David W. Hopkins (dhopkins)
Senior Member Username: dhopkins
Post Number: 120 Registered: 10-2002
| Posted on Friday, August 31, 2007 - 8:34 am: | |
Hi Craig, There is a lot about FT-NIR and Grating Units in the Discussion Group archives at this point. Also, I would recommend that you review discussions regarding noise and sampling. I will just summarize my understanding of these issues as applied to whole grain and whole pellets. There is a real problem with sampling whole grain and pellets of feed, because the product itself is not homogeneous. In the reference lab, it is important to mix the samples thoroughly and take random samples, usually of 1 gram or more for analysis. This is usually done after grinding and mixing, to obtain good reference values of a representative sample. Care has to be taken that the moisture is not changed during handling, or that the moisture basis is tracked. For the NIR measurements, it is important to measure �the same sample� as measured by the reference method. This means that the light should pass through at least a gram of sample, and more if possible. This is usually done by using transmission measurements and scanning large volumes of sample with spectral averaging. That said, the limiting source of variation in food and agricultural measurements is still usually this sampling error, not the photometric error in the instruments. Therefore, it appears to me that users obtain satisfactory results from monochromator or FT instruments. The differing noise specs for the FT and monochromator systems are not the limiting factor. You have to go to other applications, where sampling problems are not the issue, and sensitivity requirements are paramount, before you need to consider the photometric errors as dominant considerations. Best regards, Dave |
Craig Hunt (craig)
New member Username: craig
Post Number: 1 Registered: 8-2007
| Posted on Thursday, August 30, 2007 - 6:59 pm: | |
Has any body evaluated FT-NIR vs Grating units on wholegrain and whole pellets. I am told that the FT may have less signal to noise in the shorter wavelengths up to 1400nm, being a post dispersive system. Is this true or sales talk |