Multi-Color Phosphor-Converted Wide Spectrum LED Light Source for Simultaneous Illumination and Visible Light Communication

Abstract

Simultaneous illumination and communication using solid-state lighting devices like white light-emitting diode (LED) light sources is gaining popularity. The white light LED comprises a single-colored yellow phosphor excited by the blue LED chip. Therefore, color-quality determining parameters like color-rendering index (CRI), correlated color temperature (CCT), and CIE 1931 chromaticity coordinates of generic white LED sources are poor. This article presents the development of multi-color phosphors excited by a blue LED to improve light quality and bandwidth. A multi-layer stacking of phosphor layers excited by a blue LED led to the quenching of photoluminescence (PL) and showed limited bandwidth. To solve this problem, a lens-free, electrically powered, broadband white light source is designed by mounting multi-color phosphor LEDs in a co-planar ring-topology. The CRI, CCT, and CIE 1931 chromaticity coordinates of the designed lamp (DL) were found to be 90, 5114 K, and (0.33, 0.33), respectively, which is a good quality lamp for indoor lighting. CRI of DL was found to be 16% better than that of white LED (WL). Assessment of visible light communications (VLC) feasibility using the DL includes time interval error (TIE) of data pattern or jitter analysis, eye diagram, signal-to-noise ratio (SNR), fast Fourier transform (FFT), and power spectral density (PSD). DL transmits binary data stream faster than WL due to a reduction in rise time and total jitter by 31% and 39%, respectively. The autocorrelation function displayed a narrow temporal pulse for DL. The DL is beneficial for providing high-quality illumination indoors while minimizing PL quenching. Additionally, it is suitable for indoor VLC applications.

1. Introduction

Solid-state lighting devices are used massively in our day-to-day lives because of their low power consumption, high efficiency, long lifetime (>50,000 h), compactness, low cost, and being environmentally friendly, i.e., mercury-free light [1,2,3,4,5,6]. In addition, solid-state lighting devices are also explored for indoor visible light communications (VLC) due to the advantages like high speed, high security, low cost, low energy consumption, no electromagnetic interference, health safety, and license-free bandwidth [7,8,9,10,11,12].

A generic white light source is composed of a blue light-emitting diode (LED) chip exciting the Ce³⁺ doped yellow phosphor, which absorbs a part of blue spectrum light and converts it into yellow light. The output light is a mixture of unabsorbed blue and broadband yellow light, which the human eye perceives as a nearly white light. A white light source for illumination should have a color-rendering index (CRI), correlated color temperature (CCT), and CIE 1931 chromaticity coordinates at >95, 4000–6500 K, and (0.33, 0.33), respectively. The CRI classifies the white color of light, and it can be increased by broadening the wavelength spectrum [13,14,15,16,17]. The quality of light defined by the properties above is the most comfortable for human vision, the work of ciliary muscles, and autonomic nervous functioning, eventually increasing human working efficiency [18]. Recent attempts to improve LED lamp optical qualities have used red, green, and yellow phosphors pumped by blue light. Zhu et al. demonstrated wavelength division multiplexing and VLC using multi-color LEDs up to 13 m [19]. However, multiple scattering and reabsorption cause thermal quenching of emitted light, resulting in poor illumination. Another method is mixing phosphors in glass, UV adhesive, or epoxy resin by spin blade, inkjet printing, and spray coating [20]. Still, in this case, the luminance efficacy of radiation and CCT is reduced significantly, and color parameters are still not up to the mark. Vitasek et al. demonstrated how phosphor powders affected the rise and fall edges, i.e., pulse extension in the time domain. However, the petri-dish-sized phosphor layer consumes a lot of phosphor [21,22]. Jargus et al. illustrated the effect of these phosphor layer thickness and concentration on CRI, CCT, and temporal response of pulse [23]. Furthermore, most LED light sources used daily have poor color characteristics.

