1. INTRODUCTION:

Recently, a high luminance LED lighting has been introduced and it replaces existing indoor general lightings incandescent lamp and fluorescent lamps1. LED has merits of a low power consumption, eco-friendly, long life, and can reproduce a variety of color2. As a result, lightings that consider sensibility and health of human being such as sensible light and light therapy using LED lightings are produced3,4. However, with LED, wavelength shift at an around 5nm is generated due to temperature increases and aging5. Due to wavelength shift, sensible lightings and light therapy could induce vital rhythm disturbance and fatigue in the eyes for human being unlike of the beneficial purpose of LED6,7. Therefore, occupants need to check if the lighting environment is suitable for the space and activity by measuring the characteristics of the light using a measurement equipment for the light source.

For the measurement of the light source, there are methods using a tri-stimulus direct-reading light source colorimeter or spectrometer8. Tri-stimulus direct-reading light source colorimeter is a device which measures light source using an optical filter on which sensitivity similar with the color sensitivity of the human eyes is implemented. It has merits of simple structure and less volume, but cannot measure the intensity of light by the wavelength. Spectrometer is a device to measure the light by dividing the light source using a dispersive element and arrangement detector such as CCD and CMOS. It has a benefit of measuring the intensity of light by the wavelength (radiation energy), thus is being used in various fields such as materials analysis and efficiency analysis of photosynthesis9,10. However, its structure is complicated and has a large volume, it is not suitable to be used under a general environment or narrow laboratory11. In addition, integration time needs to be manually adjusted according to amount of light and dark current also should be removed as integration time is changed. In the spectrometer, integration time control and dark current removal are mandatory process to measure spectral radiation illuminance correctly12. The performance of spectrometer is degraded if integration time is short since it cannot absorb the light sufficiently. On the contrary, if integration time becomes longer, since a large amount of light is absorbed, it exceeds measurement allowance value, which directly causes data loss13. Dark current is a noise that is generated other than response to the light in the CMOS or CCD image sensor14. Such dark current is generated in proportion to the integration time, thus it should be removed according to changes of integration time in order to measure light source correctly15.

In this paper, we propose a portable spectrometer that has integration time control and dark current removal functions. The proposed spectrometer is constructed with a light measurement module, integration time automatic control module, dark current automatic removal module, and communication module. Light source measurement module is to read the light absorbed through spectral sensor as an analogue signal of 5V. That is, a raw spectrum for the current light source is measured. The integration time automatic control module is to make adequate amount of the light to be measured input into the spectral sensor. If intensity of the light is high, integration time is adjusted as short, while if intensity of the light is weak, the integration time is adjusted as long. The dark current automatic elimination module is to remove the dark current that is generated other than reaction to the light in the spectral sensor. Since dark current is generated in proportion to the integration time, it is removed after calculating dark current for the current integration time using a relation equation between dark current and integration time. The relational expression is generated using a regression analysis after obtaining large amount of dark current data by the integration time by experiment. Meanwhile, a communication module is to measure light source remotely and supports Bluetooth wireless communication between remote device and spectrometer. Performance test for the spectrometer was carried out with evaluation for the wavelength indicator scale using a mercury lamp and standard light source A and spectral radiance. Calibration and performance evaluation were also executed using CAS 140 CT and these were checked if these meet the performance requirement for spectrometer of KS.

2. DEVELOPMENT OF SPECTROMETER:

The proposed spectrometer was fabricated with the spectral sensor, Bluetooth, and MCU. C12880MA from ‘H’ company that can measure the light in around 340~800nm wavelength band was used as a spectral sensor. BT Shild and Arduino Uno were used as a Bluetooth and MCU, respectively. In addition, light source measurement module, integration time automatic control module, dark current automatic elimination module, and communication module were realized with a firmware and loaded on the MCU. Fig. 1 shows the schematic diagram of the proposed spectrometer.

