• State TI Thinks Resolved
  • Locked Locked
  • Replies 11 replies
  • Answers 1 answer
  • Subscribers 30 subscribers
  • Views 5029 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Database / Reference Graphs

Deema Dabbagh
Prodigy 10 points
Other Parts Discussed in Thread: DLP2010NIR

Hello, I currently have the DLP NIRscan Nano in my lab and want to run some sample tests with it.

- What is recommended as the most simple and straight forward food item to scan?

- Does TI have an available database with reference graphs/scans of items that they've tested before? I would like to have a reference to use in order to compare my obtained results. 

  • 0
    TI__Intellectual 1375 points

    Hi Deema,

    The NIRscan Nano EVM has a sampling module which is designed for direct contact with the sample (see section 1.2.1: Optical Engine of the NIRscan Nano User's Guide). Therefore, uniform flat samples typically work the best. For samples which could be messy (powders, cheeses, etc.), the sample can be placed in a plastic bag to keep the sample from contaminating the electronics and optics of the EVM. In this case, you can also take a reference with the reference material in the same plastic bag in order to remove the absorbance of the thin plastic from the computed absorbance of the sample.

    TI has not have a full database, but we have published some example scans in the test data available on the TI Design page for this module. We have focussed on bringing this spectroscopy tool to the market and are excited to see others develop their own chemometric models, databases, and analytics.

  • 0 in reply to Eric Pruett
    Prodigy 55 points
    Hi Eric,
    Thank you for referring me to the sample scans. I was able to scan Wonderbread and somewhat replicate the results.
    I'm having trouble with understanding the absorbance graphs. I have followed the instructions in terms of setting up the scan configuration (spectral range 900-1700, digital resolution 228, method column, and number of scans to average 6). I've scanned a piece of metal as my reference. Nonetheless I am not getting any graphs for absorbance. The intensity is showing up ranging from 46385, 92770, 139155, 185540, 231928. The absorbance graph does not seem to show anything. When I import the data into excel, the absorbance graph ranges from 0 to -0.9. What can I do to fix my absorbance graphs to match those present in the sample scans you referred me to?
  • 0 in reply to Hamed Mazhab Jafari
    TI__Intellectual 1375 points

    Hi Hamed,

    In absorbance spectroscopy, two intensity spectrums are used to generate the absorbance spectrum:

    absorbance = -log10(sample/reference)

    This is done so that the instruments do not have to be radiometrically calibrated in order to measure material absorbance. This also allows changes in lamp intensity or large environmental changes inducing absorbance from water vapor, for instance, to be accounted for by simply repeating the reference scan. Therefore, the computed absorbance spectrum will vary based not only on the sample spectrum, but also the reference spectrum.

    Typically, the standard used for the reference is a diffuse, spectrally neutral reflector like Spectralon. This allows the absorbance spectrum to describe only the sample, since the reference is known to reflect all wavelengths being measured at an equal amount.

    In the NIRscan Nano version 1.1 GUI, a feature has been added which allows you to use a factory stored reference instead of measuring one yourself. This is convenient for simple use cases and is the same method used by the iOS app developed by KST. For stringent applications which are very sensitive to small changes in the absorbance spectrum, taking a new reference scan soon before the sample is scanned can reduce absorbance measurement variability. For more information on this feature, see section 3.1.1 of the User's Guide.

  • 0 in reply to Eric Pruett
    Prodigy 55 points

    Hi Eric,

    Thanks for your response.

    I have the NIRscan Nano Version 1.1GUI installed. In the User guide [section 3.1.1] you refer to, it notes that the DLP NIRscan NON comes with two pre-loaded scan configurations from the factory "Column 1" and "Hadamard 1".  The only setting that I have which is preconfigured is "Scan Config 1". How do I correct for this? The spectral range and digital resolution all match the specifications in the User Guide.

    I'm wondering if I should purchase the reflector Sepctralon you referred to. And also, under "utilities" there is a setting to "update the factory reference calibration data" would it be useful to click "scan" in order to restore factory settings and set up the reference ?

  • 0 in reply to Hamed Mazhab Jafari
    Prodigy 55 points
    Also, there are several Spectralon reflectance standards. There's the Calibrated diffuse reflectance standard single, the unclaibrated, the color standards". Which of these do you recommend for purchase, since its not specified in the user guide?
  • 0 in reply to Hamed Mazhab Jafari
    TI__Intellectual 1375 points

    Hamed,

    Can you confirm the version of the "Tiva SW version" shown in the info page?

