Method and Numerical Tool for Optimizing Light Emitting Diode Systems

Abstract

Methods and numerical tools for designing and optimizing LED systems are provided to achieve a desired luminous performance and to increase reliability and operating lifetime. In addition, methods for designing LED illumination systems are also disclosed to determine an optimum operating power for a desired output luminous flux, given the condition of the rated power of the LED and the heatsink. By the invention, LED illumination systems can be designed with a suitable choice of LED and/or heatsink.

Description

FIELD OF THE INVENTION

This invention relates to methods and numerical tools for forming, and optimizing the performance of, light emitting diode (LED) systems having single or multiple LEDs so that the luminous flux output from a given thermal design of the LED systems is maximized.

BACKGROUND OF THE INVENTION

A light emitting diode (LED) is a semiconductor device that emits light when a current is passed through it in a forward direction. Many types of LEDs are known that emit light in various wavelengths including infra-red, visible and ultra-violet regions. Many applications for LEDs are known including as indicator lights of various colours, for use in advertising displays, and in video displays.

In the past LEDs have tended to be lower power devices that produce relatively low power outputs and have not been used for general illumination purposes. More recently, however, high-power LED devices have become known that can provide an alternative to incandescent and fluorescent light sources. LED devices produce more light per watt than incandescent light sources and may therefore be useful as energy efficient light sources, while they have a number of advantages over fluorescent light sources including being easier to dim and not requiring the use of potentially toxic and polluting elements such as mercury to create the plasma that is the source of fluorescent light.

Light emitting diodes (LEDs) have therefore emerged as promising lighting devices for the future. However, LEDs are still primarily restricted to decorative, display and signaling applications so far and have not yet entered the market for general illumination to any great extent.

In photometry, one important factor that is commonly used for comparing different lighting devices is the luminous efficacy (lumen per Watt). One major hindrance to the widespread use of LEDs in general illumination applications is that the luminous flux of LEDs decreases with the junction temperature of the LEDs. The luminous efficacy of various LEDs typically decreases by approximately 0.2% to 1% per degree Celsius rise in temperature. Due to the ageing effect, the actual degradation of luminous efficacy could be higher than this quoted figures. Accelerated aging tests show that the light output can drop by a further 45%. For aged LEDs, the efficacy degradation rate could be up to 1% per ° C. In some applications such as automobile headlights and compact lamps, the ambient temperature could be very high and the size of the heatsink is limited. The drop in luminous efficacy due to thermal problem would be serious, resulting in reduction of luminous output.

In a recent article on lighting titled “Let there be light” (Harris, Mark, IET Engineering & Technology (E&T) Magazine, Vol. 4, Issue 20, 21 Nov.-4 Dec. 2009, pp. 18-21), it was commented that, in respect of LED products, “the majority of LED A-type replacement lamps do not meet manufacturer performance claims” and that “testing reveals that these lamps produce only 10 to 60 percent of their claimed light output”. Besides the quality issues, one possible reason for such mismatch in the claimed and actual luminous performance is the understanding of the luminous efficacy figures of LED devices. LED device manufacturers usually cite high luminous efficacy figures which are only correct at a junction temperature of 25° C. In practice, luminous efficacy will decrease significantly with increasing LED junction temperature. At a junction temperature under normal operation, it is not unusual that the luminous efficacy could drop by 25% or more.

FIG. 1 shows a conventional LED. At the heart of the LED device is a light emitting semiconductor material such as InGaN though other materials will be known to those skilled in the art. In the example of FIG. 1 a light-emitting InGaN chip 1 is mounted on a silicon substrate 2 and is connected to electrodes such as cathode 3 through gold wires 4 and solder connection 5. The light-emitting chip 1 is covered by a silicone encapsulant 6 and a plastic lens 7.

When a LED of the type shown in FIG. 1 is used to generate light a substantial amount of heat is generated that will damage the light-emitting chip if not removed. Therefore a heat sink must be provided and beneath the light-emitting chip 1 is a heatsink slug 2. In practice when used to provide a source of light for illumination, conventionally multiple LEDs are provided to form a LED system as shown in FIG. 2 where multiple LEDs 10 are provided on a single heatsink 11.

SUMMARY OF THE INVENTION

The present invention provides, in a first aspect, a method of forming or optimizing an LED system to achieve a desired luminous performance over a time period, the LED system comprising a single or a plurality of LEDs on a heatsink with each LED having a junction, the method comprising the steps of: (a) calculating a predicted luminous performance over the time period with a computer-implemented dynamic model; and (b) modifying the LED system such that the predicted luminous performance substantially matches the desired luminous performance.

In a second aspect, the present invention provides a method of optimizing an LED system comprising a single or a plurality of LEDs on a heatsink with each LED having a junction, the method comprising the step of calculating junction temperature over a time period with a computer-implemented dynamic model, thereby allowing modification of the LED system if a calculated peak junction temperature exceeds a maximum rated junction temperature, and thereby optimizing the LED system.

In both the first and second aspects above, the computer-implemented dynamic model, in one embodiment, is in the form of a software tool on a computer accessible through an internet website. In another embodiment, the computer-implemented dynamic model is in the form of a software tool stored on a portable computer-readable medium.

The present invention also provides, in a third aspect, a computer-implemented numerical tool that calculates a predicted luminous performance over a time period for an LED system comprising a single or a plurality of LEDs on a heatsink with each LED having a junction, thereby allowing modification of the LED system such that the predicted luminous performance substantially matches a desired luminous performance in order to form or optimize the LED system to achieve the desired luminous performance.

In a fourth aspect, the present invention provides a computer-implemented numerical tool for use with an LED system comprising a single or a plurality of LEDs on a heatsink with each LED having a junction, the computer-implemented numerical tool calculating junction temperature over a time period, thereby allowing modification of the LED system if a calculated peak junction temperature exceeds a maximum rated junction temperature, and thereby optimizing the LED system.

In both the third and fourth aspects above, the computer-implemented numerical tool, in one embodiment, is in the form of a software tool on a computer accessible through an internet website. In another embodiment, the computer-implemented numerical tool is in the form of a software tool stored on a portable computer-readable medium.

Preferably, in both the first and third aspects above, luminous performance is one or more of luminous flux of the LED system, luminous efficacy of the LED system, heatsink temperature and junction temperature over the time period.

In a fifth aspect, the present invention provides a method of forming an LED illumination system comprising a single or a plurality of LEDs on a heatsink with a desired output flux, comprising the steps of: (a) modeling on a computer the luminous flux emitted by said LED system as a function of the thermal resistance of said heatsink and the power applied to each LED, and (b) selecting an LED system such that the maximum luminous flux is emitted at a power equal to or below a rated power of said LED system provided that said maximum luminous flux is equal to or greater than the desired output flux, or (c) selecting an LED such that the rated power of said LED system is below the power at which the maximum luminous flux is emitted, provided that the flux emitted by said LED system at said rated power is equal to or greater than the desired output flux.

Preferably, in option (c) the rated power is at between 80% and 96% of the power at which maximum flux would be output.

In a sixth aspect, the present invention provides a method of forming an LED illumination system comprising a single or a plurality of LEDs on a heatsink with a desired output flux, comprising the steps of: (a) modeling on a computer the luminous flux emitted by said LED system as a function of the thermal resistance of said heatsink and the power applied to each LED, and (b) selecting a heatsink having a thermal resistance such that the maximum luminous flux is emitted at a power equal to or below a rated power of said LEDs, or (c) selecting a heatsink having a thermal resistance such that the rated power of said LED system is below the power at which the maximum luminous flux is emitted, provided that the flux emitted by said LED system at said rated power is equal to or greater than the desired output flux.

Preferably in step (c) the rated power is at between 80% and 96% of the power at which maximum flux would be output.

In a seventh aspect, the present invention provides an LED illumination system comprising a plurality of LEDs on a heatsink, wherein said heatsink has a thermal resistance such that the maximum luminous flux is emitted at a power below a rated power of said LEDs.

The present invention also provides, in an eighth aspect, a method of forming an LED illumination system comprising a plurality of LEDs on a heatsink with a desired output flux, comprising the steps of: (a) selecting an LED system such that the maximum luminous flux is emitted at a power below a rated power of said LED system provided that said maximum luminous flux is equal to or greater than the desired output flux, or (b) selecting an LED such that the rated power of said LED system is below the power at which the maximum luminous flux is emitted, provided that the flux emitted by said LED system at said rated power is equal to or greater than the desired output flux.