SPD (288) in Fig. 3 shows the raw spectrum that was measured in advance. Satcurrent and Sattarget are the saturation and target saturation of SPD (288). ERROR is an error between current saturation and target saturation, while IT refers to an initial value of integration time. Integration time automatic control algorithm was implement to make operated as below. Firstly, the integration time is input and it is examined if the integration time is more than 11 and less than 1,000,000. If the integration time is out of the range, 11 or 1,000,000 is output and algorithm is completed. While, if the value is within the range, light source is measured with the current integration time. The saturation of spectral distribution that was obtained by measuring the light source was calculated to determine the error from the target saturation. If error is below 5%, output of current integration time is generated and algorithm is completed, while if the error is more than 5%, current saturation and target saturation are compared. If current saturation is larger, the integration time is reduced by 5%, while if the target saturation is larger, the integration time is increased by 5%, and algorithm is re-executed.

2.3. Dark Current Automatic Elimination Module

Dark current automatic elimination module is a module to remove the dark current which becomes different according to the setting of integration time. CCD and CMOS respond the light and electric charges are accumulated. The dark current generated during the process means an accumulation of electric charge with other reasons other than response towards the light. Sensor having a dark current zero is physically not existed and dark current should be minimized by designing or process¹⁶. Since dark current is generated in proportion to the integration time, it can be calculated using a correlation equation between dark current and integration time. The correlation equation is generated using the dark current data by integration time collected through experiment. Fig. 4 shows the graph representing the changes in the dark current by the integration time.

Figure 4. Dark current change curve by the integration time

Fig. 4 shows the curve for average value of dark current collected while changing the integration time as a total of 1,000 stages as 1,000, 2,000, 3,000, …, and 1,000,000 us after completely blocking the light so that the light won’t be input into the spectrometer. Here, x axis represents the integration time of the spectrometer and its unit is us(micro second). The intensity on the y axis is converted value of 0~5V analogue signal that is coming out from the spectral sensor to the digital value of 0~1023. This changes curve shows that integration time and dark current are proportional to each other. Furthermore, the correlation coefficient between the integration time and dark current was 0.995, indicating that the correlation between two data sets was very high. The regression analysis for two data sets was carried out and 1^st, 2^nd, 3^rd, 4^th, and 5^th degree regression equations were generated. Table 1 shows the residual and R-squared values for generated regression equation by each degree.

Table 1: Regression analysis between integration time and average dark current value

Regression Model	Residual					R-squared
Regression Model	Min	1Q	Median	3Q	Max	R-squared
First-order	-14.29	-8.99	-1.43	6.97	28.73	0.9901
Second-order	-9.33	-3.42	0.08	3.03	7.9	0.9986
Tertiary	-3.65	-1.24	0.03	1.22	4.92	0.9998
4th	-3.6	-1.3	0.04	1.26	4.6	0.9998
5th	-2.86	-1.04	-0.002	0.94	2.89	0.9998

Since 1^st and 2^nd degree regression equations generated large residuals and small R-squared values, compatibility of the regression equations became the worst. While, 3^rd, 4^th, and 5^th degree regression equations had the same R-squared value of 0.9998. However, the residual in the 5^th degree regression equation was smaller than those in the 3^rd and 4^th degree regression equations. Therefore, an equation to realize the dark current automatic elimination module was finally selected. The 5^th degree regression equation is as in Equation 1 and the coefficient of regression is as shown in Table 2.

(1)

Table 2: Regression equation per each order

Term	Symbol	Regression Coefficient
Constant term	C	1.092222e+02
First-order	D₁	1.844588e-04
Second-order	D₂	-9.724963e-11
Tertiary	D₃	1.198584e-15
4^th	D₄	-1.589876e-21
5^th	D₅	6.454925e-28

In the Equation 1, i is the integration time, while Darkavg(i) is the average value for the integration time. Dark current automatic elimination module calculates the average dark current for the respective integration time using the Darkavg function, and then removes the dark current from the current spectral distribution.

2.4. Communication Module:

A communication module was implemented by designing the packet for sending and receiving the Bluetooth wireless communication based data between the spectrometer and smartphone. The packet structure to send command of light source measurement from smartphone to the spectrometer is as in Table 3. Whilst, the packet structure to send the spectral distribution data from the spectrometer to the smartphone is as in Table 4. The communication module receives the packet of light source measurement request in Table 3 from the smartphone and then it examines if the packet is normal by using packet length, STX, CRC, and ETX. The spectral distribution obtained by measuring the light source is converted to the packet structure as in Table 4 and then it is sent to the smartphone.

