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.

  • TI Thinks Resolved

DLP2010: Absurd results between 1600nm and 1700nm

Prodigy 30 points

Replies: 6

Views: 117

Part Number: DLP2010

Hello there,

I have been experimenting recently with the DLP2010NIR and I consistently get absurd results above 1650nm. I tried with different kinds of samples (plastics, wood, skin) and in all cases there a huge drop in reflectance.

When looking in depth at what could cause this, it appears that the values of the reference signal are different from one measure to another while I did not touched the calibration parameters.
The shape of this reference signal stays roughly the same thought. The absurd measurements seems to come from this reference signal getting too close to the sample one on this wavelength area.

The thing is I need to study a reflection peak around 1680nm. What could fix my problem ?

I tried to use another calibration than the factory's one, and it got worse: below is an exemple of what I got.

Wavelength (nm) Absorbance (AU) Reference Signal (unitless) Sample Signal (unitless)
1674,16587 0,145373 245817 175889
1677,12038 -inf 0 156068
1680,07059 -inf 0 136024
1683,0165 nan -2147483648 116812
1685,9581 nan -2147483648 99065
1688,8954 nan -2147483648 83525
1691,82839 nan -2147483648 70496
1694,75708 nan -2147483648 59851
1697,68146 nan -2147483648 50694
1700,60154 -inf 0 43041

Thank you for your help.

Best regards

Elliot

  • Elliot,

    Welcome to the E2E forums, and thanks for your interest in our DLP technology.

    Thanks for providing this data table. Does your system behave normally in the regime below 1670nm? Also, please share what firmware version and system version you are using.

    Regards,

    Philippe Dollo

  • In reply to Philippe Dollo:

    Hi Elliot,

    "I tried to use another calibration than the factory's one, and it got worse: below is an exemple of what I got."

    Could you please share details about the reference used? How did you use at reference standard  and configuration?

    The EVM has a limitation for SNR in the the wavelength 900-950nm and 1650-1700 nm. The SNR is relatively low for the following reason:

    1. The light output level from the lamp declines significantly in the extreme band of wavelength 900-950nm  and 1650-1700nm.

    2. In addition, the sensitivity of the InGaAs detector, it’s highly variant when temperature goes for these extreme wavelength bands.

    In my own experiences, I discard data on those extreme band and use only  range form 950-1650nm for consistent performance.

    If you application relies completely depends  on 1680nm wavelength only than you may want to consider using EVM which support boarder wavelength 

    http://www.ti.com/tool/DLPNIRSCANEVM

    regards,

    Vivek

  • In reply to Philippe Dollo:

    Philippe, thanks for your answer.

    The system does behave normally below 1670nm.

    The firmware version is 2.2.0 ; the system version is 2.1.0 for both Tiva SW and GUI ; The DLPC Flash version is 2.2.0.

    Many thanks,

    Elliot

  • In reply to Vivek Thakur:

    Vivek, thanks for your answer.

    The reference tried was just a white desk that we use to put the samples on. We can have access to spectralon references thought.

    As for the configuration, here it is:

    Method: Column 1
    Host Date-Time: 13/9/19 @ 16:6:31
    Header Version: 1
    System Temp (C): 34.44
    Detector Temp (C): 32.06
    Humidity (%): 20.13
    Lamp PD: 3863
    Shift Vector Coefficients: -3.34782 0.033145 -7.99822e-5
    Pixel to Wavelength Coefficients: 1801.48 -0.921703 -0.000239148
    Serial Number: 6110032
    Scan Config Name: Column 1
    Scan Config Type: Slew
    Section 1
    Start wavelength (nm): 900
    End wavelength (nm): 1700
    Pattern Pixel Width (nm):    9.37
    Exposure (ms):   1.270
    Digital Resolution: 228
    Num Repeats: 6
    PGA Gain: 64
    Total Measurement Time in sec:   3.585

    Thank you for linking an adapted device, but it is out of budget at the moment.
    Do you know ways to somehow work around those hardware limitations ?

    Many thanks,
    Elliot

  • In reply to Elliot BERARD:

    Hi Elliot, 

    You can try taking references between scans to improve the accuracy. Also, you can also try increasing the Num Repats to average more number of scans, which may give you increased consistency. Some corporations have also designed their own ML algorithm to improve the accuracy.

    Thanks & Regards,
    Hirak.

  • In reply to Hirak Biswas:

    Hi Elliot, 

    Hope we were able to help you with your query. I'll be closing this thread due to inactivity. You can reopen anytime!

    Thanks & Regards,

    Hirak.

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.