Duty cycle distortion affects chromaticity, affecting color stability for human vision during simultaneous communication. A communication channel noise, fading, path loss, pulse spreading, and time delay at the receiver result in signal distortion. Consequently, inter-symbol interference (ISI) can occur, causing the perception of binary one as zero and vice versa [24]. The ISI restricts the number of users who may access the Wi-Fi network due to limited bandwidth. Zhang et al. [25] presented Na₂YMg₂V₃O₁₂:Sm³⁺ phosphor as a versatile element for solid-state illumination, demonstrating a CRI of 84.9, a CCT of 6637 K, and a CIE (0.314, 0.373). The study conducted by Yu et al. [26] demonstrates that multi-carbon dots can be used for illumination, with CRI, CCT, and CIE as 86.9, 5388 K, and (0.3341, 0.3075), respectively. Ying et al. [27] demonstrated the characteristics of two phosphors, namely, YAG and nitride phosphors, with CRI, CCT, and CIE as 85, 2500 K, and (0.4, 0.4), respectively. Most of those mentioned above reported multi-color phosphor lamps still could not achieve the required parameters, i.e., CRI, CCT, CIE to be >95, 4000–6500 K, and (0.33, 0.33), respectively. This paper presents a hybrid luminaire that combines lighting and communication capabilities. It achieves this by utilizing multi-color phosphors stimulated by a blue LED to meet the requirements of CRI, CCT, and CIE parameters. The designed lamp (DL) demonstrates substantial improvement in CCT (5114 K), CIE (0.33, 0.33), and CRI (90). For VLC, the evaluation of data jitter or time interval error (TIE), eye diagram, signal-to-noise ratio (SNR), and power spectral density (PSD) is presented. Due to a 31% reduction in rise time, DL transmits binary data stream faster than white LED (WL). Furthermore, the autocorrelation function displayed a narrow temporal pulse for DL. Therefore, the DL is beneficial for providing high-quality indoor illumination while minimizing photoluminescence (PL) quenching and VLC applications.

2. Theoretical Background

The color and optical characteristics of the light source are described by tristimulus values X, Y, and Z, given by Equation (1).

$$\begin{array}{c}\mathrm{X}={\mathrm{K}}_{\mathrm{m}}{\int }_{380}^{780}\mathrm{S}\left(\mathsf{\lambda }\right)\overline{\mathrm{x}}\left(\mathsf{\lambda }\right)\mathrm{d\lambda }\\ \mathrm{Y}={\mathrm{K}}_{\mathrm{m}}{\int }_{380}^{780}\mathrm{S}\left(\mathsf{\lambda }\right)\overline{\mathrm{y}}\left(\mathsf{\lambda }\right)\mathrm{d\lambda }\\ \mathrm{Z}={\mathrm{K}}_{\mathrm{m}}{\int }_{380}^{780}\mathrm{S}\left(\mathsf{\lambda }\right)\overline{\mathrm{z}}\left(\mathsf{\lambda }\right)\mathrm{d\lambda }\end{array}$$

(1)

where K_m = 683 lm/W at λ = 555 nm; S(λ) is the relative spectral power distribution, and $\overline{\mathrm{x}}$, $\overline{\mathrm{y}}$, and $\overline{\mathrm{z}}$ are the color matching functions, which are the average data of the CIE 1931 standard [28,29] observer. From these chromaticity coordinates, the color temperature can be given by the following expression: a physical quantity to define the chromaticity quality of the source [30].

$$\mathrm{C}\mathrm{C}\mathrm{T}=-437{\mathrm{n}}^3+3601{\mathrm{n}}^2-6861\mathrm{n}+5514$$

(2)

where $\mathrm{n}={\displaystyle \frac{\mathrm{x}-0.3320}{\mathrm{y}-0.1858}}$, x, and y are the CIE 1931 color coordinates obtained by normalizing tristimulus values. CRI describes the color fidelity of the light source, which helps identify minute color variations. The average CRI of the light source for color samples 1 to 8 is denoted by R_a.

$$R_a={\displaystyle \frac{1}{8}}\sum _{i=1}^8{R}_i$$

(3)

where

$$R_i=100-4.6\Delta E_i$$

(4)

∆E_i is the color difference between the sample under the reference and test sources, and $R_i$ is the special CRI. The parameters mentioned above from Equation (1) to Equation (4) are essential for characterizing the quality of the white light source. The VLC performance of DL, jitter, eye diagram, SNR, PSD, and coherence bandwidth is presented in this section. OOK (On–Off Keying) modulation is widespread because it does not require receiver phase synchronization, so a square-law-based device can be utilized for direct detection [31,32]. The intersymbol interference (ISI) is introduced when data are transmitted at symbol rates approaching the Nyquist rate [33,34]. According to Equation (5), σ_t is the RMS delay spread, which is the minimal interval between symbol transmissions to prevent interference from multipath propagation [35].

$${\mathsf{\sigma }}_{\mathrm{t}}=\sqrt{{\overline{\mathsf{\tau }}}^2-{\left({\mathsf{\tau }}_{\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{n}}\right)}^2}$$