In a ninth aspect, the present invention provides a method of forming an LED illumination system comprising a plurality of LEDs on a heatsink with a desired output flux, comprising the steps of: (a) selecting a heatsink having a thermal resistance such that the maximum luminous flux is emitted at a power below a rated power of said LEDs, or (b) selecting a heatsink having a thermal resistance such that the rated power of said LED system is below the power at which the maximum luminous flux is emitted, provided that the flux emitted by said LED system at said rated power is equal to or greater than the desired output flux.

BRIEF DESCRIPTION OF THE DRAWINGS

Some embodiments of the present invention will now be described by way of example and with reference to the accompanying drawings, in which:—

FIG. 1 shows schematically the structure of a conventional LED;

FIG. 2 shows a conventional system with multiple individual LEDs mounted on a single heatsink;

FIG. 3 is a plot of the typical relationship between efficacy of LEDs versus junction temperature;

FIG. 4 is a plot of the typical relationship between heatsink temperature and power dissipation;

FIGS. 5( a) and (b) show (a) simplified dynamic thermal equivalent circuit of N LEDs mounted on the same heatsink, and (b) a simplified steady-state thermal equivalent circuit with N LEDs mounted on the same heatsink;

FIG. 6 shows an assumed linear function of junction-to-case thermal resistance R_jc;

FIG. 7 shows calculated and measured total luminous flux versus lamp power for eight 3W LEDs mounted on a heatsink with thermal resistance of 6.3° C./W;

FIG. 8 shows calculated and measured total luminous efficacy versus lamp power for eight 3W LEDs mounted on a heatsink with thermal resistance of 6.3° C./W;

FIG. 9 shows calculated and measured total luminous flux versus lamp power for eight 3W LEDs mounted on a heatsink with thermal resistance of 4.5° C./W;

FIG. 10 shows calculated and measured total luminous efficacy versus lamp power for eight 3W LEDs mounted on a heatsink with thermal resistance of 4.5° C./W;

FIG. 11 shows calculated and measured total luminous flux versus lamp power for eight 3W LEDs mounted on heatsink with thermal resistance of 2.2° C./W;

FIG. 12 shows calculated and measured total luminous efficacy versus lamp power for eight 3W LEDs mounted on a heatsink with thermal resistance of 2.2° C./W;

FIG. 13 shows calculated and measured total luminous flux versus lamp power for two 5W LEDs mounted on heatsink with thermal resistance of 10° C./W;

FIG. 14 shows calculated and measured total luminous efficacy versus lamp power for two 5W LEDs mounted on a heatsink with thermal resistance of 10° C./W;

FIG. 15 shows calculated and measured total luminous flux versus lamp power for two 5W LEDs mounted on heatsink with thermal resistance of 6.8° C./W;

FIG. 16 shows calculated and measured total luminous efficacy versus lamp power for two 5W LEDs mounted on a heatsink with thermal resistance of 6.8° C./W;

FIG. 17 plots total luminous flux emitted as a function of LED power;

FIG. 18 illustrates in more detail one embodiment of the invention where an LED is operated below the peak value of emitted luminous flux;

FIG. 19 illustrates the operating principles in the context of forced cooling;

FIG. 20 shows an apparatus that may implement embodiments of the invention;

FIG. 21 shows a flowchart illustrating the use of an embodiment of the invention;

FIG. 22 shows a further flowchart illustrating the use of an embodiment of the invention;

FIG. 23 shows another simplified thermal equivalent circuit of N LEDs mounted on a heatsink;

FIG. 24 shows a flowchart illustrating the use of another embodiment of the invention;

FIG. 25 shows a flowchart illustrating the use of a further embodiment of the invention;

FIG. 26 shows a chart of the measured luminous flux for an LED system having eight LEDs mounted on a heatsink with thermal resistance of 1.8° C./W;

FIG. 27 shows a chart of the calculated luminous flux for the LED system referred to in FIG. 26, the luminous flux calculated in accordance with an embodiment of the invention;

FIG. 28 shows a chart of the measured luminous flux for an LED system having eight LEDs mounted on a heatsink with thermal resistance of 2.9° C./W;

FIG. 29 shows a chart of the calculated luminous flux for the LED system referred to in FIG. 28, the luminous flux calculated in accordance with an embodiment of the invention;

FIG. 30 shows a chart depicting the variation with time of the measured and calculated luminous flux for the two LED systems referred to in FIGS. 26 and 28;

FIG. 31 shows a chart of the junction temperature over time of an offline LED mounted on a heatsink with thermal resistance of 4.1° C./W, the junction temperature calculated in accordance with an embodiment of the invention; and

FIG. 32 shows a chart of the junction temperature over time of an offline LED mounted on a heatsink with thermal resistance of 2.2° C./W, the junction temperature calculated in accordance with an embodiment of the invention.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

To increase the light emitted from a conventional LED system it is necessary to increase the current applied to the LED. Continuing to increase the LED power will have diminishing returns, however, as the increasing temperature of the LED will reduce its efficiency and potentially damage the LED. The heatsink is therefore important as it is essential for heat to be carried away from the LED so as not to cause it any damage. The light emitted by a LED will increase with applied current provided that the heat produced can be removed, but for any individual LED design there will come a point where increasing power applied to the LED will not result in greater light being emitted because heat is no longer being effectively removed. Identifying the relationship between power applied and light emitted is an important aspect of the present invention.

The following is a photo-electro-thermal (PET) theory that characterizes this relationship.

Let φ_ν be the total luminous flux of an LED system consisting of N LED devices.

φ_ν=N×E×P_d  (1)

where E is efficacy (lumen/Watt) and P_d is the real power of one LED (W)

It is well known that the efficacy (E) of LEDs will decrease with increasing junction temperature of the LEDs. FIG. 3 shows a typical relationship provided by an LED manufacturer. It can be seen that:

E=E _o└1+k_e(T_j−T_o)┘ for T_j≧T_o and E≧0  (2)

where E, is the rated efficacy at the rated temperature T_o (typically 25° C.) and k_e is the relative reduction of efficacy with increase in temperature. For example, if E reduces by 20% over a temperature increase of 100° C., then k_e=0.002.

In general, the LED power can be defined as P_d=V_d×I_d, where V_d and I_d are the diode voltage and current respectively. But only part of the power will be dissipated as heat. Thus, the heat generated in one LED is defined as:

P _heat =k _h P _d =k _h V _d I _d  (3)

where k_h is a constant less than 1.

Now consider a typical relationship of the heatsink temperature and the heat generated in the LED system as shown in FIG. 4. A simplified dynamic thermal equivalent circuit of a LED system is shown in FIG. 5(a), assuming that (i) the N LEDs are placed on the same heatsink with thermal resistance of R_hs, (ii) each LED has a junction to case thermal resistance R_jc and (iii) a thermal conductor with electrical isolation (such as heatsink compound) is used to isolate the LEDs from the heatsink and which has a thermal resistance of R_ins. A distributed thermal model is used for the heatsink due to its relatively large size. The corresponding thermal capacitances are needed if dynamic response is to be studied.

Under steady-state conditions, the thermal model can be further simplified into a steady-state model as shown in FIG. 5(b). In practice, a heat sink compound or equivalent may be used between the LEDs and the heat sink to ensure good thermal contact. The thermal resistance of such thermal conductor/electric insulator is relatively small when compared with R_jc of LEDs and is neglected in FIG. 5(b) and the following equations.

Based on the model in FIG. 5(b), the steady-state heatsink temperature can be expressed as:

T _hs =T _a +R _hs(NP_heat)=′T_a+R_hs(Nk_hP_d)  (4)

where T_a=ambient temperature.

From FIG. 5(b) and (4), the junction of each LED is therefore:

T _j =T _hs +R _jc(P_heat)=T_hs+R_jc(k_hP_d)  (5a)

T _j =T _a+(R_jc+NR_hs)k_hP_d  (5b)

Now, T_j obtained in (5) can be used in (2):

E=E _o└1+k_e(T_j−T_o)┘

E=E _o{1+k_e└T_a+(R_jc+NR_hs)k_hP_d−T_o┘}

E=E _o└1+k_e(T_a−T_o)+k_ek_h(R_jc+NR_hs)P_d┘  (6)

So, the total luminous flux φ_ν is:

φ_νNEP_d

φ=N{E_o└1+k_e(T_a−T_o)+k_ek_h(R_jc+NR_hs)P_d┘}P_d

φ_ν=NE_o{[1+k_e(T_a−T_o)]P_d+k_ek_h(R_jc+NR_hs)P_d²}  (7a)

Equation (7a) can also be expressed as follows:

φ_ν=NE_o{P_d+[k_e(T_a−T_o)]P_d+k_ek_h(R_jc+NR_hs)P_d²}  (7b)

Several important observations can be made from equations (7a) and (7b).