    It sounds like perhaps you have a previous version of the firmware installed on your Nano. If the unit came with firmware version prior to 1.1.8, the default configuration will be a single config named "Scan Config 1". If the unit came with firmware version 1.1.8, it should have two default configurations named "Column 1" and "Hadamard 1". You can also define your own scan configurations as described in the User's Guide.

    For a walkthrough on upgrading the Tiva firmware if indeed your unit has an older version, see this thread:
    e2e.ti.com/.../455187

    The 'Update the factory reference calibration data' section in the utility page of the GUI allows you to take a new scan with a reference placed at the sampling window and replace the stored reference in the NIRscan Nano. So to do this procedure, you would need to have a piece of reference material. This stored reference is used as the reference when you take a scan and select 'Factory' in the 'Scan Reference Select' section.

    For general purpose use, the 99% reflective white samples are what is used. They are also available from other sources like Edmund Optics and others.

    Some people use the darker grey standards as a reference when scanning very absorbent materials or to verify instrument linearity, but these applications are more rare. The color references are used to validate color measurement of instruments in the visible region, so they are not applicable in the NIR region of the NIRscan Nano EVM (900-1700nm) or the spectrum which the DLP2010NIR DMD can support with other optical designs (700-2500nm).

  • 0 in reply to Hamed Mazhab Jafari
    TI__Intellectual 1375 points
    Hamed,

    The difference between the calibrated and uncalibrated standards is that the calibrated standard comes with reflectance scan data from a calibrated high performance spectrometer. With this, you can take into account any small deviations between 98-100% reflectance of the reference standard while processing your collected data. So the calibrated standards become useful when you need absolute absorbance accuracy down to 0.01 absorbance units. This may be the case when you are doing high-end laboratory testing, or building a model for constituent concentrations rather than material identification.
  • 0 in reply to Eric Pruett
    Prodigy 55 points
    Eric- I have the version you are referring to.

    Tiva SW version 1.1.8

    GUI version 1.1.9

    DLPC Flash Version 1.1.0


    My absorbance spectrums are still giving wack ranges [-0.9-0] when graphed. I cant seem to correct for this and I'm inclined to believe its because of the reflectance sample.
  • 0 in reply to Hamed Mazhab Jafari
    TI__Intellectual 1375 points
    Hamed,

    Have you tried scanning a sample and selecting the 'Factory' selection in the 'Scan Reference Select' section of the scan page in the GUI? As long as you haven't updated the factory reference calibration data, this should give you a good approximation of the true reference as well, and yield expected absorbances.

    One thing you can do to take your own references while you are waiting for a true reflectance reference sample is to coat something with whiteout or liquid paper and use this as a reference. It is not as good because it's not fully diffuse and has some additives in it, but the primary compound is TiO2 so it is still >90% reflective in the 900-1700nm range. This should be somewhat correct and yield an absorbance plot similar to the test data I linked to originally, but will not be as accurate as the factory reference mentioned above.

    To summarize, reference options in order of accuracy are:
    1. A true Spectralon or similar material
    2. Using factory reference (thermal and environmental effects can have a small impact on the reflectance signal measured with a reference)
    3. Using a homemade reference with whiteout or liquid paper. (Will have less reflectance than Spectralon, but the spectral shape will be similar)
  • 0 in reply to Eric Pruett
    Prodigy 55 points
    Yes I already scanned three different kinds of bread using the "factory" settings under the 'scan reference select' sections. They are all giving me the exact same graphs at the same wavelength and intensity which makes no sense. I haven't updated the factory reference calibration data so I don't understand how the same graphs are being reproduced.
  • 0 in reply to Hamed Mazhab Jafari
    TI__Intellectual 1375 points

    Hamed,

    Sorry if I missed that you had already tried with the factory reference. That does indeed sound like something is not right.

    If you wouldn't mind, I'd like to take a look at the CSV output from some of the scans to try and diagnose the issue. I have sent you a friend invitation so we can do this offline from the forum if you'd prefer and then we can report back what the resolution was. Otherwise, you can attach the CSV files from a couple of scans here and I can take a look.