(5)

where

$${\overline{\mathsf{\tau }}}^2={\displaystyle \frac{\sum _{\mathrm{k}}\mathrm{P}\left({\mathsf{\tau }}_{\mathrm{k}}\right){{\mathsf{\tau }}_{\mathrm{k}}}^2}{\sum \mathrm{P}\left({\mathsf{\tau }}_{\mathrm{k}}\right)}}, {\mathsf{\tau }}_{\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{n}}={\displaystyle \frac{\sum _{\mathrm{k}}\mathrm{P}\left({\mathsf{\tau }}_{\mathrm{k}}\right).{\mathsf{\tau }}_{\mathrm{k}}}{\sum \mathrm{P}\left({\mathsf{\tau }}_{\mathrm{k}}\right)}}$$

${\mathsf{\tau }}_{\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{n}}$ is the mean excess delay, i.e., the first moment of the power profile, and P is the power received with corresponding time delay $\mathsf{\tau }$. RMS delay spread estimates coherence bandwidth B_c, mathematically expressed using Equation (6), where k can vary depending on the power delay profile, i.e., 50 when the correlation between frequencies is greater than 90% and 5 when greater than 50%. Coherence bandwidth determines channel dynamics such as flat or frequency-selective fading [35,36].

$${\mathrm{B}}_{\mathrm{c}}={\displaystyle \frac{1}{\mathrm{k}·{\mathsf{\sigma }}_{\mathrm{t}}}}$$

(6)

The central limit theorem states that adding multiple independent random variables produces a Gaussian profile [37]. However, other noise sources, such as thermal, shot, and flicker noise, affect the Gaussian response [38]. Flicker noise is the “trap and release” theory in which impurities, defects, and spaces in phosphor octahedral and tetrahedral sites trap charge carriers [39]. The colored channels induce crosstalk at the receiver, quantified by jitter in the time domain. Clock time interval error (TIE) is the digital signal reference-to-received edge time interval. This error occurs between all binary bits of signal (2¹¹ − 1 bits of random ones and zeros). The accumulated error over these bits is called periodic or phase jitter and quantifies phase measurement accuracy. In general, total jitter (TJ) present in the signal is the convolution of random jitter (RJ) and deterministic jitter (DJ) [40]. The RJ component is inherently present due to shot and thermal noise and cannot be ignored. The DJ can be removed using software-based algorithms. The probability density function of RJ follows the Gaussian distribution given by Equation (7), where $\sigma$ denotes the standard deviation of normal Gaussian distribution, τ is the jitter. Random jitter is characterized by its root-mean-square value ($\sigma )$ or standard deviation. The DJ is introduced by ambient interfering sinusoidal signal [40]. Time deviations depending on the used data pattern (e.g., PRBS) are regarded as one of the types of DJ results in duty cycle distortion arising from the rising and falling edge timing differences or single-ended system ground shifts, and inter-symbol interference (ISI), arising from limitations of component and system bandwidth. DJ is measured in terms of peak-to-peak values since it has been defined to have upper and lower bounds [41,42].

$$\mathrm{R}\mathrm{J}\left(\mathsf{\tau }\right)={\displaystyle \frac{1}{\mathsf{\sigma }\sqrt{2\mathsf{\pi }}}}\mathrm{e}\mathrm{x}\mathrm{p}\left(-{\displaystyle \frac{{\mathsf{\tau }}^2}{2{\mathsf{\sigma }}^2}}\right)$$

(7)

The TJ is mathematically represented in Equation (8):

$$\mathrm{T}\mathrm{J}\left(\mathsf{\tau }\right)=\mathrm{R}\mathrm{J}\left(\mathsf{\tau }\right)\ast {\displaystyle \frac{{\mathrm{A}}_{\mathrm{L}}\mathsf{\delta }\left(\mathsf{\tau }-{\mathsf{\mu }}_{\mathrm{L}}\right)+{\mathrm{A}}_{\mathrm{R}}\mathsf{\delta }\left(\mathsf{\tau }-{\mathsf{\mu }}_{\mathrm{R}}\right)}{{\mathrm{A}}_{\mathrm{L}}+{\mathrm{A}}_{\mathrm{R}}}}$$

(8)

A_L and A_R designate the amplitudes of the left and right DJ components, and µ_L and µ_R denote the mean of left and right jitter, respectively. Dirac-delta function is indicated by δ. Equation (9) represents the TJ for a bit error rate of 10⁻¹². DJ is often measured in peak-to-peak values [40,42,43,44].

$$\mathrm{T}\mathrm{J}=\mathrm{R}\mathrm{J}+\mathrm{D}\mathrm{J}=14\mathsf{\sigma }+\mathrm{D}\mathrm{J}$$

(9)

The tolerance of two signals to additive white Gaussian noise can be compared when signals have the same energy per bit, E, but the noise level can differ for the exact value of E or the same average power level.

