Professional Documents
Culture Documents
ISC NIRScan User Manual - EN - V1.1
ISC NIRScan User Manual - EN - V1.1
User Manual
Ver. 1.1
2021/02/05
www.inno-spectra.com
Thank you for choosing InnoSpectra! Please read
this manual carefully and follow the operating
instructions and application guidelines.
Index
Chapter 1 Introduction ...................................................................................................................... 1
1.1 Near-infrared Spectroscopy ................................................................................................ 1
1.2 NIRScan ............................................................................................................................... 2
1.2.1 Diffuse Reflective Module & Spectrometer ............................................................. 4
1.2.2 Transmissive Module ................................................................................................ 6
1.2.3 Fiber Input Module .................................................................................................. 8
1.3 NIR Measurements............................................................................................................ 10
Chapter 2 Getting Start and Software Operation............................................................................ 12
2.1 System Requirement ......................................................................................................... 12
2.1.1 Download Software ................................................................................................ 12
2.1.2 Install Software....................................................................................................... 12
2.1.3 USB Requirement ................................................................................................... 13
2.1.4 Getting Start ........................................................................................................... 13
2.2 ISC_NIRScan_Win_WinForms_GUI ................................................................................... 14
2.2.1 Scan Configuration ................................................................................................. 15
2.2.2 Scan Setting ............................................................................................................ 17
2.2.3 Saved Scans ............................................................................................................ 18
2.2.4 Utility ...................................................................................................................... 19
2.3 Getting Excellent Spectrum ............................................................................................... 20
2.3.1 Reference Signal ..................................................................................................... 20
2.3.2 PGA Gain Control.................................................................................................... 23
2.3.3 Pattern Width and Number of Sampling Points ..................................................... 28
2.3.4 Exposure Time ........................................................................................................ 30
2.3.5 Slew Scan................................................................................................................ 32
2.3.6 Operating Point ...................................................................................................... 34
Chapter 3 Operating ISC NIRScan .................................................................................................... 38
3.1 Diffuse Reflective Module ................................................................................................. 38
3.2 Transmissive Module......................................................................................................... 41
3.3 Fiber Input Module ........................................................................................................... 43
Chapter 4 Spectrum Analysis and Application ................................................................................ 45
4.1 Spectrum ........................................................................................................................... 45
4.2 Resampling ........................................................................................................................ 47
4.3 Application ........................................................................................................................ 48
4.3.1 Drug Identification ................................................................................................. 48
4.3.2 Brix Screening......................................................................................................... 50
Chapter 1 Introduction
1.1 Near-infrared Spectroscopy
Near-infrared spectroscopy (NIRS) is a spectroscopic method based on the
absorption of electromagnetic radiation at wavelengths from 700 to 2500nm.
NIRS is a form of molecular spectroscopy. NIR radiation is absorbed by molecules
through a mechanism involving molecular vibration. Not all bonds in a molecule are
capable of absorbing infrared radiation. Only those bonds that have a dipole
moment that changes as a function of time are capable of absorbing infrared
radiation. For example, O–H, N–H, C–H, S–H bonds are strong NIR absorbers. The
most prominent absorption bands occurring in the NIR region are related to
overtones and combinations of fundamental vibration. In vibrational spectroscopy,
an overtone band is the spectral band that occurs in a vibrational spectrum of a
molecule when the molecule makes a transition from the ground state (n=0) to the
second excited state (n=2) (Figure 1.1). The NIR absorption bands are shown in
Figure 1.2.
1
Figure 1.2 NIR absorption bands
NIR has unique properties which makes it very useful for the analysis of most solid,
slurry and liquid samples. NIRS in conjunction with chemometrics has revolutionized
much of the general area of analyses.
1.2 NIRScan
InnoSpectra Corporation (ISC) is devoted to the development of NIR analysis
solution. ISC is committed to providing a series of compact and cost-effective
DLP-based NIR spectrometers including fiber input modules, diffuse reflective
modules and transmissive modules in different wavelength range to support all sorts
of applications. The optical engine of the DLP-based NIR spectrometer adopts a
post-dispersive architecture, consisting of slits, collimating lenses, bandpass filters,
gratings, focusing lenses, DMD (digital micromirror device), light receiving lenses and
single-point InGaAs detectors. DLP-based spectrometers replace the traditional
linear array detector with a DMD for wavelength selection and a single point
detector as shown in Figure1.3. DLP technology in NIR spectroscopy provides the
following advantages:
1. Higher performance through the use of a larger single point detector (ϕ=1mm)
in comparison to a linear array with very small pixels (e.g. 50μm).