• • 1. Equation (7) relates the luminous flux (φ_ν) to the electrical power of the LED (P_d) and the thermal resistance of the heatsink (R_hs) and the LED junction (R_jc) together. It is an equation that integrates the photometric, electrical and thermal aspects of the LED system together.
  • 2. For a given heatsink (that may be restricted in size by a specific application), the operating point P_d* at which maximum φ_ν occurs can be determined. Alternatively where there is flexibility in designing the heatsink, the equations can be used for thermal design to optimize the size of the heatsink (R_hs) for a given LED array.
  • 3. Because k_e is negative and less than 1, (7) is in the form of φ_ν=α₁P_d−α₂P_d² where a₁ and a₂ are two positive coefficients. As P_d is increased from zero, φ_ν increases almost linearly because the second term is negligible when P_d is small. As P_d increases, the second negative term which is proportional to the square of P_d will reduce φ_ν significantly. After reaching the maximum point, the φ_ν will drop faster as P_d and R_jc increase (due to the increasing significance of the negative terms in (7b)). This means that the parabola of φ_ν is not symmetrical. Since the luminous flux function is a parabola and therefore has a maximum value, this maximum point can be obtained from

$\frac{{\phi }_v}{P_d}=0.$

By differentiating (7) with respect to P_d,

$\begin{array}{cc}\frac{{\phi }_v}{P_d}={\mathrm{NE}}_o\left\{\begin{array}{c}\left[1+k_e\left(T_a-T_o\right)\right]+2k_ek_h\left(R_{\mathrm{jc}}+{\mathrm{NR}}_{\mathrm{hs}}\right)P_d+\\ \left[\left(T_a-T_o\right)P_d+k_h\left(R_{\mathrm{jc}}+{\mathrm{NR}}_{\mathrm{hs}}\right)P_d^2\right]\frac{k_e}{P_d}+\\ \left[k_e\left(R_{\mathrm{jc}}+{\mathrm{NR}}_{\mathrm{hs}}\right)P_d^2\right]\frac{k_h}{P_d}+\left(k_ek_hP_d^2\right)\frac{R_{\mathrm{jc}}}{P_d}\end{array}\right\}& \left(8\right)\end{array}$

It should be noted that the first two terms on the right hand side of (8) do not have derivatives, while the remaining three terms do. Strictly speaking, k_e, k_h and R_jc are not constant. It must be noted that R_jc will indeed increase significantly with lamp power.

The above equations can usefully be simplified for practical applications. As a first approximation, it is assumed that k_e, k_h, and R_jc are constant for the time being. It is known that k_h will reduce slightly for a few percent under dimming conditions. From LED manufacturer data sheets the degradation of the efficacy with junction temperature is usually assumed to be linear and thus k, is assumed to be constant. This assumption is acceptable for k_e and k_h, and will be relaxed to accommodate the changing nature of R_jc in the analysis later. Based on this assumption, (8) can be simplified as:

$\begin{array}{cc}\frac{{\phi }_v}{P_d}={\mathrm{NE}}_o\left\{\left[1+k_e\left(T_a-T_o\right)\right]+2\left[k_ek_h\left(R_{\mathrm{jc}}+{\mathrm{NR}}_{\mathrm{hs}}\right)\right]P_d\right\}& \left(9\right)\end{array}$

Therefore, maximum-φ_ν point can be obtained by putting

$\begin{array}{cc}\frac{{\phi }_v}{P_d}=0\mathrm{and}P_d^{*}=-\frac{\left[1+k_e\left(T_a-T_o\right)\right]}{2k_ek_h\left(R_{\mathrm{jc}}+{\mathrm{NR}}_{\mathrm{hs}}\right)}& \left(10\right)\end{array}$

where P_d* is the LED power at which maximum φ_ν occurs. (Note that k_e is a negative value.)

From (3), the corresponding LED current at which maximum φ_ν occurs can be obtained as:

$\begin{array}{cc}{I}_d^{*}=-\frac{\left[1+k_e\left(T_a-T_o\right)\right]}{2k_ek_h\left(R_{\mathrm{jc}}+{\mathrm{NR}}_{\mathrm{hs}}\right)V_d}& \left(11\right)\end{array}$

Several significant observations can be made from (10) and (11).

• • 1. Equations (10) and (11) relate the optimal P_d and I_d, respectively, to the thermal design of the LED system (i.e. thermal resistance of the heatsink R_hs and R_jc.
  • 2. The maximum luminous flux will occur approximately at a lamp power P_d* specified in (10). This P_d* will shift to a lower value if (R_jc+NR_hs) is increased. This leads to the possibility that the P_d* may occur at a power level that is less than the rated power P_d(rated) of the LED.
  • 3. Based on the above comment, one should expect that the P_d* could be shifted to higher power level if a larger heatsink with lower R_hs is used.
  • 4. For many applications such as head lamps of vehicles and compact lamps for replacement of incandescent lamps, the size of the heatsink is highly restricted and the ambient temperature is high. In these cases, there is a high possibility that P_d* (at which maximum luminous flux is produced) will occur at a power level less than the rated power.

In practice, R_jc of the LED increases with lamp power. Therefore, a vigorous equation can be obtained from (8) as:

$\begin{array}{cc}\frac{{\phi }_v}{P_d}={\mathrm{NE}}_o\left\{\begin{array}{c}\left[1+k_e\left(T_a-T_o\right)\right]+\\ 2k_ek_h\left(R_{\mathrm{jc}}+{\mathrm{NR}}_{\mathrm{hs}}\right)P_d+\left(k_ek_hP_d^2\right)\frac{R_{\mathrm{jc}}}{P_d}\end{array}\right\}& \left(12\right)\end{array}$

The function of R_jc is highly complex and it depends on several factors such as thermal resistance of the heatsink, ambient temperature, the LED size and mounting structure and even the orientation of the heatsink. Equation (7b) in fact provides the physical meaning of effects of the temperature-dependent R_jc. Since R_jc increases with lamp power P_d, the two negative terms (with k_e which is negative) in (7b) will accelerate the reduction of the luminous flux as P_d increases. This effect should be noticeable when P_d exceeds the P_d*, resulting in a slightly asymmetric parabolic luminous flux function.

In order to verify the theory two types of LEDs are used: 3W cool white LEDs and 5W cool white LEDs from Luxeon K2 Star series. They are mounted on several heatsinks with thermal resistances of 6.3° C./W, 3.9° C./W and 2.2° C./W so that experiments can be performed to evaluate their luminous output under different lamp power operations.

Since the junction-to-case thermal resistance R_jc is a complex and nonlinear function of the lamp heat dissipation P_heat (which is equal to k_hP_d) and the thermal design of the mounting structure, the theoretical prediction is based on a simplified linear function as follows:

R _jc =R _jco(1+k_jcP_d)  (13)

where R_jco is the rated junction-to-case thermal resistance at 25° C. and k_jc is a positive coefficient. A typical linear approximation of R_jc is shown in FIG. 6.

If equation (13) is used in (7b), a more accurate luminous flux equation can be derived as:

φ_ν=NE_o{[1+k_e(T_a−T_o)]P_d+[k_ek_h(R_jco+NR_hs)]P_d²[k_ek_hk_jcR_jco]P_d³}  (7c)

A: Tests on 3W LEDs

(i) On a Heatsink with Thermal Resistance of 6.3° C./W

A group of eight identical Luxeon K2 Cool-white 3W LEDs are mounted on a standard heatsink with a thermal resistance of 6.3° C./W. The efficacy of the LEDs is measured at rated power in an integrating sphere. The parameters required for the equation (7) are:

k_e=−0.005, k_h=0.85, T_a=28° C., T₀=25° C., E₀=41 Lumen/Watt, N=8, R_hs=6.3° C./W, R_jco=10° C./W and k_jc=0.1° C./W².