Error probability for binary unipolar signal elements is given by Equation (10) [45].

$$P_e=Q\sqrt{{\displaystyle \frac{E}{N_o}}}$$

(10)

Q is the quality factor, and No/2 is the magnitude of the power spectral density of Gaussian white noise. The idea from cellular systems defined in [37] can be used for n number of sources here, seven LEDs, and Equation (10) can be modified to Equation (11). The bigger the number of interferers, the lower the error probability, as a Gaussian interferer is better than a single interferer.

$$P_e=Q\sqrt{{\displaystyle \frac{E}{N_o+n\left(E/N_o\right)}}}$$

(11)

The theoretical bit rate of LED is estimated by [46].

$$bit\_rate= {\displaystyle \frac{1}{Rise time+Fall Time}}$$

(12)

Signal rise time is the time it takes the leading edge to reach 10% to 90% amplitude. Similarly, fall time is the trailing edge’s 90% to 10% change. The human detection thresholds for flickering of light sources at different sinusoidal frequencies are defined in CIE recommendations [47].

3. Experimental Details and Results for Phosphor-Converted White Light Systems

The procurement of commercial phosphor powders of average diameter of 15 microns was made from Zhuhai Hanbo Co., Ltd. (Zhuhai, China) The red phosphor (RP) SrAlSiN₃: Eu⁺² has a quantum efficiency of approximately 82.3% [48]; the green phosphor (GP) SrBaSiO₄: Eu⁺² has a quantum efficiency of 88.6% [49], and the yellow phosphor (YP) Y₃Al₅O₁₂: Ce⁺³ has a quantum efficiency greater than 85% [50]. The initial step involved combining YP with transparent optical adhesive (NOA61-MIL-A-3920 (Norland Products Inc., South Brunswick Township, NJ 08831, USA)) in a proportion of approximately 7:2. Subsequently, this mixture was applied onto a transparent glass slide measuring 2.5 cm × 2.5 cm using the screen-printing process. The slide was subjected to UV irradiation for 10 min to remove moisture. Similarly, the two remaining layers were built and activated for a few seconds remotely using a 6 W blue LED placed 1 cm apart. Moreover, these slides were stacked together to form a three-layer combination and then activated remotely using the same source. A portable spectroradiometer (Lisun-6000 (LISUN ELECTRONICS (SHANGHAI) CO., Ltd., Shanghai, China)) measured the CRI, CCT, and CIE 1931 parameters. The integration time for the measurements was set to 1 s.

Figure 1a shows the fitted spectrum of single phosphor-coated slides with GP, RP, and YP. It was observed that most of the down-converted photons are surrounded near the peak-emission wavelength of 521 nm, 621 nm, and 552 nm, corresponding to the activator cations of GP, RP, and YP, respectively. Figure 1b shows the spectrum of a three-layer combination of these slides stacked together and excited with the same source again. A large band of down-converted frequencies is observed when GP is placed first at the optical axis, followed by YP and RP. However, the spectrum is not flat compared to sunlight (visible region) due to thermal and PL quenching in multi-layered structures. Figure 1d,e shows CRI, CCT, and CIE coordinates (in parentheses) associated with Figure 1a,b, respectively. A lamp is made to add seven LEDs using a reflector and diffuser assembly to create a broadband spectrum to overcome thermal quenching and limited optical bandwidth. The diffuser reduces optical power by 30% at a distance of 30 cm. These seven LEDs of six colors (green used twice) are mounted equidistantly on a heat sink in a side-by-side (co-planar) ring structure. Figure 1c compares the spectrum of DL to conventional WL, and Figure 1f shows the CRI, CCT, and angular uniformity of DL.

LEDs are of 2B3C-3W, surface mount device type, with 300 lumens with forward current of 300 mA, having dimensions 12 mm × 12 mm, tolerance of ±0.2 mm, max power dissipation of 4560 mW, forward voltage in the range 9–11.4 V, and maximum reverse current of 20 µA with half angle 70 degrees. All the properties were reported in the datasheet at 25 degrees temperature.