Table 3: Packet structure for request of light source measurement

	STX	DEV_ID	CMD	DATA	CRC	ETX
Size (bytes)	1	1	1	4	1	1
Value	0x02	0x00~ 0xFF	‘M’	Integration Time	CheckSum	0x03

Table 4: Packet structure to send spectral distribution

	STX	DEV_ID	CMD	DATA	CRC	ETX
Size (bytes)	1	1	1	676	1	1
Value	0x02	0x00~ 0xFF	‘S’	Spectrum	CheckSum	0x03

The communication module receives the light source measurement request in Table 3 from the smartphone, and then it examines if the packet is normal by using the packet length, STX, CRC, and ETX. After that, the spectral distribution obtained by measuring the light source is converted to the packet structure as in Table 4, and then it is sent to the smartphone.

3. EXPERIMENT AND ANALYSIS:

3.1. Calibration of Wavelength Indication Scale and Performance Test:

Calibration of wavelength indication scale is a task to convert 288 numbers of pixel of spectral sensor into the wavelength (nm). This process requires a light source for calibration such as already known mercury and helium lamp. Mercury and helium lamps are suitable for calibrating wavelength indication scales in spectrometers because they emit a spectroscopic distribution of atomic intrinsic wavelengths with a size around 0.05 nm. Therefore, calibration and evaluation were carried out using mercury lamp and CAS 140 CT which have appropriately distributed wavelength. The CAS 140 CT instrument is a spectrometer that can be used as a calibration instrument. The mercury lamp used in the experiment has a total of eight reference wavelengths at 340 ~ 800nm. Lighting was turned on with a DC power source for accurate calibration and evaluation. Table 5 summarizes the eight reference wavelengths obtained using the CAS 140 CT instrument and the peak pixels obtained using the proposed spectrometer.

Table 5: Peak pixel and wavelength of each equipment

INDEX	Wavelength [nm] (CAS 140CT)	Pixel (Mini Spectrometer)
Peak 1	360.64	16
Peak 2	400.09	31
Peak 3	431.09	43
Peak 4	481.45	63
Peak 5	539.37	87
Peak 6	578.53	104
Peak 7	607.35	117
Peak 8	704.75	165

The reference wavelength of the Peak 1 is 360.64nm and the peak pixel is 16. This means that the analogue signal of no. 16 pixel in the proposed spectrometer is the intensity of light having wavelength 360.64nm. Equation 2 shows the relation of the number of the pixel of the spectral sensor with the respective wavelength and is generally generated using various regression methods¹⁷.

(2)

Where, λ is the wavelength, and p is the pixel number. C₀, C₁, C₂, and C₃ are the coefficient of polynomial to calculate the wavelength. It can be improved as the number of reference wavelength used in the calibration is increased. The wavelength indication scale as in Equation 2 was generated based in Table 5. The reference wavelength used in the calibration process is referred as a standard wavelength and the calculated wavelength by the Equation 2 is set as the direction wavelength, the error can be expressed as in Table 6.

Table 6: Evaluation result of wavelength indication scale

INDEX	Standard Wavelength [nm]	Direction Wavelength [nm]	Absolute Error [nm]
Peak 1	360.64	360.81	0.17024
Peak 2	400.09	400.06	0.028576
Peak 3	431.09	430.86	0.228378
Peak 4	481.45	480.9	0.550489
Peak 5	539.37	538.59	0.777544
Peak 6	578.53	577.75	0.77868
Peak 7	607.35	606.66	0.685174
Peak 8	704.75	704.85	0.09974

Comparison analysis for absolute error of the wavelength indication scale showed that it was the highest in the Peak 6 with 0.77868nm and the lowest at the Peak 8 with 0.099744nm. the average of the absolute error throughout entire reference wavelength was 0.414853nm, which was in conformity with the KS Performance requirement of the spectrometer (within 0.5nm).

