number: 29 Registered: 2-2006 |
Jim Burger, when we talk about expensive lamps from high-end vendors, you need to keep a few things in mind: 1) A NIR specialist with extensive knowledge can probably evaluate his own lamps and rebuild his setup if necessary to get the right amount of light etc. 2) Instrument vendors has to make sure that their instruments are stable and transferable. The lamp is generally replaced by the customer who in most cases only has a basic idea of what the instrument does. So it basically has to work out of the box or the instrument will not pass diagnostics, and will produce funny results. So vendors assume a big responsibility when selling lamps which is reflected in the price. | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 474 Registered: 9-2001 |
A long time ago I read an article, I forget by who and I forget where although there's a good chance it was in Appl. Spect., where the author investigated fluctuations in sources. One of the unexpected results I remember was that putting a cylindrical housing (chimney) around the lamp, to control and direct the flow of air, caused more fluctuations than leaving the lamp in "free air" as long as the air was still. He attributed that result to the enhanced cooling effect of air flowing past the lamp in the chimney, creating more cooling of the lamp envelope, but still in an uncontrolled manner so that the envelope temperature was less constant than in free air. \o/ /_\ | ||
| Joelle Rahm� (rahm�) New member Username: rahm� Post Number: 4 Registered: 10-2011 |
Thank you so much for all your answers. Yes, the lamp reflector was the one that was giving me the funny sinusoidal shape. I tried today with another lamp with no reflector and the spectrum from the spectralon looked much better. Still, I am concern because the top of the spectrum is not as flat as the manufacturers usually show in their booklets (possible detector non-linearity?) and Howard, I am also concern about this periodical high frequency periodical signal you noticed. I keep on having these peaks. The frequency seems constant: up and down. I have used two spectrometers from the same company, the same model, and they both do this. Removing the black measurement (black noise) does not lead to a smooth spectra. I am wondering if mean averaging smoothing is going to be fine, and if taking every other wavelength would be ok too given that the resolution I have is kind of low. And a last question for Jim, do you think the diffusor would make a big difference? I have heard is common for imaging units. Could I survive without one for single point measurements? Thank you all again! | ||
| Gustavo Figueira de Paula (gustavo) Senior Member Username: gustavo Post Number: 29 Registered: 6-2008 |
Joelle, You can try using a polished aluminum reflector made in house. Aluminum is easy to polish to a mirrored surface and will not create artifacts on the spectrum. The baseline drift due to poor lamp quality may be (at least partially) overcome by frequent measurements of the white standard - one measurement before each sample measurement on a extreme limit. Gustavo. | ||
| David Semmes (dsemmes) Junior Member Username: dsemmes Post Number: 10 Registered: 6-2010 |
Jim, thanks for sharing your experience! When you write about the importance of regular, correct, calibration with standard reference materials for instrument standardization, are you referring to absolute reflectance or reflectance relative to an internal standard? Could you write a little about what types of changes you�ve observed, and the cause of those changes? Do you calibrate with typical sintered PTFE standards? Could you write a little about your process? Thanks! | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 473 Registered: 9-2001 |
Joelle - I think you have more than one problem. If you look closely at the region between 1500-1650 nm you'll notice a second, high "frequency" cycle superimposed on the low-"frequency" mountain. To a lesser extent, there's another one on the peak between 1200-1350 nm, too. Howeard \o/ /_\ | ||
| venkatarman (venkynir) Senior Member Username: venkynir Post Number: 149 Registered: 3-2004 |
Hi Joelle You spectrum look like damping wave ( sinsoidal nature ).I agree with Mr.burger statement .I used ordinary "auto lamp" .I have not find any sinusoidal wave form.Please check | ||
| Jim Burger (jburger) Junior Member Username: jburger Post Number: 10 Registered: 11-2010 |
Hi Joelle, I have experimented with a lot of NIR light sources for use in hyperspectral imaging. In this case, both the spectral output AND spatial distribution are important factors. I tend to agree with 'most' of the previous comments: The power-supply is critical: use one that is 'current' regulated, not voltage; and Direct current (DC) not alternating current (AC). And yup, all cables should have good fittings and make very good connection. Cooling can also be an issue: try to create a static environment for the light source which reaches equilibrium after time 'x'. But the lamps themselves? The important thing is the reflector. Many 'household' lamps have a dichroic reflector, purposely designed to minimize IR reflectance (heat). Obviously these lamps should be avoided for NIR work. But otherwise, I have had excellent results with 'cheap' lamps... everything from 