Comprehensive Evaluation System for LED Driver Concerned of Visual Comfort

Special features and varied applications make complex claims for LED driver selection. The aim of study is to establish a comprehensive evaluation system for LED driver for selection under varied applications. This study first analyzes the importance of key elements for driver evaluation concerned of visiual comfort. Besides, research shows that the change of driver current, driver mode and dimming procedure will have impact on the LED SPD (Spectrum Power Distribution), color and visual comfort. Thus, visual comfort index including dimming Linearity, dimming stability and strobe is added to the evaluation system besides the traditional evaluation index of circuits. As the complex evaluation factors can not be described using one equation, we structure a comprehensive evaluation system containing these three factors system performance, driver performance and visual performance based on the Analytic Hierarchy Process (AHP) method. Finally, an application case is implemented and applied to clarify the proposed method.


INTRODUCTION
With the improvement of whole social living standardard and process of semiconductor technology, people are demanding more for the lamp of energy saving (Rubinstein, 2007;Aldrich et al., 2010), personalized illumination (Muthu et al., 2002) concerned about psychological mood regulation (Figueiro, 2010) and circadian rhythm, rather than just as extension of natural light. Some parameters for traditional circuit evaluation such as cost, power factor and harmonic content can be applied in the construction of evaluation system for LED driver. However, considering the special current-luminous flux characteristic of LED and complex visual applications, we should pay more attention for other influencing factor. For instance, the life difference between the driver and LED lamp will reduce the application value of LED as long life light source 5 , thus it is necessary to take the life and stability of driver into account to build an evaluation system. The impact on the visual comfort caused by driver is another problem that is easily overlooked.
Visual channel, as the main method for human to get information, has already been adapt to the sun's rays and developed a special spectrum sensitive characteristics called V(λ) after long evolutionary in nature, as Fig. 1 showed below.
People's visual perception is decided by the spectrum of light source, which is affected by the power of drive current. Therefore, the thesis will focused on the study of relation between LED driver and visual comfort and breaking the professional barriers between circuit and visual, then use it as the basis for buildingup LED driver evaluation system, As Fig. 2 showed. As a complicated system evaluative process, the evaluation of LED driver involve many factors, which are difficult to be expressed by unified equation. Thus, Analytic Hierarchy Process is chosen to help establish evaluating model as objective guidance for different drive application choice. As the complex evaluation factors can not be described using one equation, we structure a comprehensive evaluation system containing these three factors system performance, driver performance and visual performance based on the Analytic Hierarchy Process (AHP) method.

THE EFFECT OF DRIVER CIRCUIT ON THE VISUAL PERCEPTION
The role of lamp is to convert the electrical energy to the light energy, thus the change of light energy form is closely related to the input power. Considering the difference between LED and rational light source, this study will discuss the relation between drive and visual comfort from the following points: • The effect of light spectrum on the visual: Due to the differences between the artificial light spectrum and the daylight spectrum, human vision will be some inadequacies, such as information misreading and visual discomfort. To take the color rendering as example, which is fully decided by the spectrum power distribution of light, the lamp with continuous spectrum have better color rendering. The change of light spectrum will make an impact on the visual perception.
The study shows, the dominant wavelength will shift toward short wavelenghth with the increase current, due to band filling and QCSE dominated effects (Gu et al., 2006). At the same time, it takes very little shift in wavelength, in the range of 1 to 6 nm, before the color shift crosses the 3-step MacAdam ellipse boundary, as Fig. 3 showed.
LED has fast response times within nanosecond and is a kind of current driven device. With current control, the LED brightness L (cd/m 2 ) will be roughly linearly proportional to current IF: (1) Therefore, the subtle change of the drive current will cause the LED brightness change. In order to Fig. 3: Color change in CIE1931(x, y) coordinate system of PC white and mixed-color LED system under continues current and PWM dimming mode realize perfect luminous effect, it is necessary to control current within ideal range and set a evaluation indicator for the effect of driver current on stability of light wavelength at the same time.
• The effect of dimming process on vision: The LED is commonly used in different dimming occasions for its feature such as fast response and changing color, providing variety combinations of brightness and color.
As the nerve structure accepted stimulation is divided into individual units according function, eyes are very sensitive to subtle changes of light under low brightness, while difficult to detect greater change under lighter condition due to the saturation of the human visual system. It means that people are aware of stimulate when incremental amount is large enough to excited an additional neural quantum unit. So we can see that visual psychological feeling and the actual Control parameters of light source and practical brightness have the following relationship: When feeling evenly brightness changes from B 1 to B 2 , actual brightness changes from L 1 to target brightness L2 according to the formula 2. Thus we get the following result: In the process of light dimming, the drive control should meet the request of formula 4 in order to realize the light smooth transition.
Professor Jinno Ya.'s study turns out that when imposed 5% duty cycle, the cycle frequency of about 60Hz high-speed pulse voltage on LED, the experience brightness of human eye is about 2 times compared with DC voltage condition 11 . This illustrates further that LED driven by the same way will have different performance while dimming.
• The effect of driver waveforms on visual perception: With special current-voltage driving characteristics, LED driver need proper design and topology selection to maximum the advantage of LED. According to the final output drive current forms, different driving scheme is shown as Table 1.
Due to different driving waveform, LED takes different time to keep lighten each cycle and produces different heat to change junction temperature, which cause changes caused band widen, wavelength shifted then color changed. According to the study of Marc Dyble, the chromaticity shifts of LED system with the PWM performing are slightly better than the continuous current dimming scheme 9 . The driving waveform is concerned of driving topology; however the change of control information will have an impact on the driving waveform in the same topology, as Fig. 5 showed.
To sum up, the capture of human eye for visual information is closely related with light source spectral distribution. Driving current, driving topology and dimming process will all change the spectral distribution and color of LED, even because the visual discomfort. These various connections between driving waveform and the visual comfort should be taken into account of LED driver evaluation system. Specific index will be given in the next chapter.
Establish evaluation index about the driver's effect on visual perception: Set up the index as follows according to the discussion above: linear correlation of dimming current: The change light output will lag behind the control parameter and increase the instability of dimming system, when the linear relation is poor. So the index is definite as followed: X represents the control signal of dimming system; Y represents the average output in one cycle. The linear correlation is bigger when |r| is bigger. | r | = 1 means completely linear correlation. Dimming stability: Although LED has the advantages of quick response, it also becomes a disadvantage for causing fluctuation when brightness changes' fast, as Fig. 6 showed.
There is a kind of psychophysics of vision that the intermittent light stimulus appears to be completely steady to the observer, as long as modulation frequency is above the flicker fusion threshold (or flicker fusion rate), besides the perceived intensity can be changed by changing the relative periods of light and darkness. However, the light above the flicker fusion threshold can also cause visual fatigue, headache and other symptoms. It depends on the fluctuant depth and frequency. As Daniele Gallo's study shows, the perception of light varies with the frequency and presents maximum spectral response between (Fig. 7). Fluctuation depth is used as an index to represent the depth of light disturbance. As formula 6 shows, Emax, Emin, Eav express the maximum, minimum and average change values in one dimming cycle. The proportion of the light output can be equal with the proportion of the current for LED brightness is proportional to the current: Color stability: Luminous characters and even the color of LED will change with the current fluctuation for band filling effect. The PWM dimming schemes performs slightly better than the continuous current dimming scheme, as the output of light keeps still when lighting on under PWM schemes. Besides, Fig. 8 and 9 shows that dominate wavelength will decrease, related color temperature increase linearly with the current increase. Therefore, the absolute value of the current changes during conduction is used to measure the color stability of driver. Taking buck, half-bridge and SCR controller topology as example, compare of different topology by evaluation index is shown as Table 2.