Now two equations can be derived from (7). If the R_jc is assumed to be constant as a first approximation (i.e. R_jc=R_jco)=

φ_ν=323.08×P_d−84.2×P_d²  (14)

If R_jc is assumed to obey (13),

φ_ν=323.08×P_d−84.2×P_d²−1.39P_d³  (15)

The luminous flux is measured in an integrating sphere. The measured total luminous flux for eight LEDs is used for comparison with calculated values. The measured and calculated total luminous flux values are plotted, not against the total power sum of eight LEDs but against one LED power because the eight LEDs are identical and are connected in series. Using the power of one LED in the x-axis allows one to check easily if the optimal operating point is at the rated LED power or not. The measured results and calculated results from (14) and (15) are plotted in FIG. 7. Several points should be noted:

• • 1. The theoretical curves generally have the same trends as the measured curve. This confirms the validity of the general theory.
  • 2. The maximum lumen/Watt point occurs at about P_d=1.9W, which is less than the rated power of 3 W. This result shows that the general theory can predict accurately the P_d* operating point which may not be the rated power. Equation (10) indicates that a large NR_hs term will shift P_d* to the low power level of the LED.
  • 3. The two negative terms in this example can also be seen in (15). The asymmetry after the peak luminous output point is more noticeable in the theoretical curve obtained from (15) than from (14). Comparison of (14) and (15) shows that the effect of the variation of R_jc, which is reflected in the extra third term in (15) is the reason for the obvious asymmetry of φ_ν.
  • 4. In summary, the simplified model (7b), which is the basis for (14), has the form of φ_ν=α₁P_d−α₂P_d², while the more vigorous model (7c), which is the basis for (15), has the form of φ_ν=α₁P_d−α₂P_d²−α₃P_d². Therefore, the model in (7c) is more accurate than the model (7b) particularly when P_d has exceeded P_d*. However, since both simplified and vigorous models are accurate enough for the power less than P_d*, which is also the recommended useable power range of LEDs, both equations can be used in the design optimization procedure to be explained.

Based on (6), the efficacy function can also be obtained.

E=40.39−10.52P_d assuming is constant  (16)

E=40.39−10.52P_d−0.17P_d² assuming R_jc obeys (13)  (17)

The measured efficacy values and the calculated values from (16) and (17) are displayed in FIG. 6. It is noted that the calculated values are consistent with measurements. The results obtained from (17) are more accurate than those from (16) when P_d is large.

(ii) On a Heatsink with Thermal Resistance of 4.5° C./W

Eight identical 3W LEDs are mounted on a larger heatsink with thermal resistance of 4.5° C./W. The measured and calculated total luminous output as a function of single LED power P_d are shown in FIG. 9. It is noted that the calculated values are generally consistent with measurements, except at very low power where the light output is low and the relative measurement error is large. The P_d* is 2.4W at an efficacy of 21 lumens/Watt in this case. The use of a larger heatsink with a smaller thermal resistance means that the NR_hs term in the denominator of (10) is smaller than that in the previous case (with R_hs6.3° C./W). Therefore, P_d* has increased from 1.9W to 2.4W as expected from (10) and the efficacy from 21 lumens/Watt to 23 lumens/Watt.

The corresponding measured and calculated efficacy are shown in FIG. 10 and it can be seen that they are in good agreement.

(iii) On a Heatsink with Thermal Resistance of 2.2° C./W

Another eight 3W LEDs are mounted on an even larger heatsink with thermal resistance of 2.2° C./W for evaluation. The measured and calculated luminous output as a function of LED power P_d are shown in FIG. 11 and the corresponding results of the efficacy are included in FIG. 12.

The theoretical P_d* is now about 3.5W, which is higher than the rated power of 3 W. This again confirms the prediction by the theory (10) that P_d* will shift to the higher power level with a decreasing term of NR_hs (i.e a larger heatsink with a lower R_hd). Therefore, the theory can be used to design the optimal heatsink for a particular operating power. On the other hand, it can also be used to predict the optimal operating power for a given heatsink.

B: Tests on 5W LEDs

In order to ensure that the theory can be applied to other LEDs, 5W LEDs are used for evaluation. They are mounted on two heatsinks with thermal resistance of 6.8° C./W and 10° C./W respectively

(i) On a Heatsink with Thermal Resistance of 10° C./W

Two 5W LEDs are mounted on a heatsink with thermal resistance of 10° C./W. For the theoretical calculation, the parameters used in (10) are k_e=−0.00355, k_h=0.85, T_a=28° C., T₀=25, E₀=38 Lum/W, N=2, R_hs=10° C./W, R_jc=13° C./W and k_jc=0.13° C./W². Fitting these parameters into (7) and assuming that R_jc will rise linearly with temperature, the luminous flux equation and the efficacy equation are expressed as (18) and (19), respectively, and they are plotted with practical measurements in FIG. 13 and FIG. 14, respectively. Despite only two 5W LEDs being used, the theoretical predictions based on the averaged values are in general agreement with the measurements.

φ_ν=75.2P_d−7.57_d²−0.296P_d³  (18)

E=37.6−3.78P_d−0.149P_d²  (19)

(ii) On a Heatsink with Thermal Resistance of 6.8° C./W

The previous experiments are repeated by mounting the two 5W LEDs on a larger heatsink with a thermal resistance of 6.8° C./W. FIG. 15 and FIG. 16 show the comparisons of the measured and calculated luminous flux and efficacy, respectively. In general, calculated and measured results are in good agreement. Comparisons of the peak luminous flux in FIG. 13 and FIG. 14 confirm once again that using a larger heatsink (with lower thermal resistance) can shift the optimal operating point to a higher lamp power level. For the heatsink with R_hs=10° C./W, the optimal point occurs at about 3.8W. For the heatsink with R_hs=6.8° C./W, this optimal point has shifted to about 6W.

An important conclusion can be drawn from these results. The peak luminous flux (i.e. maximum φ_ν) occurs at a LED power P_d* that depends on the thermal design (i.e. the heatsink thermal resistance). In general, the larger the heatsink (the lower the heatsink thermal resistance or the better the cooling effects), the higher the peak luminous flux can be achieved. Since operating the LEDs at a power higher than their rated power will shorten the lifetime of LEDs drastically, the theory can be used to project the maximum luminous flux for a given thermal design. It can also be used to predict the optimal thermal design for maximum luminous flux output if the LEDs are designed to operate at rated power.

P_d* can be controlled by using different heatsinks with different thermal resistance. For a larger heatsink, R_hs becomes small and therefore P_d* will be shifted to the higher power level as shown in FIG. 17, where the values of P_d* are labeled as A, B, C and D as the size of heatsink (or cooling effect) increases. By assuming R_hs=0, a theoretical limit can be plotted as shown in FIG. 17. It is important to note that the operating LED power must not exceed the rated LED power (P_rated) otherwise the lifetime of the LED will be shortened. Therefore, the intersection points of these curves with the rated power limit indicate how the light output can be maximized.

It should be noted that a reduction of R_hs corresponds to an increase in the cooling effect. One way to achieve increased cooling is to increase the size of the heatsink. In FIG. 17, it can be seen that two curves with maximum points marked by C and D have relatively high luminous flux at the rated power. The curve with maximum point D has a smaller R_hs and thus a larger heatsink than that with maximum point C. The increase of luminous flux at the rated power from using curve C to curve D is small, but the increase in the size of the heatsink is much larger in proportion.

Three important points are highlighted here:

• 1. The maximum φ_ν is the point of inflexion of the luminous flux function (7b) or (7c). As P_d increases from zero, the positive slope of the curve

$\left(i.e.\frac{{\phi }_v}{P_d}\right)$

is gradually decreasing to zero when the peak of the curve is reached. A large positive slope means that a relatively small increase of P_d can result in a relatively large increase of φ_ν. So the initial linear portion of the curve results in good efficacy. As P_d is moved to the region at and around P_d*, the slope is zero or relatively small. Therefore, a relatively large increase in P_d will give a relatively small increase in φ_ν.

• 2. The LED power P_d must not exceed the rated LED power P_d(rated). Otherwise, the lifetime of the LED will be shortened. Therefore, the intersection points of these curves with the rated power limit should indicate how the light output can be maximized. The intersection points of these curves and the rated power line are denoted as “a”, “b”, “c” and “d” as shown in FIG. 17.
• 3. The values of φ_ν at “c” and “d” are higher than that at “b”. But the curve with peak φ_ν at D requires a much larger heatsink than that with peak φ_ν at C. The difference of φ_ν at “c” and “d” may not be significant enough to justify an increase in cost and size of the heatsink.

The following rules are proposed as an optimization.

Rule 1:

The function of the luminous flux versus LED power P_d is a parabolic curve with a maximum point. The operating point P_d should be chosen at or below the maximum point P_d*. This means that for a given luminous flux output, the lower LED power should be chosen. Within this recommended power range, either (7) or (14) can provide sufficiently accurate predictions.

Rule 2:

If the thermal design is restricted by limited space for the heatsink so that the P_d* occurs at a power less than or equal to the rated power P_(rated), then the LED system should be operated at P_d* for each LED device. [For example, points A and B are optimal operating points for the respective curves as their P_d* values do not exceed P_(rated).]

Rule 3:

If the thermal design is flexible, then the LED system should be designed in such a way that (i) the theoretical maximum φ_ν point (or P_d*) occurs at a power higher than P_(rated) of the LED and (ii) the intersection point of the theoretical φ_ν-P_d curve and the rated power line should have a value of about 80% to 96% of the theoretical maximum φ_ν value. The rated power should be chosen as the operating power for each LED.