Figure 2a shows the PL spectrum of six LEDs with a peak wavelength of blue (451 nm), pink (612 nm), yellow (601 nm), white, red (632 nm), and two green (523 nm) lights. Two green LEDs were used to improve CIE color coordinates. Figure 2b compares the difference in R1-R15 color patches responsible for CRI and special CRI. Figure 2c(i) shows the lighting photograph of conventional WL; Figure 2c(ii) shows WL with reflector and diffuser. Figure 2c(iii) shows seven LEDs used to design the lamp; Figure 2c(iv) shows the lamp with reflector and diffuser assembly at 500 lux.

Figure 3 compares the optical properties of DL in contrast with conventional WL over an angular span of 120 degrees, such as CRI in Figure 3a, with an average value of 90 and 77.7 for DL and WL, respectively. Figure 3b shows CCT with an average value of 5114 K and 4930 K for DL and WL, respectively, and Figure 3c shows CIE 1931 coordinates of DL with an average value of (0.337, 0.337) and (0.35, 0.37) for DL and WL, respectively. It is important to note that DL shows a 16% increase in CRI and sunlike coordinates, which are rounded to two decimal places (0.33, 0.33). These results are recorded by rotating the spectroradiometer with the help of a rotating table. Figure 4 compares the image of the test object illuminated by the proposed light source and conventional white LED. The DL shows better color-rendering ability due to a 16% increased CRI.

Figure 5a shows the digital pulse of 1 s received at the photodiode and each LED’s amplitude vs. time response. The SNR depends on the responsivity of the silicon-based photodetector (shown in Figure 6), phosphor concentration, beam divergence, and channel response of these LEDs [51,52,53]. Figure 5b shows the PSD of individual LEDs to check other peaks or ripples in the 0–200 Hz band. Only the power supply frequency at 50 Hz (and its harmonic at 150 Hz) is evident with a power of nearly −30 dBm. According to the time-bandwidth trade-off [54], the broadband spectrum shows a narrow response in the temporal domain, i.e., short pulse width. The frequency response of ‘All LEDs’ and WL are nearly equal. The silicon-based photodetector depicted in (Figure 6) exhibits higher responsivity for red wavelengths (609–619 nm) relative to other predominant wavelengths emitted by our lamp.

Figure 7 shows the electroluminescence (EL) spectra of (a) WL and (b) DL by varying currents from 10 to 55 mA. The optical bandwidth of DL is increased with the wide range of down-converted wavelengths in 400–700 nm. Also, parallel combinations of LEDs require lower input voltage for the same amount of current. Consequently, the product of voltage and current, i.e., input power for driving DL, is lower than that of WL when achieving the same current at the photodiode. The DL operates at 120 mA (60 lumens) with an input power of 1.2 Watts, while the WL works at 90 mA (~1 Watts) to achieve a 5 µV (rms) current at the photodiode. The lumens can be increased accordingly by increasing the input power. The signal propagates through ambient air and is collected onto the photodiode by a lens with a focal length of 2.5 cm. A spectroradiometer positioned at the receiver plane (aligned in the line of sight without a lens) measures 21 Lux when integrating lighting and communication simultaneously. Figure 7a demonstrates that the spectral peak at 450 nm undergoes a blueshift of 1 nm, while the peak at 556 nm remains unchanged as the current increases from 10 to 55 mA. In Figure 7b, 450 nm shows a blueshift of 1 nm; the peak at 524 nm remains constant, and 609 nm is redshifted to 619 nm. The full-width half maxima of the WL and DL are 142 nm and 162 nm, respectively. Consequently, the spectral bandwidth is expanded by 14%. These spectral changes may occur due to increased applied electrical current leading to increased temperature. Hence, the refractive index and scattering property will change.