ESTABLISHMENT OF EVALUATION MODEL AND CASE ANALYSIS
Establishment of comprehensive evaluation system and relative weight: The objective of the synthetic assessment system is to assess the best LED drivers for specific occasions. This study builds an evaluation system which is based on the match degree between driver and lamp, performance of the driver and visual comfort. Determination matrix can be generated based on the evaluation model using 1-9 scale method and finally get the weight of each level for the overall target LEDs can be used in many occasions and each factor has different significance in different occasion to be allocated different weight. In this study , the example is about traditional lighting system of a studio replaced by LED lighting system. The flicker of the lighting sources used in videos will cause pictures fuzzy. The tint of the lighting sources should be as steady as possible. So, lighting sources used in videos has high requirements in visual comfort. Fig. 10 is the evaluation model and the weight of different factors.
Example introduction: Two dimming methods were used in this study. In method A, LM3445 was used which is compatible with traditional SCR dimming system to drive the LEDs. The driver turns TRIAC signal into PWM control signal whose duty cycle matches the conduction angle of SCR. The PWM control signal was then used to control the BUCK circuit in order to generate specific current. So, finally the relationship between the output voltage of the driver and the conduction angle is linear. And the value of the conduction angle is controlled by the RC circuit. The SCR turns on, only when the voltage of the capacitor reaches a fixed value. As we can see, the relationship between specific conduction time and adjustable resistor is linear. That means the output voltage is linear to the adjustable resistor. Light intensity of the LEDs is in proportion to the drive current, but the relationship between drive current and voltage is nonlinear. So, the linear correlation of the driver is unacceptable. The rated current of the designed driver is 400 mA and the fluctuation is usually 15%~30% of the rated current.   The frequency of the fluctuation is 250 kHz. So, the maximum value of the fluctuation is 400×30% = 120 mA. This fluctuation will cause the color to be unstable. The driver used in method B is designed based on a half-bridge resonant converter. So, the output luminous flux is totally linear to the duty cycle of the control signal. The current is constant when the LED is on, so the color is stable.

Comparison of different driving modes:
The parameter of two kinds of driver is listed in Table 3 as follows: Due to the different properties and dimensiol, quantitative index and qualitative indexes is hard to express in unified digital quantity. This evaluation system brings in the Fuzzy theory to dismiss the effect of different index on the judging result. The judging indexs are divided in five levels to get fuzzy score. According to the above model, the indicators set is V = {1, 3, 5, 7 and 9}, while comments set is V = {Extremely Low, Low, Medium, High, Extremely High}.
Evaluation score of each index is listed By the biggest membership principle, PWM dimming method is proper for the studio lighting considering driver, system performance and visual comfort, although LM3445 has the advantage of cost and the PWM dimming is too complex to design.