LED Panel

[sherman]

In fact, this means that by choosing between two fixtures of same power but a three-times price difference, one may yield less light per dollar from the cheaper fixture .
At the same time, the less efficient fixture will spend much more in power bills in a fixture’s lifetime.
Worst of all, it is almost impossible to accurately measure LEDs efficiency at home, and manufacturers are tempted to indicate a higher efficiency for the LEDs they use.

We believe that the choice of most efficient LEDs is very important when building a fixture. As we have shown, their higher cost will be justified in time.

[sherman]

There are a few other important issues which are often poorly addressed in existing fixtures.
We will look into them in another few days.
We hope you digest and understand what manufacturers never tell you about led

[sherman]

It is almost impossible to accurately measure LEDs efficiency at home, and manufacturers are tempted to indicate a higher efficiency for the LEDs they use: serious errors in LED fixture specifications are common.

We have heard claims from many experienced reef keepers that LED lighting has not yet matured for aquarium use. We believe that this notion is caused by two main shortcomings of the existing fixtures: spectrum deficiency and poor color mixing which results in multicolored blotches and shades. (Normal know as disco effect and shadow)

[sherman]

The first problem is relatively easy to address by combination of  LEDs of all spectrum in order to cover the desired spectrum fully (in our previous article we have shown that violet spectrum plays an important role but is neglected or poorly represented by most manufacturers, mainly due to the high cost).

[sherman]

The second problem requires a complex of efforts to achieve an acceptable solution. Some people spot these color blotches at once, while others may require time to notice the problem. Still, it is very important to eliminate this effect, as it affects our perception of the reef tank directly.

[sherman]

Challenge number one: Getting rid of color shadows (disco effect) 
The so-called disco effect is witnessed when a tank is lighted by several power LEDs (especially multi-colored) which are positioned on a certain distance between each other. Each LED is a point light source, resembling a small sun. When illuminated by several LEDs from different directions, every object produces multiple shadows. When these LEDs are of different color, these shades look colored like disco lights.
The only way to overcome this effect is to place LED crystals as closely as possible. We cannot place all the LEDs of the fixture in one place, as we want to light up the whole surface more-or-less uniformly. Therefore, we need to gather all differently colored LEDs in groups, in the required proportion. These groups can then be evenly spread over tank’s surface

[sherman]

One option is to collect a group of LED crystals in a small space by creating a custom COB (Chips On Board) with many LED crystals closely packed in one case. A COB is a matrix of individual LED crystals attached to a common platform and covered by a protective compound (usually silicone). They have several advantages over SMD LEDs and are becoming quite widespread for general lighting use. Indeed, they are easy to mount, provide a relatively low cost per emitted light, and are undemanding to optics – they are, indeed, very good when you need lots of light from an inexpensive fixture. COBs can be very good for the lighting of roads, warehouses, or other large premises.

[sherman]

However we believe that COBs are not the best choice for a reef tank. Three factors are most important when building a reef fixture: spectrum, efficiency, and optics
None of the top ranking manufacturers, neither Cree nor Philips Luxeon, cover the whole spectral range required for reef lighting. Cree, for example, does not make violet (“true actinic”), or deep red LEDs, Philips Luxeon also doesn’t cover the violet range. Of course, there are smaller companies which make a wider range of LEDs, and their crystals cover the required spectrum.

[sherman]

Our second requirement – efficiency – would suffer. Theoretically, crystals from different manufacturers can be combined in one COB package, but this can be practically impossible for similar crystals, made by different brands. As a result, the COB will have to use crystals with significantly lower efficiency, compared with world’s leading manufacturers. Our calculations based on the datasheets from leading Taiwanese manufacturers :
Epistar/Epileds, show that there can be as much as 54% difference in efficiency compared with best individual LEDs from top manufacturers.
Even Cree declares about 34% less efficiency for their best COBs today, compared with individual LEDs

[sherman]

A shorter thermal path in COBs is usually listed as an advantage, however, this is not as simple as it seems. As an example, let us consider Cree’s newest COBs. As one of World’s top leaders in LED technologies, this company pays closest attention to improving the thermal characteristics of their products. According to the datasheets, thermal resistance of Cree’s COBs varies from 2.5°C/W for CXA1507 to an impressive 0.8°C/W for CXA2530. Seems to be pretty good, huh? Only until we notice that, say, CXA2530 may consume up to 50W of power
According to the datasheet (see the picture below), this will add a significant 40С to junction temperature above the temperature of the LED case. For CXA2530, with 61W power consumption, junction temperature will be even higher

[sherman]

See the datasheet

 

[sherman]

Note that the thermal resistance of COBs from Asian manufacturers is usually even higher than those of Cree. This can result in serious overheating of the crystal (even though the heatsink temperature may stay in the acceptable range).

