Author |
Message |
Pierre Dardenne (dardenne)
Senior Member Username: dardenne
Post Number: 46 Registered: 3-2002
| Posted on Wednesday, May 05, 2010 - 3:18 am: | |
Christian, David is right; a natural sample can change slightly over time. We have had this problem too. The predicted values depend on the equation model and the one provided by Foss was sensitive to small changes (moisture, temp., �). If you repredict your check cell records with another model, you could reach to another conclusion with no trend at all. This is what we did using our own large model (7000 samples) for feed. Pierre |
David von Boisman (david_von_boisman)
Intermediate Member Username: david_von_boisman
Post Number: 16 Registered: 2-2006
| Posted on Tuesday, May 04, 2010 - 11:43 am: | |
Hi Christian, I understand from your profile that you're based in chile, is that correct? Have you asked the local FOSS company for assistance? They should know. Its normal that you see a light drift over longer periods of time, since the check cell is based on a natural sample and it will dry slighty over time. Also, never leave it in the sun or something like that. If you see a sudden change in the level of one of the constituents, then you may have a problem with the instrument. Id be happy to have a quick look at the scans, they usually say a lot about the condition of the check cell and the instrument. You can send me a note on david(at)msanjose.net. (that is, right-click on the check cell, select export / export NIR/cal) Best regards David von Boisman |
Christian Mora (cmora)
Advanced Member Username: cmora
Post Number: 25 Registered: 2-2007
| Posted on Tuesday, May 04, 2010 - 11:11 am: | |
Dear list members; I've been running checkcell readings periodically for around 5 months in our FOSS 6500 spectrometer. The checkcell is an Infrasoft International standard check cell kit 21468 S, part IH-0324A. For collecting the spectra we use ISIscan. Based on the help info (ISIscan) the check cell (at room temp) must have certain values for constituents such as protein and fat content +/- deviation. Based on our readings, the average values for all the constituents are different from what is shown in the help page (although the deviation is low) and what worries me out is that some of them show a sort of trend in time. We collect the spectra at 20 C degrees and approx 40% RH. Even when the t-stats reported by the software are within the suggested values, that trends are something that I would like to understand and maybe (statistically) quantify. Does anyone (I'm pretty sure that the aswer is yes) have experience with this kind of repeatibility tests that can provide some advice? If so, I would like to contact him/her off list and send perhaps some snapshots of what I'm trying to explain. Thanks in advance |
|