3.2. Calibration and Evaluation of Spectral Irradiance

Calibration and spectral irradiance is a process to convert intensity coming out from the spectral sensor into irradiance [W/m²]. The standard light A was used to calibrate the spectral irradiance. Standard light source A is a light source having irradiation energy at all the wavelength and is suitable for calibration and evaluation of spectral irradiance. Therefore, calibration and evaluation were carried out using a 60W bulb-type standard light source A and CAS 140 CT. Since wavelength gaps between the proposed spectrometer and CAS 140 CT are different each other, wavelength interpolation has to be carried out. As a wavelength interpolation, Lagrangian interpolation and Spline interpolation were used⁷. Lagrangian interpolation was introduced to calibrate and evaluate the spectral irradiance by interpolating the wavelength band of 340~800nm with 1nm gap. Table 7 shows data for the standard light source A using the proposed spectrometer and CAS 140 CT.

Table 7: Data for the standard light source A by each measuring equipment

Wavelength [nm]	Intensity	Irradiance [W/m²]
340	50	5.85E-05
341	53	6.12E-05
342	57	6.34E-05
343	61	6.48E-05
···	···	···
797	340	0.005838
798	339	0.005867
799	335	0.005881
800	338	0.005867

Intensity is digital values of 0~1023 converted from the analogue signal from the spectral sensor. Irradiance is the spectral irradiance and represents the energy per unit area. The ratios between these two data are used as calibrated data for the spectral irradiance illumination. When the measurement data by CAS 140 CT for the standard light source is set as a standard irradiance and the calibrated data by the proposed spectrometer is set as a corrected irradiance, the error rate can be expressed as in Table 8.

Spectral irradiance was evaluated and the highest error 3.301% between the standard irradiance and corrected irradiance was found at 349nm, while the lowest error 0.0002% was obtained at 495nm. The average error was 0.35%, satisfying the performance requirement of KS (within 0.5%).

4. CONCLUSION:

In this paper, a portable spectrometer was developed in order to support measurement of lighting environment easily. Conventional mini-spectrometer have inconvenienced manual adjustment of the integration time according to the amount of light. We have proposed a portable spectrometer that realizes the integration time automatic control. The portable spectrometer was constructed with a light source measurement module, integration time automatic control module, dark current automatic elimination module, and communication module. The light measurement module to measure a raw spectrum for the light source was developed by implementing a spectral sensor (C12880MA). The absorbed light was processed with the input of 288Pixel, 5V analogue signal. For the Integration Time Automatic Control Module, an adequate amount of the light was to enter through the spectral sensor. In the method of integration time automatic control module, the target saturation is set for the incident amount of light and the integration time is adjusted according to the intensity of light. Furthermore, it was confirmed that the dark current was generated in proportion to the integration time. The dark current automatic elimination module was realized by obtaining sufficient dark current data as per the integration time. The relationship equation between the dark current and integration time was drawn using a regression analysis, and then dark current could be removed. The proposed spectrometer was then fabricated as a portable type so that users even under general environment and narrow laboratory can operate it easily. In addition, the remote control of the device was realized based on Bluetooth wireless communication.

We evaluation of wavelength indication scale and spectral irradiance was performed to investigate performance requirement of the proposed portable spectrometer. The performance of the portable spectrometer was compared with the reference device CAS-140CT. The errors of wavelength indication scale and spectral irradiance were displayed as 0.414853 nm and 0.35 %, respectively, indicating that these values were satisfactory with the performance requirement of the spectrometer as specified in KS. In the future study, it is necessary to develop a smartphone application that can easily check the characteristics of the light and confirm the adequacy of the current lighting environment for the general users.

5. ACKNOWLEDGMENT:

This work was supported by Institute for Information & communications Technology Promotion(IITP) grant funded by the Korea government(MSIP) No. R71181610160002003, Development of Interlocked Multi Wearable Sensor Smart Device and Service Platform for Effective Personal Training.