Rule 3 is an important idea. Where the theoretical maximum (P_d* for maximum φ_ν) occurs at a point higher than the rated power, one should still operate the LED system at the rated power. As can be seen from FIGS. 17 and 18 the slope of the curve shows that as P_d* is approached the increase in luminous flux becomes very small and in terms of efficiency the gain in flux is not worth the additional power used. By using Δφ_ν|_p11=0.04˜0.20 of the maximum φ_ν point in the curve the 4%-20% range for Δφ_ν from the maximum φ_ν point offers a good compromise of the light output and the size and thus cost of the heatsink.

If forced cooling is applied, the φ_ν-P_d curve will change dynamically. This can be visualized as having a dynamically changing thermal resistance R_hs. The optimal operating point should follow the three rules explained previously. It should be kept along the operating lines as highlighted in the bold solid lines in FIG. 19 in order to maximize the luminous flux output.

FIG. 20 shows schematically an apparatus that may be used in embodiments of the invention. The apparatus comprises a microprocessor control unit (MCU) 20 that performs functions to be described below, database 21, user input means 22 which may be a keyboard, touchscreen or any other means that enables a user to display data, and output display means 23 which may be a screen, print out or any other means for data output to be communicated to a user. Database 21 includes details obtained from datasheets of the physical and electrical parameters of all known commercially available LEDs. This database may be provided directly as part of the apparatus or may a database kept elsewhere and accessed remotely. A user will input selected parameters of a desired LED lighting system using input means 22. These parameters will include at least the desired luminous flux output required from the system, and may further include any other parameters that the user wishes to fix, including for example the number N of LEDs and the size of the heatsink if that is fixed.

MCU 20 is programmed to carry out the steps shown in the flowcharts of FIGS. 21 and 22. Beginning with FIG. 21, a user may input a required flux output φ_ν. The MCU may then select a candidate LED from the database and calculated the maximum flux that the LED is capable of achieving from Eq.7 above. If this maximum flux is obtained at a power below the rated power, and if this maximum flux is equal to or greater than the required flux, then such an LED is capable of being used in an LED system meeting the desired flux output and the process can stop. If the answer to either question is negative, ie if the maximum flux would only be obtained at a power greater than the rated power, or if the maximum flux is insufficient, then another LED is chosen and the process repeats.

If no LED can be found by the process of FIG. 21, either after a number of attempts or after all LEDs in the database have been exhausted, the MCU runs the process of FIG. 22 (or alternatively a user may go directly to the process of FIG. 22 if preferred). This process corresponds to the situation where the peak of the flux output occurs at a power above the rated power. In such cases, as discussed above, it is preferable to select an LED with a peak flux output such that the maximum rated power is between 80% and 96% of the power at which the maximum flux is obtained. As can be seen from lines C and D in FIG. 17 the output of these lines where they cross the rated power is higher than the equivalent flux of an LED that has its peak flux output exactly when operated at the rated power (the condition of line B). In FIG. 21 a peak output is calculated by the MCU at a power higher than a rated power and then various LEDs are chosen from the database until one is found where the flux output φ_ν at the rated power is equal to or greater than the required power φ_req.

As mentioned above, unlike traditional power electronic circuits, the main factor that needs to be optimized in a lighting system is the luminous efficacy instead of energy efficiency. Good luminous efficacy would automatically imply good efficacy, but not vice versa. The general (steady-state) photo-electro-thermal (PET) theory for LED systems, as described above, has pointed out both theoretically and practically that an LED device will not generate the maximum amount of luminous output at its rated power unless the LED system is optimally designed in an integrated manner. Optimal design of LED systems can only be achieved with the proper choice of LED devices and array structures, LED drivers, operating powers and thermal designs. Various aspects of LED systems such as the thermal management, non-linear behavior of junction-to-case thermal resistance, LED drivers and current sharing techniques for LED strings have been reported. While the interactions of photometric, electric and thermal aspects of the LED systems have been linked together, very limited research on the time-dependency of the luminous performance of LED systems has been reported.

This time factor is in fact highly relevant to the actual luminous output of any LED system. For LED systems designed for continuous operation, such as road lighting systems and LED bulbs, the luminous output will drop from the initial to steady-steady operation. The main reason is due to the gradual increase in the LED junction and heatsink temperature values. Therefore, the time factor is critical in understanding how much luminous reduction an LED system will have so that proper LED systems can be designed. For LED systems with momentarily or discontinuous operation, such as traffic lights and signal indicators of vehicles, the design criteria would be different from those for continuous operation.

The PET theory describes the steady-state performance of an LED system. The following is a photo-electro-thermo-temporal (PETT) theory that incorporates the time domain into the PET theory so that the dynamic behavior of the LED systems can be studied. The PETT theory can also be termed a dynamic PET theory. This PETT theory or dynamic PET theory will converge to the steady-state theory under steady-state conditions. Since the luminous efficacy is the main design factor for LED systems, this dynamic theory provides a useful tool not only for studying the behavior of LED systems, but more importantly, for optimizing LED system designs for both continuous and discontinuous operations.

As a special application of this PETT theory, the dynamic variation of the junction temperature can be predicted accurately. This theory therefore provides a valuable numerical design tool for LED systems.

In the following description of the PETT theory, the symbols and labels for various quantities and terms in formulae refer primarily to those quantities and terms under the described PETT theory. However, it will be appreciated that some of the symbols and labels refer to equivalent quantities and terms under the PET theory described above. Furthermore, where a formula under the PETT theory has an equivalent formula under the PET theory described above, the reference numeral of the equivalent formula under the PET theory appears in square brackets “[ ]” and in italics after the reference numeral of the formula under the PETT theory, so that the equivalent formulae can be easily compared.

FIG. 23 shows a simplified thermal equivalent circuit of N LEDs mounted on a heatsink. The heat source P_heat is the amount of heat generated by the LED. The internal junction temperature of the LED device is T_j. The thermal resistor and thermal capacitor of the power device are labeled as R_jc and C_jc respectively. T_hs is heatsink surface temperature, R_hs and C_hs are heatsink thermal resistor and thermal capacitor respectively. In practice, the LED packages are mounted on the heatsink with some form of electrical insulation or with the use of thermal paste to ensure good thermal contact. The thermal resistance of this thermally conductive layer is usually much smaller than R_jc and R_hs and is thus ignored in the following analysis.

The amount of heat generated by an LED can be expressed as:

P _heat =k _h P _d  (T1)

where k_h is the heat dissipation coefficient that represents the portion of input power that is dissipated as heat and P_d is the input power of each LED.

In FIG. 23, some heat (P₁) flows from the equivalent heat source through the equivalent thermal resistor, and another portion of the heat (P_Cj) flows into the equivalent thermal capacitor of the LED package. Based on the circuit analysis technique, it can be shown that:

$\begin{array}{cc}{P}_{C_j}={\mathrm{Nk}}_hP_d-P_1& \left(\mathrm{T2}\right)\\ P_{C_j}={\mathrm{NC}}_{\mathrm{jc}}\frac{\left(T_j-T_a\right)}{t}& \left(\mathrm{T3}\right)\\ P_1=\frac{T_j-T_{\mathrm{hs}}}{R_{\mathrm{jc}}/N}& \left(\mathrm{T4}\right)\end{array}$

Based on (T2), (T3) and (T4),

$\begin{array}{cc}{\mathrm{NC}}_{\mathrm{jc}}\frac{\left(T_j-T_a\right)}{t}={\mathrm{Nk}}_hP_d-\frac{T_j-T_{\mathrm{hs}}}{R_{\mathrm{jc}}/N}& \left(\mathrm{T5}\right)\end{array}$

Rearranging (T5) gives

$\begin{array}{cc}\frac{T_j}{t}+\frac{1}{C_{\mathrm{jc}}R_{\mathrm{jc}}}T_j=\frac{k_hP_d}{C_{\mathrm{jc}}}+\frac{T_{\mathrm{hs}}}{C_{\mathrm{jc}}R_{\mathrm{jc}}}& \left(T6\right)\end{array}$

In reality, both of the junction temperature and the heatsink temperature will change with time under normal operation. In practice, the thermal time constant of the LED package (τ_jc=R_jcC_jc which is typically tens of seconds) is much smaller than that of the heatsink (τ_hs=R_hsC_hs which is typically hundreds of seconds). The heatsink temperature T_hs will change much slowly than LED junction temperature T_j. In order to avoid confusion, the terms ‘fast transient’ and ‘slow transient’ refer to the time frames in the order of the τ_jc and τ_hs, respectively. Under fast transient situation, T_hs can be considered as a constant. Based on this argument, the dynamic relationship between T_j and T_hs under the fast transient condition can be obtained from (T6) as