10 - 20W 'mini-lamp' bulbs, to 55W car lights, to 300W 'garage' illumination lamps. (all with DC current regulated power!!) I don't buy the marketing claim from the high-end vendors. I don't believe the 'expensive' bulbs are manufactured any differently - and any 'quick test' to sort their quality is no different than what you can achieve in the lab with a few samples. Instrument standardization is achieved with correct calibration. ALL quartz-halogen bulbs are going to age, and show a drop in output. That's the nature of the halogen cycle. What is critical, is that your system is REGULARLY calibrated with known standard reference materials. My trusty demo imaging system is powered by Ikea! Still going strong after 3.5 years of abuse. You mentioned measuring small animal feed pellets. Keep in mind that your sample has a height of 4 - 5 mm, and light intensity at two different path lengths may also contribute to signal variance. You might consider using some kind of spherical or hemi-spherical diffuse reflector to help minimize this effect. | ||
| Gabi Levin (gabiruth) Senior Member Username: gabiruth Post Number: 66 Registered: 5-2009 |
Hi Joelle, The signal curve you are getting indicates clearly that the intensity you send to highly the "flat" response (spectrlaon?) sample is sinusoidal in nature. I would look for different reflectors that will have a flat wavelength reflectivity coating - or even better - just polished metal without any coating to minimize reflectivity variability with wavelength. The posibility of constructive and destructive interference of the light at different wavelengths when launched and reflected is not out of the question - to minimize this - if indeed it plays a role - you can play with the incident light angle with respect to the collecting angle - this could help to remove such an effect. If the wavelengths at which the max and min signal occur shifts with angle - it could indicate that interference is an issue. Other aspects of lamp quality are of course very important, but do not cause such phenomena. Gabi Levin | ||
| Joelle Rahm� (rahm�) New member Username: rahm� Post Number: 3 Registered: 10-2011 |
Thank you so much for you fast answers! Brian, indeed my spectrum have sinusoidal features (  ). David and Art, I have and use normally a Tungsten Halogen lamp from a recognized NIR manufacturer, but the signals I was getting were very low. These cheap lamps were just for experimenting with more light (so the cabling is also not so good, the power supply not very controlled...). Would you know how else can I improve the intensity of my signals? Thank you very much! | ||
| Art Springsteen (artspring) Senior Member Username: artspring Post Number: 43 Registered: 2-2003 |
Both David and Brian are correct. You should get your lamps from a reputable supplier (in the US, someone like Gilway, although there are many others). Make sure the reflector is not coated- in fact, no reflector at all is a good idea. And be sure to have a very good, current controlled power supply for stability. Otherwise, you'll never be able to hold a good baseline. | ||
| David von Boisman (david_von_boisman) Senior Member Username: david_von_boisman Post Number: 28 Registered: 2-2006 |
You will not get a stable signal from a cheap lamp, in particular if you use on with a reflector. NIR Manufacturers either buy lamps with very tight specifications, or select them from large batches bases on criteria such as stability and light level. That means, they are not cheap. You also need to make sure that your connectors and cables are REALLY good as anything that creates a bad connection, or a change in current, will generate drift. And there are more things to consider, I believe other members on the forum can fill in on that. | ||
| Brian Curtiss (briancurtiss) New member Username: briancurtiss Post Number: 4 Registered: 6-2006 |
The problem may be the reflector. Many reflectors have coatings that result in interference fringes - does your spectrum have sinusoidal features like these: http://tiny.cc/4opjs ? | ||
| Joelle Rahm� (rahm�) New member Username: rahm� Post Number: 2 Registered: 10-2011 |
Hello, I want to use an external halogen lamp for my set-up (reflectance mode). I bought one from a conventional store to try it, 35 watts and it says something like a 10 degree angle of the beam, it is metallic-coated. When I measure the blank reference material (spectralon), I get a funny spectrum (in raw sensor counts) with 3 or 4 broad peaks. I tried positioning the probe several ways, and moving the light up and down, but the spectra keeps looking like that. When I move the light and the main beam does not hit the reference material, but the softer light, then the spectrum looks more like it should (montain shape). But... I get a too low intensity overall, reaching below 50% of the achievable signal by the detector (or saturation limit). So my questions are: what is happening with this light? does it have anything to do with the degree of the light beam? can I work with the light directly without using any lenses to colimate? Should I use any light diffuser? Any kind of cheap lamp that you could recommend me that gives a stable signal from wavelengths 900 to 1600? My end use are going to be small samples, like small animal feed pellets. Thank you in advance! |