This research was supported by the Functional Districts of the Science Belt support program, Ministry of Science, ICT and Future Planning(2016K000298)

6. REFERENCES:

1. Byong-Moon Yang, Woo-jin Jang, and Jae-Sang Cha, Development of The LED FDL for Replacing Fluorescent Lamp, Journal of the Korean Institute of Illuminating and Electrical Istallation Engineers, 2015, 29 (1), pp. 9-14.

2. X Luo, R Hu, S Liu, and K Wang, Heat and fluid flow in high-power LED packaging and applications, Progress in Energy and Combustion Science, 2016, 56, pp. 1-32.

3. Ortiz-García-Cervigón, Vicente, et al., LED strips for color-and illumination-based emotion regulation at home. International Workshop on Ambient Assisted Living. Springer International Publishing, 2015, pp. 277-287.

4. Profita, Halley, Asta Roseway, and Mary Czerwinski, Lightwear: An Exploration in Wearable Light Therapy. Proceedings of the Ninth International Conference on Tangible, Embedded, and Embodied Interaction. ACM, 2015, pp. 321-328.

5. Tsai, Chun-Chin, et al., Decay mechanisms of radiation pattern and optical spectrum of high-power LED modules in aging test, IEEE Journal of Selected Topics in Quantum Electronics, 2009, 15 (4), pp. 1156-1162.

6. Brainard, George C., et al., Action spectrum for melatonin regulation in humans: evidence for a novel circadian photoreceptor, Journal of Neuroscience, 2001, 21 (16), pp. 6405-6412.

7. Glickman, Gena, et al., Light therapy for seasonal affective disorder with blue narrow-band light-emitting diodes (LEDs), Biological psychiatry, 2006, 59 (6), pp. 502-507.

8. Mothods of colour measurement for light sources (KS A 0068), Korean Standards Service Network, 2015.

9. Devereaux, Zachary J, et al., Matrix-Assisted Ionization on a Portable Mass Spectrometer: Analysis Directly from Biological and Synthetic Materials, Analytical chemistry, 2016, 88 (22), pp. 10831-10836.

10. Lashgari, Mohsen, Sanaz Soodi, and Parisa Zeinalkhani, Photocatalytic back-conversion of CO 2 into oxygenate fuels using an efficient ZnO/CuO/carbon nanotube solar-energy-material: Artificial photosynthesis, Journal of CO2 Utilization, 2017, 18, pp. 89-97.

11. Bacon, Christina P, Yvette Mattley, and Ronald DeFrece, Miniature spectroscopic instrumentation: applications to biology and chemistry, Review of Scientific instruments, 2004, 75 (1), pp. 1-16.

12. Andersen, Hans Villemoes, H Wedelsback, and Per Waaben Hansen, NIR spectrometer technology comparison. White Paper from FOSS, 2013, pp. 1-14.

13. Ren, Zhong, et al., Research of spectrophotometer for bio-chemical analyzer based on automatic adjustment of the integration time of linear CCD. Bioinformatics and Biomedical Engineering, 2008. ICBBE 2008. The 2nd International Conference on. IEEE, 2008, pp. 2164-2168.

14. Abbas El Gamal, Helmy Eltoukhy, CMOS Image Sensors. IEEE Cirsuits&Devices Magazine.

15. Burkart, Andreas, et, al., A novel UAV-based ultra-light weight spectrometer for field spectroscopy, IEEE sensors journal, 2014, 14 (1), pp. 62-67.

16. Porter, William C, et al., Dark current measurements in a CMOS imager. Proc. SPIE. 2008.

17. Mirapeix, J, et al., In-process automatic wavelength calibration for CCD-spectrometers. Photonics Europe. International Society for Optics and Photonics, 2008.

Received on 20.05.2018 Modified on 11.06.2018

Research J. Pharm. and Tech 2018; 11(10): 4619-4626.

DOI: 10.5958/0974-360X.2018.00845.4

Development of Portable Spectrometer supporting, Automatic Control of Integration Time

Figure 1. Schematic diagram of the spectrometer

Figure 2. Timing Chart for C12880MA

Figure 3. Flow chart of Integration time automatic control algorithm