$\left[\mathrm{cf}.\left(5a\right)\right]$ $\begin{array}{cc}{T}_j=R_{\mathrm{jc}}k_hP_d\left(1-{}^{-\frac{t}{C_{\mathrm{jc}}R_{\mathrm{jc}}}}\right)+T_{\mathrm{hs}}& \left(\mathrm{T7}\right)\end{array}$

For the heatsink, heat flows into the thermal capacitor C_hs and the thermal resistor R_hs of the heatsink. The heat flow component P₁ can be expressed as:

P ₁ =P _{C hs} +P ₂  (T8)

where P_Chs is the heat flowing into the thermal capacitor of heatsink, and P₂ is the heat through the thermal resistor of the heatsink. They can be formulated as:

$\begin{array}{cc}{P}_{C_{\mathrm{hs}}}=C_{\mathrm{hs}}\frac{\left(T_{\mathrm{hs}}-T_a\right)}{t}& \left(\mathrm{T9}\right)\\ P_2=\frac{T_{\mathrm{hs}}-T_a}{R_{\mathrm{hs}}}& \left(\mathrm{T10}\right)\end{array}$

From (T4), (T8), (T9) and (T10), the heatsink temperature is therefore:

$\begin{array}{cc}\frac{T_j-T_{\mathrm{hs}}}{R_{\mathrm{jc}}/N}=C_{\mathrm{hs}}\frac{\left(T_{\mathrm{hs}}-T_a\right)}{t}+\frac{T_{\mathrm{hs}}-T_a}{R_{\mathrm{hs}}}& \left(T11\right)\end{array}$

which can be rewritten as:

$\begin{array}{cc}\frac{T_{\mathrm{hs}}}{t}=\frac{N}{C_{\mathrm{hs}}R_{\mathrm{jc}}}T_j-\frac{{\mathrm{NR}}_{\mathrm{hs}}+R_{\mathrm{jc}}}{C_{\mathrm{hs}}R_{\mathrm{hs}}R_{\mathrm{jc}}}T_{\mathrm{hs}}+\frac{T_a}{C_{\mathrm{hs}}R_{\mathrm{hs}}}& \left(\mathrm{T12}\right)\end{array}$

Now, the relationship of T_j and T_hs obtained in (T7) can be used in (T12):

$\begin{array}{cc}\frac{T_{\mathrm{hs}}}{t}=\frac{N}{C_{\mathrm{hs}}R_{\mathrm{jc}}}\left(-R_{\mathrm{jc}}k_hP_d{}^{-\frac{t}{C_{\mathrm{jc}}R_{\mathrm{jc}}}}+R_{\mathrm{jc}}k_hP_d+T_{\mathrm{hs}}\right)-\frac{{\mathrm{NR}}_{\mathrm{hs}}+R_{\mathrm{jc}}}{C_{\mathrm{hs}}R_{\mathrm{hs}}R_{\mathrm{jc}}}T_{\mathrm{hs}}+\frac{T_a}{C_{\mathrm{hs}}R_{\mathrm{hs}}}& \left(\mathrm{T13}\right)\end{array}$

Solving equation (T13), the heatsink temperature T_hs can be obtained as:

$\begin{array}{cc}{T}_{\mathrm{hs}}\left(t\right)=\frac{-R_{\mathrm{jc}}C_{\mathrm{jc}}{\mathrm{NR}}_{\mathrm{hs}}k_hP_d}{C_{\mathrm{jc}}R_{\mathrm{jc}}-C_{\mathrm{hs}}R_{\mathrm{hs}}}{}^{-\frac{t}{C_{\mathrm{jc}}R_{\mathrm{jc}}}}+{\mathrm{NR}}_{\mathrm{hs}}k_hP_d+A{}^{-\frac{t}{C_{\mathrm{hs}}R_{\mathrm{hs}}}}+T_a& \left(\mathrm{T14}\right)\end{array}$

where A represents a constant, which can be determined from the physical boundary condition of the heatsink. The boundary condition of a heatsink is that at t=0 the heatsink temperature is equal to ambient temperature. That is,

T _hs(t=0)=T_a  (T15)

Putting (T15) into (T14), the coefficient A can be obtained as:

$A=-\left(\frac{-R_{\mathrm{jc}}C_{\mathrm{jc}}{\mathrm{NR}}_{\mathrm{hs}}k_hP_d}{C_{\mathrm{jc}}R_{\mathrm{jc}}-C_{\mathrm{hs}}R_{\mathrm{hs}}}+{\mathrm{NR}}_{\mathrm{hs}}k_hP_d\right)$

Putting A into equation (T13), the heatsink temperature can be obtained as:

$\begin{array}{cc}\left[\mathrm{cf}.\left(4\right)\right]& \\ T_{\mathrm{hs}}\left(t\right)=\frac{-R_{\mathrm{jc}}C_{\mathrm{jc}}{\mathrm{NR}}_{\mathrm{hs}}k_hP_d}{C_{\mathrm{jc}}R_{\mathrm{jc}}-C_{\mathrm{hs}}R_{\mathrm{hs}}}{}^{-\frac{t}{C_{\mathrm{jc}}R_{\mathrm{jc}}}}-k_hP_d\left(\frac{-{\mathrm{NR}}_{\mathrm{hs}}R_{\mathrm{jc}}C_{\mathrm{jc}}}{C_{\mathrm{jc}}R_{\mathrm{jc}}-C_{\mathrm{hs}}R_{\mathrm{hs}}}+{\mathrm{NR}}_{\mathrm{hs}}\right){}^{-\frac{t}{C_{\mathrm{hs}}R_{\mathrm{hs}}}}+{\mathrm{NR}}_{\mathrm{hs}}k_hP_d+T_a& \left(\mathrm{T16}\right)\end{array}$

By putting the result of (T16) into (T7), the complete dynamic relationship of T_j and T_hs is:

$\begin{array}{cc}\left[\mathrm{cf}.\left(5b\right)\right]& \\ T_j=\left\{\begin{array}{c}-R_{\mathrm{jc}}k_hP_d\left(\frac{C_{\mathrm{jc}}{\mathrm{NR}}_{\mathrm{hs}}}{C_{\mathrm{jc}}R_{\mathrm{jc}}-C_{\mathrm{hs}}R_{\mathrm{hs}}}+1\right){}^{-\frac{t}{C_{\mathrm{jc}}R_{\mathrm{jc}}}}-\\ k_hP_d\left(\frac{-{\mathrm{NR}}_{\mathrm{hs}}R_{\mathrm{jc}}C_{\mathrm{jc}}}{C_{\mathrm{jc}}R_{\mathrm{jc}}-C_{\mathrm{hs}}R_{\mathrm{hs}}}+{\mathrm{NR}}_{\mathrm{hs}}\right){}^{-\frac{t}{C_{\mathrm{hs}}R_{\mathrm{hs}}}}+\\ \left(R_{\mathrm{jc}}+{\mathrm{NR}}_{\mathrm{hs}}\right)k_hP_d+T_a\end{array}\right\}& \left(\mathrm{T17}\right)\end{array}$

The luminous efficacy (E) has the following relationship with the junction temperature T_j of the LED.

E=E _o└1+k_e(T_j−T_o)┘  (T18)

Now, T_j obtained in (T17) can be used in (T18):

$\begin{array}{cc}\left[\mathrm{cf}.\left(6\right)\right]& \\ E=E_o\left[\begin{array}{c}1+k_e\left(T_a-T_o\right)+k_ek_h\left(R_{\mathrm{jc}}+{\mathrm{NR}}_{\mathrm{hs}}\right)P_d-\\ k_eR_{\mathrm{jc}}k_hP_d\left(\frac{{\mathrm{NR}}_{\mathrm{hs}}C_{\mathrm{jc}}}{C_{\mathrm{jc}}R_{\mathrm{jc}}-C_{\mathrm{hs}}R_{\mathrm{hs}}}+1\right){}^{-\frac{t}{C_{\mathrm{jc}}R_{\mathrm{jc}}}}-\\ k_ek_hP_d\left(\frac{-R_{\mathrm{jc}}C_{\mathrm{jc}}{\mathrm{NR}}_{\mathrm{hs}}}{C_{\mathrm{jc}}R_{\mathrm{jc}}-C_{\mathrm{hs}}R_{\mathrm{hs}}}+{\mathrm{NR}}_{\mathrm{hs}}\right){}^{-\frac{t}{C_{\mathrm{hs}}R_{\mathrm{hs}}}}\end{array}\right]& \left(\mathrm{T19}\right)\end{array}$