2. The high resolution DMD (854*480 pixels) allows custom patterns to optimize
the optical performance of each individual system.
2
3. With programmable patterns, a DLP spectrometer can use Hadamard patterns
to improve SNR.
Slit 25 μm
Detector 1 mm InGaAs (non-cooled)
Scanning type Linear (Column) / Hadamard / Slew
Module Fiber Input / Transmissive / Diffuse Reflective
Interface USB / BLE / UART
Sensor Humidity / Temperature
Power USB(5V/1A) / Built-in Li-ion battery (3.7V/1000mAh)
3
1.2.1 Diffuse Reflective Module & Spectrometer
1. NIR-S-G1 (900-1700nm)
NIR-S-G1 is a diffuse reflective spectrometer that integrates two built-in 0.7W
tungsten filament lamps as illumination sources, a sampling module with a sapphire
scan window, a DLP-based post-dispersive optical engine, a BLE module, an
operation panel, a rechargeable lithium-ion battery pack and an EMC shielding case.
4
Figure 1.4 NIR-S-G1
2. NIR-M-R2 (900-1700nm)
NIR-M-R2 is a diffuse reflective module that integrates two built-in 0.7W tungsten
filament lamps as illumination sources, a sampling module with a sapphire scan
window, and a DLP-based post-dispersive optical engine. Both power and signal can
be transmitted through USB or UART interfaces, while BLE module and operation
panel are optional. NIR-M-R2 is suitable for secondary development by system
integrator.
5
Dimensions: 76mm * 57.6mm * 26.8mm
Weight: < 80 g
6
Power: Micro USB / UART / Li-ion battery
Dimensions: 76mm * 91.8mm * 41.2mm
Weight: < 106 g
2. NIR-M-T11 (1350-2150nm)
NIR-M-T11 is a transmissive module that integrates one built-in 0.7W tungsten
filament lamp as illumination source, a sampling module with cuvette holder and
receiving lens, and a DLP-based post-dispersive optical engine. Both power and
signal can be transmitted through USB or UART interfaces, while BLE module and
operation panel are optional. NIR-M-T11 is suitable for secondary development by
system integrator.
7
Dimensions: 48mm * 96mm * 38.2mm
Weight: < 100 g
9
Weight: < 65 g
There are four standard measurement modes for the acquisition of NIR spectra from
a sample: transmission, reflection, transflection, and interaction (interactance).
Illustrations of these are shown in Figure 1.10.
10
Figure 1.10 Sample presentation for NIR spectroscopy
11
Chapter 2 Getting Start and Software Operation
12
2.1.3 USB Requirement
All ISC NIRScan can be powered by USB cable. The USB specification is as below:
USB 2.0: 5V /0.5A
USB A male to micro-USB B male
Figure 2.2 LEDs: (a) initial status, (b) connecting via USB cable
13
Table 2.1 The details of LED indication
LED CONDITION DESCRIPTION
2.2 ISC_NIRScan_Win_WinForms_GUI
After connecting ISC NIRScan, PC, tuning on the power switch (figure 2.3) and
executing the ISC-NIRScan-GUI.exe, the software will check ISC NIRScan
automatically, as shown in figure 2.4. You can read the details of software in ISC
NIRScan GUI User's Guide which can be downloaded from the following path.
https://github.com/InnoSpectra/ISC_NIRScan_Win_Winform_GUI
In order to get the best GUI display, we recommend a screen resolution of 1366 x
14
768. The GUI displays three tabs on the left:
Scan: In this page, you can select Scan Setting, Scan Config, and Saved Scans on
right side.
Utility: You can realize the device information and perform firmware update in
this item.
About: You can learn about the license agreement and ISC contact information.
Figure 2.3 Power switch position: (a) standard module, (b) extension module
15
"Hadamard 1". You can customize the following parameters as needed. See figure
2.5 for the Scan Configuration Dialog.
Name: This item is used to name the defined scan configuration.
Num Scans to Avg.: This item is helpful to improve SNR (signal to noise ratio)
with increasing of Num Scans.