4. Experimental Details and Results for VLC Using Designed Lamp

The DL was simultaneously used for illumination and data transmitter through a digital switching network. From Figure 8a, LED was modulated with a MOSFET (IRF 540) for sending a 2-kbps pseudo-random bit sequence (PRBS) signal, 5 V (pp) with an offset of 2.5 V, length 2¹¹ − 1 (11 is the length of shift register), and rise time of 12 ns, generated by a function generator (Rigol DG952, 250 MSa/s). The PRBS signal is a non-return-to-zero (NRZ), an on-off-keying signal used to modulate the light. It is a bitstream mapped onto a time scale that serves as input to the base of MOSFET. At the collector terminal of MOSFET, the constant current power supply is connected to (Lisun, DC3005S) for biasing. A source terminal of MOSFET is kept at ground potential. Power supply stability is tested by gated-FFT of its voltage level, which shows a 48 MHz ripple frequency that cannot interfere with the transmitted signal. The FFT of background intensity on the photodiode exhibited a peak at 104 Hz with a power level of −28 dBm and another at 50 Hz with a power level of −36 dBm. The Telefunken S-153 P photodiode collects light from ambient air. It has a radiant sensitive area of 7.5 mm², diode capacitance of 75 pF, half angle sensitivity of 55°, rise time of 100 ns, spectral bandwidth of 600–1050 nm, responsivity of 0.6 A/W, and noise equivalent power of 1.4 × 10⁻¹⁴ W/√Hz. A trans-impedance amplifier converts photocurrent to electrical voltage and is displayed on a real-time oscilloscope.

In Figure 8a, the left side shows the schematic arrangement of multi-color LED(s) with reflector cup and acrylic diffuser. LEDs were fixed on an aluminum metal-based heat sink. According to the datasheet [56], five-watt LEDs require a heat sink with a minimum surface area of 10,300 mm². From this data, we estimated that one three-watt LED would require a heat sink of size 6180 mm^2, and seven such LEDs would require a minimum surface area of 43,260 mm². Hence, the LEDs were mounted on the heat sink with a surface area of 60,000 mm², which is the optimum requirement. The right side of this figure shows the schematic diagram representing the p-i-n photodiode and its corresponding readout circuits. It shows the circuit diagram of the receiver designed using MCP606, low power, unipolar power supply of 5 V, rail to rail, phase retaining op-amp. The feedback resistor sets the required gain by considering the maximum reverse current of the photodiode. The desired output swing from 0 to 5 volts is achieved by the feedback resistor (R), which is computed using Ohm’s law by a ratio of voltage and photocurrent, i.e., 5–0 V/5 µA gives 1 MΩ. Theoretically, the feedback capacitor (C) limits the signal bandwidth from 0 Hz to 80 kHz (to capture the PRBS sequence and its harmonics with clock cycle 2 kHz) by C = ½ × pi × R × f. By experimental validation, a small feedback capacitor of 2 pF was added to reduce ringing oscillations caused by poles created due to output load [57].

Figure 9 shows the time delay measured using an oscilloscope between the reference PRBS signal and the received signal as the power is increased for blue light and the coherent bandwidth of the system. The absolute power at the receiver plane was recorded by a power meter by Thor Labs (PM100USB). Using Equations (5) and (6), the delay spread (rms) of 5.97 µs, mean delay spread of 70 µs, and coherence bandwidth, B_c was found to be 33.5 kHz. The channel can be assumed to be flat-fading since B_c is significantly larger than the input signal bandwidth of 4 KHz.

Duty cycle distortion like ISI can be caused by reflections or bandwidth limitations of the channel or receiver [58]. The DL is driven at 120 mA and WL at 90 mA to obtain 5 µV current (rms) at the photodiode. Figure 10 shows the eye diagram and jitter of the received PRBS signals using WL and DL as transmitters while maintaining the same photocurrent output from the photodiode (5 µA). The root-mean-square current was measured using a Fluke-115 multi-meter (Fluke Corporation PO Box 9090, Everett, WA 98206, USA). The eye diagram was constructed using a real-time (EXR054A, 500 MHz, 16 Gs/s, Keysight Technologies, Penang, Malasiya) oscilloscope at the sampling rate of 1 GSa/s at a memory depth of 10 million points. Here, note that an explicit clock signal of 1 kHz (Agilent 33521A Function Generator) is used. Figure 10a,b depicts the superimposed tracing of 1000-bit intervals. The eye pattern of the WL signal is noisy because of the edges’ slow rising time. Conversely, the fast rise time of DL gives more noise margin (eye-opening), which may contribute to a lower ISI.