So the total luminous flux φ_ν is

$\begin{array}{cc}\left[\mathrm{cf}.\left(7b\right)\right]& \\ {\phi }_v={\mathrm{NEP}}_d{\phi }_v=\mathrm{NE}\left\{\begin{array}{c}\left[1+k_e\left(T_a-T_o\right)\right]P_d+k_ek_h\left(R_{\mathrm{jc}}+{\mathrm{NR}}_{\mathrm{hs}}\right)P_d^2-\\ k_ek_hR_{\mathrm{jc}}\left(\frac{{\mathrm{NR}}_{\mathrm{hs}}C_{\mathrm{jc}}}{C_{\mathrm{jc}}R_{\mathrm{jc}}-C_{\mathrm{hs}}R_{\mathrm{hs}}}+1\right){}^{-\frac{t}{C_{\mathrm{jc}}R_{\mathrm{jc}}}}P_d^2-\\ k_ek_h\left(\frac{-R_{\mathrm{jc}}C_{\mathrm{jc}}{\mathrm{NR}}_{\mathrm{hs}}}{C_{\mathrm{jc}}R_{\mathrm{jc}}-C_{\mathrm{hs}}R_{\mathrm{hs}}}+{\mathrm{NR}}_{\mathrm{hs}}\right){}^{-\frac{t}{C_{\mathrm{hs}}R_{\mathrm{hs}}}}P_d^2\end{array}\right\}& \left(\mathrm{T20}\right)\end{array}$

Equations (T16) to (T20) now form the dynamic equations that describe the essential variables of the LED system. These dynamic equations can converge to the steady-state equations as time variable t approaches infinity. As t→∞

eqn.(T16) becomes

T _hs(t)=NR_hsk_hP_d+T_a  (T21)[cf.(4)]

eqn.(T17) becomes

T _j=(R_jc+NR_hs)k_hP_d+T_a  (T22)[cf.(5b)]

eqn.(T19) becomes

E=E _o└1+k_e(T_a−T_o)+k_ek_h(R_jc+NR_hs)P_d┘  (T23)[cf.(6)]

eqn.(T20) becomes

φ_ν=NE{[1+k_e(T_a−T_o)]P_d+k_ek_h(R_jc+NR_hs)P_d²}(T24)[cf.(7b)]

It can be seen that (T21), (T22), (T23) and (T24) are identical to the steady-state equations of the general PET theory described previously (the reference numerals of the equivalent equations of the PET theory are shown in square brackets and italics directly after the reference numerals of these PETT equations).

Because the thermal time constant of the LED device package is typically in the order to tens of seconds (and that of the heatsink is typically tens of minutes), the junction temperature of the LED does not change significantly within tens of micro-seconds. Therefore, it can be assumed that the junction temperature remains constant if the time frame is small compared with the thermal time constant of the LED device package. For example, for an off-line LED system powered by the ac mains, the mains frequency is either 50 Hz or 60 Hz. So the period of the mains half cycle is typically 10 ms or 8.33 ms which is small compared with the time constant of the LED package of tens of seconds. Thus, the steady-state equation (T24) can be rewritten as:

$\begin{array}{cc}{R}_{\mathrm{jc}}=\frac{{\phi }_v}{{\mathrm{NE}}_ok_ek_h}P_d^{-2}-\frac{1+k_e\left(T_a-T_o\right)}{k_ek_h}P_d^{-1}-{\mathrm{NR}}_{\mathrm{hs}}\mathrm{for}P_d>0& \left(\mathrm{T25}\right)\end{array}$

It is important to note that R_jc in (T25) can be determined from the measured LED power P_d and luminous flux φ_ν and other measurable coefficients, heatsink temperature and temperature. With this R_jc obtained, the dynamic junction temperature can be obtained from (T17):

$\begin{array}{cc}{T}_j=\left\{\begin{array}{c}-R_{\mathrm{jc}}k_hP_d\left(\frac{C_{\mathrm{jc}}{\mathrm{NR}}_{\mathrm{hs}}}{C_{\mathrm{jc}}R_{\mathrm{jc}}-C_{\mathrm{hs}}R_{\mathrm{hs}}}+1\right){}^{-\frac{t}{C_{\mathrm{jc}}R_{\mathrm{jc}}}}-\\ k_hP_d\left(\frac{-{\mathrm{NR}}_{\mathrm{hs}}R_{\mathrm{jc}}C_{\mathrm{jc}}}{C_{\mathrm{jc}}R_{\mathrm{jc}}-C_{\mathrm{hs}}R_{\mathrm{hs}}}+{\mathrm{NR}}_{\mathrm{hs}}\right){}^{-\frac{t}{C_{\mathrm{hs}}R_{\mathrm{hs}}}}+\\ \left(R_{\mathrm{jc}}+{\mathrm{NR}}_{\mathrm{hs}}\right)k_hP_d+T_a\end{array}\right\}& \left(\mathrm{T17}\right)\end{array}$

For the steady-state conditions, eqn.(T17) becomes

T _j=(R_jc+NR_hs)k_hP_d+T_a  (T22)

The equations of the dynamic model described above can form the basis of numerical design tools and methods as follows.

Provided the junction thermal resistance is known or approximated, the flowchart of FIG. 24 shows an example of using the above dynamic model in a numerical design tool or a method to determine:

(i) the luminous flux;(ii) the luminous efficacy;(iii) the heatsink temperature; or(iv) the LED junction temperature.

This approach gives a good estimation of the luminous performance and can thus be used iteratively to fine tune LED system designs.

Further important embodiments of a numerical design tool and a method provide data on the junction temperature in a practical LED system in order to make sure that the junction temperature will not exceed the maximum temperature limits as specified in the LED device data sheets. FIG. 25 shows a flowchart of an embodiment of a prediction method or a numerical design tool based on practical measurements of the luminous flux and LED power. If the peak junction temperature exceeds the maximum limit, then the design should be modified.

Thus, the dynamic model described above can be embodied as numerical tools and methods, or form part of numerical tools and methods. Preferably, these numerical tools and methods are implemented, or partly implemented, on a computer. These computer-implemented numerical tools and methods can be in the form of a software tool, or can involve the use of a software tool. In some embodiments, the computer-implemented numerical tools, or the computer-implemented parts of the methods, are accessible from an internet website. In another embodiment, the computer-implemented numerical tools, or the computer-implemented parts of the methods, are stored on a portable computer-readable medium, such as a CD, DVD, or a flash memory device. These can then be sold separately for implementation on a user's computer.

Tests were carried out as practical confirmation of the accuracy of the numerical tools described above.

(i) Tests on CREE X Lamp XR-E 3W LEDs

Eight CREE 3W LEDs were mounted on heatsinks in two sets of tests with the heatsink thermal resistance equal to 1.8° C./W (Sample A) and 2.9° C./W (Sample B). Instead of using the constant R_jc value in the LED data sheet, the curves formed by the calculated R_jc values based on (T25) are used for predicting relationships of the luminous flux and LED power at different time intervals. Table 1 below shows the measured data and the corresponding calculated R_jc values when the two heatsinks are used.

TABLE 1
Parameter list for the calculation of Rjc of CREE X lamp XR-E LEDs mounted
on heatsinks with thermal resistances of 1.8° C./W and 2.9° C./W.
Heatsink	Power(watt)	1.94	2.27	2.65	3.00	3.36	3.73	4.10
with	Luminous	1247	1367	1476	1546	1617	1674	1711
1.8° C./W	flux (lumen)
thermal	Calculated	9.4	17.5	23.4	28.1	29.9	31.1	31.8
resistance	Rjc (° C./W)
Heatsink	Power(watt)	1.91	2.26	2.60	2.96	3.31	3.67	4.02
with	Luminous	1184	1294	1384	1443	1490	1521	1533
2.9° C./W	flux (lumen)
thermal	Calculated	10.4	18.5	23.6	28.0	30.1	31.3	32.1
resistance	Rjc (° C./W)

(a) On a Heatsink with Thermal Resistance of 1.8° C./W:

The curves of the luminous flux are measured and recorded at different time intervals as shown in FIG. 26. The parameters τ_LED=C_jcR_jc, τ_hs=C_hsR_hs, R_hs and R_jc are listed in Table 2 below. Based on the dynamic theory (PETT), the theoretical curves of the luminous flux obtained from (T20) at the same time intervals are plotted in FIG. 27.

TABLE 2
LED system parameters.
ke	kh	Eo	Ta	To	N	Rhs	τLED = CjcRjc	τhs = ChsRhs
−0.0025	0.87	90	28	25	8	1.8	45	520

(b) On a Heatsink with Thermal Resistance of 2.9° C./W.

The measured luminous flux curves are shown in FIG. 28. For the theoretical curves, the parameters τ_LED=C_jcR_jc, τ_hs=C_hsR_hs and R_hs are listed in Table 3 below. The theoretical luminous flux curves versus LED power at different time intervals generated by (T20) are plotted in FIG. 29. Very good agreements between the measured and theoretical curves are observed.