Num Sections: The wavelength can be separated into 1-5 parts between 900nm
to 1700 nm or 1350nm to 2150 nm, and all parameters can be set individually in
each part.
Scan Type: You can select Column or Hadamard mode to scan sample in this
item.
Spectral Range: You can select spectral range between 900 nm to 1700 nm or
1350 nm to 2150 nm.
Width in nm: This number selects the width of the groups of pixels in the
generated Column or Hadamard patterns.
Exposure time: The exposure time can be set in the range of 0.635 to 60.960 ms.
Digital Resolution: This number defines how many wavelength points are
captured across the defined spectral range. In general, we suggest the
resolution to oversample at least twice of the width. For example, for a 7.03 nm
width between 900 to 1700 nm, use 2 * (1700 - 900) / 7.03 ≈ 228 wavelength
points.
All parameters will be explained in detail in the following section “2.3 Getting
Excellent Spectrum”.
16
Figure 2.5 ISC_NIRScan_Win_WinForms_GUI Scan Configuration
17
Scan Average: This function allows users to select the amount of scan based on
the quality of spectrum. Generally, with increase of Num Scans, the SNR can be
improved.
Continuous Scan Select: For a series of repeated scans, this function allows users
to set amount of scan and interval time.
Save Scan As: This function allows users to select the data format and dictionary
to save the spectrum. We recommend that you have to choose the CSV file to
avoid data lost.
18
Figure 2.7 ISC_NIRScan_Win_WinForms_GUI Saved Scans
2.2.4 Utility
The Utility tab is shown in figure 2.8, it indicates:
Model Name
Serial Number
Date and Time
Sensors:
19
Advertising Name.
Device: This function allows users to update and restore the reference signal
when necessary.
20
New/Previous shown in Table 2.2 and this reference signal is stored in PC.
(a) Condition #1: The Built-in factory reference is used as reference signal, and
the pattern width, digital resolution of sample and reference are the same,
the absorbance is very closed to ideal situation.
(b) Condition #2: The Built-in factory reference is used as reference signal, but
the configuration of sample and reference are not the same, the absorbance
is not corresponding to zero.
(c) Condition #3: It is similar to condition #1, but the Built-in factory reference is
replaced with the local new reference, the absorbance result is also very
closed to ideal situation.
21
Figure 2.9 The effect of configuration: (a)both configurations are the same, (b)both
configurations are different, (c)both configurations are the same by scanning new
reference.
The Built-in reference signal is suitable for user doing a scan when there is no
reference sample available in hand. In order to match with the Built-in reference
signal, please choose same scan configuration to minimize the absorbance variation.
You can choose any scan configuration which perfectly matches with your sample
22
condition, please perform New reference scan regularly. You can select Previous
reference signal to match with your current scan configuration, then the absorbance
variation will be minimized.
Don not replace the Built-in reference signal unless you are sure that a Spectralon
sample is available to do reference scan.
Built-in reference signal could be replaced by the following way.
But we do not suggest you to replace this reference signal unless user has a reliable
reference material in hand and is very familiar with the purpose of this factory
reference signal.
In order to get correct absorbance, we suggest scanning new reference white before
sample when you start to use NIRScan every time.
There are two options for PGA setting in ADC:Auto or Fixed PGA Gain.
For auto PGA, the Auto box is ticked and software will assign the maximum PGA
value to the device according to the signal intensity automatically. The options of
PGA values are 1, 2, 4, 8, 16, 32 and 64.
For specified PGA value, software can set a PGA value to the device. The specified
PGA value cannot be larger than the PGA value obtained in the Auto PGA mode,
otherwise the signal will become saturation. For example, if the Auto PGA value is 16,
then you would not set PGA to 32.
If the reference signal is scanned at one PGA value (e.g. 64) but the sample signal is
scanned at another PGA value (e.g. 16), then the reference signal will be scaled
down (x 1/4) to meet the PGA value of sample signal. However, the actual reference
signal at PGA 16 is not equal to quarter of the reference signal at PGA 64, this will
induce absorbance error. To make an accurate measurement, the same PGA value
for both reference signal and sample signal is required. The relationship of PGA
setting and ADC reading is investigated in following report. The ADC input is supplied
23
with a constant voltage source. By changing the PGA setting and reading the ADC
data, a linear relationship between the actual ADC reading and PGA setting can be
calculated.