Jitter strongly affects link reliability by deviating edges due to signal crosstalk frequencies, making clock recovery difficult [40,58,59]. The jitter analysis of TIE is carried out in this experiment at the rising and falling edge of the signal to check the flickering of a light source. Figure 10c shows that TJ is the convolution of RJ and DJ. From the signal itself, the clock is recovered using a software-based clock recovery method of Keysight (DSO60002A, 1 GHz, 20 GSa/s), which aligns the received PRBS data edges with the virtual clock edges. From Figure 10d,e, peak-to-peak (pp) jitter is less for DL (41 µs) than that of WL (68 µs). This jitter (pp) represents the deterministic component (DJ) present in the signal, which is some ambient modulation frequency or interference present in the signal. This DJ component can be used to guess the transmitted pattern. The RJ (considering the left peak only) is 1.1 µs for DL, which was higher than that of WL (708 ns) because seven LEDs are modulated together; hence, random noise might have increased due to color crosstalk. From Equation (9), we calculated a TJ of 56 µs for DL and 78 µs for WL. Notably, the total jitter is 39% reduced using DL as an information source with simultaneous illumination.

Table 1. Experimental parameters of the received signal were obtained from its eye diagram, jitter or TIE histogram, and FFT.

Evaluation Parameters	White LED	Designed Lamp
3 dB bandwidth of a signal	987 Hz	994.6 Hz
Rise Time (avg)	41 µs	28 µs
Q factor	55	60
Extinction Ratio	22.64 dB	23.3 dB
SNR	64.05 dB	61.8 dB
Deterministic Jitter (DJ, peak to peak)	68 µs	41 µs
Random Jitter (RJ)	708 ns	1.1 µs
Total Jitter (TJ)	78 µs	56 µs

indicates signal parameters such as extinction ratio, SNR, rise time, 3-dB bandwidth, DJ, RJ, TJ, and Q (quality) factor. The 3 dB bandwidth is 1/√2 times the maximum amplitude of the PRBS test signal without the dc component. Notably, WL has a mean rise time of 41 µs, while DL has 28 µs. Hence, DL reduced the rise time by 31%. Based on these findings, we can say that bandwidth efficiency or theoretical bit rate (from Equation (12)) is higher for DL as rise time is inversely related to bandwidth. A higher extinction ratio also signifies a lesser bit error rate.

The 3 dB bandwidth of the test signal on an oscilloscope screen was captured at a sampling rate of 200.0 kSa/s and a memory depth of 60 k points. The maximum magnitude for both signals was 40 dBm at dc-frequency. Typically, the 3-dB bandwidth of white LEDs is in the MHz range. The 3-dB bandwidth for the WL was 984 Hz, while for the DL, it was 994 Hz, as indicated in Table 1. There is a slight increase in bandwidth for (DL) compared to (WL), possibly due to either the color crosstalk in DL or increased Q-factor. Bandwidth typically increases with current density [60,61,62]; however, LEDs experience efficiency droop [63]. LEDs’ −3 dB modulation bandwidth rises with higher current density due to a reduction in carrier lifetime. Conversely, the increase in −3 dB modulation bandwidth decreases with rising current density due to a slower rate of carrier concentration growth and enhanced electron leakage [64]. Figure 11 shows MATLAB (R2020b) plotting the PSD from 10 M data samples. The frequency domain representation of a PRBS test signal is a line spectrum bounded by a sinc function whose initial null point equals its bit rate. The baseband spectrum is symmetrical about dc-frequency, and the null-to-null bandwidth is twice the bit rate [51]. Figure 11 shows that a 2-kbps data transfer rate corresponds to the first frequency spectrum null at 2 kHz and a 4 kHz null-to-null bandwidth. Figure 12 shows the signal autocorrelation of the received signal to confirm DL bandwidth increment.

We have fabricated the light source using multi-colored LEDs to improve CRI (90), CCT (5114 K), and CIE 1931 coordinates (0.33, 0.33) of conventional white light. Keeping the illuminance conditions the same, nearly 21 Lux for these sources were used. Signal integrity is analyzed using an eye diagram, jitter analysis, and PSD (FFT and autocorrelation).