TABLE 3
LED system parameters.
ke	kh	Eo	Ta	To	N	Rhs	τLED = CjcRjc	τhs = ChsRhs
−0.0025	0.87	90	28	25	8	2.9	45	510

Based on (T20), the measured and calculated luminous flux versus time curves of the two examples above are plotted in FIG. 30. These results confirm the accuracy of the method and numerical tool for forming and optimizing the performance of LED systems as proposed in the present invention.

(ii) Off-Line LED Bulb

In another example, an off-line LED bulb (with an electronic LED driver and LED devices) mounted on a heatsink with thermal resistance of 4.1° C./W is simulated. In the simulation, R_jc is determined from equation (T25) using the practical luminous flux measurements. The variations of the junction temperature are predicted and recorded at different intervals (from 2 minutes to 27 minutes) after the LED bulb is turned on.

These junction temperature curves are plotted in FIG. 31. It is clear that this numerical tool can predict the gradual increase in junction temperature (due to the gradual temperature increase in the heatsink which has a longer thermal time constant of tens of minutes). From FIG. 31, it is important to note that, under steady-state conditions, the peak junction temperature exceeds 150° C., which is the maximum temperature of many LED devices. Such operation will definitely reduce the lifetime of the LED product. Therefore, this numerical tool allows the product designer to predict this important factor.

Now the heatsink with thermal resistance of 4.1° C./W is replaced by a large heatsink with thermal resistance of 2.2° C./W and the corresponding junction temperature curves are plotted in FIG. 32. It can be seen that the peak junction temperature now falls below the maximum temperature of 150° C. and therefore the product will be more reliable.

The internal junction temperature of LED packages cannot be easily assessed directly. In LED system design, it is important to make sure that the junction temperature of an LED does not exceed its maximum rating in order to prolong its lifetime. As shown above, the present invention offers a method and a numerical tool that can be used to predict the junction temperature in a computer simulation or other computational environment. It can be used as a design tool for LED systems.

Although the invention has been described with reference to specific examples, it will be appreciated by those skilled in the art that the invention can be embodied in many other forms. It will also be appreciated by those skilled in the art that the features of the various examples described can be combined in other combinations.

Claims

1. A method of forming or optimizing an LED system to achieve a desired luminous performance over a time period, the LED system comprising a single or a plurality of LEDs on a heatsink with each LED having a junction, the method comprising the steps of: (a) calculating a predicted luminous performance over the time period with a computer-implemented dynamic model; and(b) modifying the LED system such that the predicted luminous performance substantially matches the desired luminous performance.

2. A method according to claim 1 wherein luminous performance is one or more of luminous flux of the LED system, luminous efficacy of the LED system, heatsink temperature and junction temperature over the time period.

3. A method according to claim 2 wherein the computer-implemented dynamic model comprises the following equations for calculating the luminous flux of the LED system, the luminous efficacy of the LED system, the heatsink temperature and the junction temperature over the time period:

4. A method according to claim 1 wherein the computer-implemented dynamic model is in the form of a software tool on a computer accessible through an interne website.

5. A method according to claim 1 wherein the computer-implemented dynamic model is in the form of a software tool stored on a portable computer-readable medium.

6. A computer-implemented numerical tool that calculates a predicted luminous performance over a time period for an LED system comprising a single or a plurality of LEDs on a heatsink with each LED having a junction, thereby allowing modification of the LED system such that the predicted luminous performance substantially matches a desired luminous performance in order to form or optimize the LED system to achieve the desired luminous performance.

7. A computer-implemented numerical tool according to claim 6 wherein luminous performance is one or more of luminous flux of the LED system, luminous efficacy of the LED system, heatsink temperature and junction temperature over the time period.

8. A computer-implemented numerical tool according to claim 7 that calculates one or more of the luminous flux of the LED system, the luminous efficacy of the LED system, the heatsink temperature and the junction temperature over the time period using the following equations:

9. A computer-implemented numerical tool according to claim 6 wherein the computer-implemented numerical tool is in the form of a software tool on a computer accessible through an internet website.

10. A computer-implemented numerical tool according to claim 6 wherein the computer-implemented numerical tool is in the form of a software tool stored on a portable computer-readable medium.

11. A method of optimizing an LED system comprising a single or a plurality of LEDs on a heatsink with each LED having a junction, the method comprising the step of calculating junction temperature over a time period with a computer-implemented dynamic model, thereby allowing modification of the LED system if a calculated peak junction temperature exceeds a maximum rated junction temperature, and thereby optimizing the LED system.

12. A method according to claim 11 wherein the computer-implemented dynamic model comprises the following equation for calculating the junction temperature:

13. A method according to claim 12 wherein the computer-implemented dynamic model comprises the following equation for calculating Rjc:

14. A method according to claim 11 wherein the computer-implemented dynamic model is in the form of a software tool on a computer accessible through an internet website.

15. A method according to claim 11 wherein the computer-implemented dynamic model is in the form of a software tool stored on a portable computer-readable medium.

16. A computer-implemented numerical tool for use with an LED system comprising a single or a plurality of LEDs on a heatsink with each LED having a junction, the computer-implemented numerical tool calculating junction temperature over a time period, thereby allowing modification of the LED system if a calculated peak junction temperature exceeds a maximum rated junction temperature, and thereby optimizing the LED system.

17. A computer-implemented numerical tool according to claim 16 that calculates the junction temperature over the time period using the following equation:

18. A computer-implemented numerical tool according to claim 17 that calculates Rjc using the following equation:

19. A computer-implemented numerical tool according to claim 16 wherein the computer-implemented numerical tool is in the form of a software tool on a computer accessible through an Internet website.

20. A computer-implemented numerical tool according to claim 16 wherein the computer-implemented numerical tool is in the form of a software tool stored on a portable computer-readable medium.

21. A method of forming an LED system comprising a single or a plurality of LEDs on a heatsink with a desired output flux, the method comprising the steps of: (a) modeling on a computer the luminous flux emitted by said LED system as a function of the thermal resistance of said heatsink and the power applied to each LED, and(b) selecting an LED system such that the maximum luminous flux is emitted at a power equal to or below a rated power of said LED system provided that said maximum luminous flux is equal to or greater than the desired output flux, or(c) selecting an LED such that the rated power of said LED system is below the power at which the maximum luminous flux is emitted, provided that the flux emitted by said LED system at said rated power is equal to or greater than the desired output flux.

22. A method as claimed in claim 21 wherein in step (c) the rated power is at between 80% and 96% of the power at which maximum flux would be output.

23. A method of forming an LED system comprising a single or a plurality of LEDs on a heatsink with a desired output flux, the method comprising the steps of: (a) modeling on a computer the luminous flux emitted by said LED system as a function of the thermal resistance of said heatsink and the power applied to each LED, and(b) selecting a heatsink having a thermal resistance such that the maximum luminous flux is emitted at a power equal to or below a rated power of said LEDs, or(c) selecting a heatsink having a thermal resistance such that the rated power of said LED system is below the power at which the maximum luminous flux is emitted, provided that the flux emitted by said LED system at said rated power is equal to or greater than the desired output flux.

24. A method as claimed in claim 23 wherein in step (c) the rated power is at between 80% and 96% of the power at which maximum flux would be output.

25. An LED system comprising a plurality of LEDs on a heatsink, wherein said heatsink has a thermal resistance such that the maximum luminous flux is emitted at a power below a rated power of said LEDs.

26. A method of forming an LED system comprising a plurality of LEDs on a heatsink with a desired output flux, the method comprising the steps of: (a) selecting an LED system such that the maximum luminous flux is emitted at a power below a rated power of said LED system provided that said maximum luminous flux is equal to or greater than the desired output flux, or(b) selecting an LED such that the rated power of said LED system is below the power at which the maximum luminous flux is emitted, provided that the flux emitted by said LED system at said rated power is equal to or greater than the desired output flux.

27. A method as claimed in claim 26 wherein in step (b) the rated power is at between 80% and 96% of the power at which maximum flux would be output.

28. A method of forming an LED system comprising a plurality of LEDs on a heatsink with a desired output flux, the method comprising the steps of: (a) selecting a heatsink having a thermal resistance such that the maximum luminous flux is emitted at a power below a rated power of said LEDs, or(b) selecting a heatsink having a thermal resistance such that the rated power of said LED system is below the power at which the maximum luminous flux is emitted, provided that the flux emitted by said LED system at said rated power is equal to or greater than the desired output flux.

29. A method as claimed in claim 28 wherein in step (b) the rated power is at between 80% and 96% of the power at which maximum flux would be output.