The constant voltage source is supplied for ADC reading to reach and close to the
maximum value at PGA=64. After that, the ADC readings of smaller PGA values are
captured to compare the linear regression result.
By comparing the PGA and ADC reading, the linear regression result of PGA setting
vs. ADC reading ratio is very close to each other but not equal to multiple. When
PGA is larger, the deviation of the ratio is more obvious.
As above, the ADC reading at different PGA setting does not follow the ratio of PGA
values. For example, the ADC reading at PGA32 is not the double of ADC reading at
PGA16.
When we assume the ADC reading at PGA32 is exactly equal to the double of ADC
reading at PGA16, we will see the absorbance inconsistence problem. You can
compare the two different procedures in figure. 2.11 and the results are shown in
figure 2.12. Procedure #A will ensure absorbance consistence at different PGA
setting while procedure #B will have consistent reference signal ratio but
inconsistent absorbance result.
24
Figure 2.11 Comparison of PGA setting for procedure #A vs. procedure #B
For example, we use NIR-S-G1 and scan SRS-99, the auto PGA is 64 for Column 1 and
the auto PGA is 16 for Hadamard 1(Figure 2.13)
25
Table 2.3 PGA Gain configuration: Column mode and Hadamard mode
Scan Config Name Column 1 Hadamard 1
Scan Type Column Hadamard
Spectral Range Start 900 900
Spectral Range End 1700 1700
Width (nm) 7.03 7.03
Exposure Time (ms) 0.635 0.635
Dig. Resolution 228 228
Num Scans to Avg 6 6
PGA Gain 64 16
Figure 2.13 The result of Auto PGA Gain : (a)Column mode, (b)Hadamard mode
PGA gain affects the signal intensity obviously, a larger PGA gain will increase higher
signal intensity, but the specified PGA gain can’t be set more than auto PGA gain.
Next, we will demonstrate the effect of different PGA options. To evaluate a new
sample, please tick “auto” first. The auto PGA gain is 16. After you are experienced
with your device or sample condition, you can choose Fixed PGA Gain. Fixed PGA
Gain has different gain settings including 1, 2, 4, 8, 16, 32, and 64. We select PGA
gain as 8、32 and 64. Now we can realize the PGA setting effects the absorbance
26
Table 2.4 PGA Gain configuration
(a) (b) (c) (d)
Scan Config Name Hadamard 1 Hadamard 1 Hadamard 1 Hadamard 1
Scan Type Hadamard Hadamard Column Hadamard
Spectral Range Start 900 900 900 900
Spectral Range End 1700 1700 1700 1700
Width (nm) 7.03 7.03 7.03 7.03
Exposure Time (ms) 0.635 0.635 0.635 0.635
Dig. Resolution 228 228 228 228
Num Scans to Avg 6 6 6 6
PGA Gain 16 8 32 64
Figure 2.14 PGA gain effect in Hadamard mode, (a) PGA Gain: Auto, (b) PGA
Gain: 8, (c)PGA Gain: 32, (d)PGA Gain: 64
As above, not only scan configuration but also PGA setting will affect the
measurement results (absorbance consistence). To achieve consistent result, please
fix scan configuration and PGA setting during your test. If you make any change on
either scan configuration or PGA setting, please redo reference signal scan before
sample signal scan.
27
2.3.3 Pattern Width and Number of Sampling Points
Except light source intensity, pattern width can be increased to enhance signal level.
You can increase pattern width to the maximum width which still produces good
enough spectrum resolution shown in figures 2.15. For example, a plastic sample is
scanned with different pattern width settings as follows. The AU still looks fine when
pattern width is not increased too much.
Number of sampling points which means the Dig. Resolution in GUI can be increased
to improve spectrum resolution. Number of measurement points can also be
28
reduced to save scanning time if resolution is still good enough. As example in figure.
2.17, 128pts and 160pts do not make much difference.
However, keep increasing the number of sampling points, for example, 32pts and
64pts will have distortion in AU displayed in figure. 2.18.
The number of sampling points could refer to the formula.
800
× 2 = Dig. _ resolution
Pattern _ width
29
2.3.4 Exposure Time
In general, traditional spectrometer may be used an analog charge integrator, which
integrates the charge from a photodiode with a capacitor before measuring the
voltage across it with an ADC. With this charge integration topology, the voltage
continually rises as the integration time increases.