[sherman]

A better approach would be, instead of COBs, to use carefully selected individual LED emitters (the ones with most efficient bins, to minimize heat generation), then mounted on an advanced MCPCB to provide the best heat dissipation from the crystal. This can yield less than a 10°C temperature difference between the junction and the MCPCB, and about 12°C between junction and heatsink

 

 

 

 

 

 

 

 

 

 

[sherman]

Therefore, we believe that the best approach to avoid the Disco effect is through the use of individual LEDs packed tightly on a custom MCPCB. This way the light beams from differently colored LEDs will mix up well, without forming color shadows. The smaller is the distance between individual LEDs, and the better it will be able to overcome the color shade effect.
The distance between individual LEDs, however, cannot be smaller than the size of secondary optics. Fortunately, compact and efficient LED lenses are available today. The best effect is achieved through the use of hybrid optics, combining a TIR lens and a special light diffusing material.

[sherman]

Tomorrow we look at the second issue. 

Battling the heat

[sherman]

Challenge number two: Battling the heat 
Modern LEDs are quite efficient at the conversion of electric energy into light. However, even the most efficient LEDs today waste about half of the consumed energy in the form of heat. Since the crystal is quite small (a power LED’s crystal usually has the surface of only 1-2mm2 or 0.0015-0.003in2), its generated heat density is quite large. In fact, it is about 30 times as large as the heat density through the soleplate of a household iron!

[sherman]

Heat removal from LEDs is very important since their lifespan and performance depend on the operating temperature. When the temperature of the crystal increases from 25 to 100C the performance of the most heat tolerant LEDs (Luxeon Rebel ES royal blue) decreases by 10%. In the same time, less heat tolerant LEDs may lose up to 75% of their efficiency (Luxeon Rebel amber). This drop of efficiency is even more pronounced in  LEDs from not well known manufacturers 

[sherman]

Unfortunately, commonly known technologies today do not permit mounting the LED crystal directly on the heatsink. The crystal requires a special package, and its heat transferring capability is described by a parameter called thermal resistance, usually measured in the degrees of temperature increase per watt of generated heat (C/W). The thermal resistance of the best individual LEDs is about 2.5C/W today.

[sherman]

Let us consider the influence of thermal resistance on crystal temperature. Suppose we have two 3W LEDs of the latest-generation. Let us consider, for example, a violet LED Semileds N35L-U-A with a thermal resistance of 4.4C/W, and a Cree XP-E2 green LED with 15C/W thermal resistance. When operating at 700mA, the respective voltage drop will be, approximately, 3.3V and 3.5V, and hence their overall power consumption will be around 2.31W and 2.45W respectively. In the first case, the LED package will add 2.31*4.4=10.16С to crystal temperature, and in the second case the temperature difference is much higher: 2.45*15=36.75С.
We need to supply current to the LED and the conductors should be somehow isolated from the heatsink. Therefore, the LED emitter is first installed onto a special PCB, which is then mounted on the heatsink. Thermal resistance of least expensive PCBs can be as high as 60C/W or more (see the datasheet for Cree XLamp _PCB_Thermal.pdf). This, however, is quite unacceptable, and therefore special metal PCBs (МСРСВs) have been developed. 

[sherman]

MCPCB cross-sectional geometry
 

 

[sherman]

Since we are planning to pack the LEDs tightly, in order to avoid the “Disco” effect, we shall assume a larger figure, towards 5C/W. This will add another 11.5С to 12.25С to the temperature of our crystals. These calculations are valid for a traditional MCPCB, but we can improve the situation significantly by using an advanced MCPCB like SinkPAD or similar.

[sherman]

The МСРСВ is not a part of the heatsink and is attached to it through a layer of heat conducting compound, which is required to fill in the gaps resulting from manufacturing imperfections of the attaching surfaces. The compound’s thermal resistance depends on material but, in any case, the thinner is the layer, the less will be the resistance. As a rough approximation, for a high-quality compound it is around 2C/W. This will add 4.6С and 5С respectively to our crystals.
Once the heat from the LED has reached the heatsink, it cannot be instantly transferred to the environment. A heatsink is usually selected in such a way that its average temperature would not exceed 50C when the surrounding air temperature is 25C. Heatsink temperature at the contact point with the МСРСВ will be around 60C. Thus, the crystals temperature will be 60+10.1+11.5+4.5=86.1С and 60+37.5+12.25+5=114.75С respectively. Note that if we attach such LEDs to a cheap FR-4 PCB with thermal resistance about 60C/W, crystal temperature may reach 252С. Since LEDs are attached with a solder which melts at about 217С, the LED would probably unsolder due to its own heat!

[sherman]

This explain why some led switch off or blink after running

[sherman]

Many  manufacturers avoid indicating this parameter in their datasheets, just because the figure is even worse. Also note that real-life temperatures will be worse than we calculated above, because of mounting imperfections of the МСРСВ, an uneven layer of the heat conductive compound, etc.

[sherman]

As we have seen, there are two main obstacles to removing excess heat from the LED crystal: the LED package and the MCPCB. Manufacturers are constantly working to overcome the first obstacle and the best LEDs on the market are using packages with reduced thermal resistance. Overcoming the second obstacle is more of a challenge since, until recently, there was no technology that would allow engineers to get rid of the dielectric layer between the LED thermal pad and the МСРСВ metal.