5. Discussion

From Grassmann’s second law of additive mixing, two green LEDs are placed to improve the CRI property of DL. Still, the eye perceives identical colors due to metamerism. Green color enhances the CRI as human scotopic vision is defined for predominantly green regions. Figure 1a shows that most down-converted photons of individual slides are surrounded near the peak-emission wavelength of their respective activator cations. However, in multi-layered structures, the spectrum in Figure 1b is not flat compared to sunlight (visible regions) due to thermal and PL quenching. A designed lamp attempts to solve the problem of PL quenching and limited optical bandwidth is fabricated to perform spectral addition of seven LEDs using reflector and diffuser assembly, yielding a resultant broadband spectrum whose average CRI, CCT, and CIE are 90, 5114 K, and (0.33, 0.33). The DL shows 16% increased CRI with 120 degrees of angular uniformity than conventional WL (CRI 78 in this experiment) and near-perfect sunlike coordinates, which are rounded to two decimal places (0.33, 0.33), as shown in Figure 3a. Light mixing inside light reflectors can be assumed to follow the central limit theorem, which says that if many independent random processes with arbitrary probability distribution functions are added, the sum approaches Gaussian distribution. This lamp achieved a maximum CRI of 95.5 in the 1–2 A range of current, but from an eye safety and thermal safety points of view, we chose to record all the data at low current. However, in our present system, we have used multiple blue chip LEDs (surface mount device, SMD), with a color phosphor mounted on top of each one. Since we have mounted these chips in the side-by-side planar configuration, there are negligible chances of reabsorption. Absorption of phosphors requires a wavelength of 450 nm or below. The downconverted set of wavelengths from each LED possess fewer residual blue (pump) photons; hence, we can say that neighboring phosphors are less likely to re-absorb.

Four methods are shown to analyze signal integrity: first, the signal–eye diagram, jitter analysis, and PSD (FFT and autocorrelation function). From Equation (11), it is clear that the more LEDs there are, the more interference there will be. Still, Gaussian noise (from DL) is better than a single noise source (WL), as the error probability of unipolar NRZ signal is inversely proportional to noise. Figure 11 shows that a data transfer rate of 2 kbps corresponds to the first null of frequency spectrum at 2 kHz and a null-to-null bandwidth of 4 kHz. In Figure 12, we observed the autocorrelation of the received PRBS signal to confirm the bandwidth increment using DL.

The authors introduced a high CRI light source for visible light communications in the proposed work. The slew rate transforms amplitude error to phase error (jitter). The designed lamp shows a fast communication response and enhanced color quality parameters. It is worth noting that DL is more suitable for low-loss transmission of a test signal than WL. We compared the communication characteristics without using a spectrum analyzer and an arbitrary waveform generator. Jitter analysis is imperative from a cyber-security point of view as a deterministic jitter component can be used to guess the transmitted pattern.

A 31% decrease in rise time was observed for DL, as shown in Table 1. The total jitter content of DL is 39% less than that of WL as an information source, but the RJ component is high as we have modulated seven LEDs together; hence, color crosstalk may occur. It is important to note that the Q-factor increases when DL transmits a signal. Crosstalk can be mitigated by positioning free-form optics before the detector [65]. Alternatively, one may opt for a laser-based solid-state lighting device wherein ballistic blue photons emitted by the laser facilitate detection for communication in the line of sight to the detector. We can say that data traveling at a higher speed induce more jittering (beam wandering) of the edges of the digital signal. It is tested up to 10 kbps speed by cascading two high-speed integrated circuits TL07 as a comparator for signal conditioning. A proof of concept is demonstrated up to a limited distance with a significantly lower driving current of 120 mA. However, DL can be driven up to 2.1 A, and a higher-power lamp can be developed. A large aperture lens can be used before the detector to focus light to increase the distance. Hence, designing an anti-jitter circuit is also suggested for future VLC work. This study finds the importance of color quality of light for human vision and visible light communication. Encouraged by these findings, future work will include eye-safe testing of the designed lamp from a photometric point of view with increased distance, including deep red phosphor and quantum dots, to enhance the color quality and signal properties.

6. Conclusions

In this paper, we have demonstrated simultaneous illumination and indoor visible light communication using a multi-colored lamp. To avoid the quenching effect, we have designed a multi-color LED lamp with a side-by-side arrangement of LEDs, leading to minimum quenching. Experimental results show that the average CRI, CCT, and CIE values are 90, 5114 K, and (0.33, 0.33), respectively, making it a good quality indoor lighting lamp and 16% better CRI than conventional LED.

For VLC, we obtained jitter, eye diagram, SNR, PSD, and autocorrelation. The DL shows a fast rise and fall times, more bandwidth with a good Q factor, SNR, and bit error ratio below the forward error correction limit of the order of 10⁻¹². In summary, DL shows a faster communication response than conventional WL due to a 31% decrease in rise time and a 39% decrease in total jitter. Also, the autocorrelation function demonstrated a fast temporal pulse for DL. This paper finds future applications in mining optical bandwidth with spectral modulation schemes like multi-channel frequency coding and color shift keying.