In contrast, for DLP spectrometer, photo diodes operate by absorption of photons
and generate a flow of current in an external circuit, proportional to the incident
power. The incident power is proportional to its radiation frequency. The current
generated from the photo diode should not change when exposure time increases. If
we expect that the total number of photons will give a higher voltage signal. This is
only true if the analog front-end is accumulating that current in a capacitor. In our
implementation, we convert this current directly into a voltage and sample the
result multiple times during each DMD pattern.
Exposure time = DMD pattern display time.
Integration time = the time for analog front-end to accumulate the charge into a
capacitor.
With the NIRScan design, we use a transimpedance amplifier which converts the
current to a voltage without accumulating it. Therefore, increased exposure time
increases the number of samples that we collect and average, but does not increase
the voltage level at the ADC input.
30
In DLP spectrometer, which uses hardware trigger input mode of the DLPC150, so
each pattern stays displayed on the DMD for the entire exposure time selected. The
hardware trigger is generated from TIVA.
The increase in samples used is not linear because for each exposure we must wait
until the signal stabilizes before taking samples. This delay is fixed, so increasing the
exposure time adds samples at the ADC rate of 30kHz. The SNR will improve SQRT
(5.08/0.635) =2.8 when increasing exposure time from 0.635ms to 5.08ms.
In theory, Hadamard mode should still also benefit from increased exposure time.
However, it may be less noticeable if the SNR is already high. Depending on the
Hadamard scan configuration, it will likely increase the SNR by 2x-7x. Increasing the
exposure time will increase the SNR relative to the square root of the exposure time.
The difference between 0.635ms pattern exposure time Column scan and 0.635ms
pattern exposure time Hadamard scan may be similar to the difference between
0.635ms exposure time Column scan and 5.08ms exposure time Column scan. This
function may be cooperated with slew scan to enhance the spectra quality.
31
Figure 2.21 The effect of scan mode and expose time to SNR.
32
Figure 2.23 Slew Scan
33
Figure 2.24 Single InGaAs detector photosensitivity
34
Table 2.5 Scan configuration setting
Scan Config Name Hadamard 1
Scan Type Hadamard
Figure 2.25 The sample signal variation in continue scan from cold-start
By using the initial reference signal R as the reference signal and Sn (n=1 to 200) as
sample signal to calculate AU, the 1st AU and the 200th AU are shown below. The
200th AU not only has level shift but also has spectrum distortion.
35
Figure 2.26. AU(1st) vs. AU (200th)
The spectrum intensity at 1300nm during 200 scans is shown below. Obviously, it is
not proper to take R (reference signal at cold-start) as the reference signal for
calculating the AU for all the following 200 signals because the spectrum intensity
will change rapidly from cold-start. By using R (reference signal at cold start) as the
reference signal, the calculated AU is affected by the warm-up process obviously. AU
(S10 vs. R) shows nonlinear behavior shown in figure 2.28.
36
Figure 2.28 The absorbance is affected in warm up procedure.
If S100 is used as reference signal, the result will be much better. By using S100 as
reference signal, the AU deviation will be within 0.0005 for the following 100 scans.
As above, if both reference and sample scans are taken after enough warm-up time,
the calculated AU will be much stable. If device is operated randomly among
different operating points, then a frequent reference scan is required to get stable
AU.
37
Chapter 3 Operating ISC NIRScan
In this chapter we will describe how to use ISC NIRScan to get a good spectrum.
Upon launching the ISC_NIRScan_Win_WinForms_GUI, the software checks ISC
NIRScan through USB and you can start to perform scan operation.
For example, we use NIR-S-G1 to scan tablet and the operation as following.
1. Set Scan Configuration: you can adjust scan parameters as shown in figure 3.1.
38
Figure 3.2 Reference scan
3. Fixed PGA and scanned sample: we put tablet directly on scan window and fix
PGA Gain. After scanning tablet, we check spectrum and find some variation with
wavelength of 900 -1000 nm and 1650-1700 nm. This configuration is not suitable
for tablet and we adjust scan configuration again by slew scan as step 4.
4. Reset Scan Configuration: we use slew scan and adjust scan parameters to try to
find out an optimal scan configuration.
39
Figure 3.4 Adjust scan parameters
5. Checked spectrum: the spectrum is better than we scanned before, hence, we can
use this method to scan tablet.
40
3.2 Transmissive Module
For transmissive module, such as T1, the instruction of GUI is the same with diffuse
reflective module. In order to get good spectrum, please note the following
suggestions we mention:
1. The height of sample: The Z-dimension of cuvette holder is 4.75mm, please fill the
cuvette with enough sample to more than 10mm shown in Figure 3.6.
41
3. Scan type: with transmissive NIRScan, we do not suggest using “Hadamard mode”
because it is very easy to make reference signal saturated even the PGA=1. Once
signal is saturated, the absorbance calculation will never be correct.
4. The fit of cuvette: please use the Allen wrench to adjust the ball plunger screws
until the ball end contacts the cuvette and starts to compress, do not over tighten
the ball plunger screws.
42
3.3 Fiber Input Module
Fiber input module is the most flexible depending on the application fields you want
to measure. For example, a liquid measurement device is shown in figure 3.10, and
there are some suggestions as following:
1. Fiber: We suggest that you can use linear fiber bundle to couple incident light into
the slit. DLP spectrometers typically have taller entrance slits than array detector
systems because the DMD is taller than the array detector and this allows for
increased signal throughput. In order to take advantage of the DLP spectrometer’s
large etendue, a round-to-linear fiber bundle should be used. These optical fibers
have multiple cores arranged in a circular pattern at one end, and a linear
configuration at the opposite end. The round end can be utilized at the sample as in
typical spectroscopy illumination modules, while the linear end can be more
effectively coupled to the slit and you can refer to products from Thorlabs.
2. Connection: We provide SMA interface for you to connect fiber cable into the
fiber input module and the mark on fiber linear has to be parallel to the groove on
spectrometer like figure 3.11.
43
Figure 3.11 The illustration of connection for fiber and spectrometer
44
Chapter 4 Spectrum Analysis and Application
4.1 Spectrum
The wavelength of spectrum is transferred from DMD pixel, and the mapping
between each pixel and wavelength is determined by the pixel-to-wavelength
coefficients (a, b, and c). The wavelength of each pixel is calculated by λ=aP2+bP+c,
where λ is wavelength and P is DMD pixel location. A standard light source, AR-1 and
a standard absorbance material, RM-NIR are scanned for wavelength calibration.
There are 7 known peaks from AR-1 and 2 known peaks from RM-NIR, which will be
used to solve the equation of λ=aP2+bP+c. After wavelength calibration, if given x
(DMD pixel location), then y (wavelength point) can be determined.
45
Figure 4.2 Resampling preprocessing
In data analysis, it is very important to make the step of wavelength uniform. For
example, both spectra of SRM2036 are very similar collected by FOSS 6500 and
NIR-S-G1 in figure 4.3(a). We analyze data directly by first order derivative, and the
result indicates a significant difference in figure 4.3(c). Hence, the first step for the
spectrum collected from NIR-S-G1 is to do resampling, and the result shows a good
consistency in figure 4.3(b). In summary, we suggest that you have to take
resampling preprocessing first before analysis.
46
Figure 4.3 Spectrums comparison: (a)raw spectrum, (b) first derivative with
resampling preprocessing, (c) first derivative without resampling.
4.2 Resampling
An appropriate interpolation method is benefit for analysis. For example, we use
linear interpolation and cubic spline interpolation to treat the raw spectrum and the
result shows the cubic spline interpolation is better than linear interpolation in
figure 4.4.
47
Figure 4.4 Interpolation: (a) Linear interpolation (b) Cubic spline interpolation
4.3 Application
The advantage of NIR is that it can be used for rapid quantitative or identification
analysis to understand the properties of materials quickly. Here we show two
applications as follows.
48
(a). Spectrum collection: all spectrums of 80 anti-hyperlipidemic agents are
collected by NIR-S-G1 through the best scanning configuration.
(c). Model training and validation: All spectrums are divided into training set,
validation set and test set, and the LDA model is established through PCA analysis
and posterior probability.
49
Figure 4.7 PCA analysis
(d). Application: Finally, the model can be applied to App and identify drugs well.
50
collected from low temperature to normal temperature.
51
Figure 4.11 Model training and validation
52