本帖最后由 ambiva 于 2013-11-19 20:13 编辑

The Meaning of life 寿命的含义 —— 这个月 Lux Live 杂志中不错的文章http://www.luxmagazine.co.uk/?utm_source=iContact&utm_medium=email&utm_campaign=Robert%20Bain&utm_content=#!/issue/28/page/142
原作者 Mike Simpson (飞利浦照明英国技术总监，IEC LED规范起草小组成员)Mike Simpson 阐述了IEC对于LED产品寿命的新定义，以及如何帮助我们更好地理解LED性能。
有空我会翻译一部分，先贴原文。注意红字The life of any lighting product has always been the source of a certain amount of mystery, and the advent of LEDs has turned the mystery into confusion. Through its work developing standards for all LED products, the IEC has realised that a clear, concise explanation of LED life is necessary, and it is now developing this as part of its ongoing standardisation work. This article is based on the IEC’s latest work to help advance understanding of LED lifetimes.
A new approachBefore looking at LEDs,let’s reflect on the approach to establishing life for traditional lightsources. There are two important characteristics: a gradual reduction in lightoutput with time and lumen depreciation, and a known rate at which physicalfailures occur. These are often shown by a graph with a line curving downward.Data for these curvescomes from measuring the characteristics of batches of lamps. They representmedian values from the batch measured so there will always be some lamps thatperform better and some that perform worse. In fact, the curves are the probabilitythat at least half of the lamps in the batch will be above a particular valueat a given time. With traditional lamps we have taken the survival curve as thebasis of the claimed life and given lumen depreciation as extra information.

With LEDs we take theopposite approach. The life of an LED product is based on its lumendepreciation and the survival curve is given as extra information. Confusionhas been caused in the past by the LxFy term with which manufacturers tried tocombine lumen depreciation and survival in a single figure. You might haveseen, for example, L70F10 as a quoted lifetime where L70 is the time it takeslight to depreciate to 70 per cent and F10 represents the point at which 10 per cent of the lamps have failed. This isnot possible because both are unlikely to occur together. If you take the timeto a light depreciation of 70 per cent then the failures at that time areunlikely to be exactly 10 per cent. You can only specify one factor and theother has to follow from the data.

So let’s try to simplify things. The useful life of an LED productwill be the time to which a given light output remains (Lx) for a givenpercentage of the batch measured (By). So if I want to know how long it will bebefore the light has depreciated to 70 per cent for half of the measured batch,then we get L70B50 .Or we might decide that we actually only want 10 per centof the measured batch to be below 70 per cent light output so we get L70B10.If we compare this to the way we measure traditional lamps then itis most common to have 50 per cent above and 50 per cent below the quotedvalue, so the median is specified. To create a simple LED life definitionthat is equivalent to the way we deal with traditional lamps we use L70B50. Ifwe call this the median useful life, B50 is assumed and we just call it L70.For most applications we expect this to be the basic definition oflife in literature and packaging. This could become the standard way to definelife for LED products. However, in some applications – streetlighting, forexample – Lx may have to be achieved for more than half of the batch, in whichcase the useful life can be used with the y specified.Fade away or burn out?As with traditional lamps, extra information on abruptfailures is useful when evaluating a product’s performance. An abrupt failureis when the light output completely stops, and may be the result of a failureof a LED module, LED driver or even the complete LED luminaire. Again weare talking about statistics and probability. So the time to abrupt failure isthe length of time to a given percentage y of a batch failing to produce anylight. This is given by the term Cy.The time to abrupt failureis independent of the useful life, but it would be useful to have life andfailure values linked. At the median useful life (which is determined bygradual loss of light) we can have the abrupt failure value (which is determinedby complete loss of light). The abrupt failure value is the percentage of LED products that nolonger operate at the median useful life.When designing installations, professionals allow for loss oflight output, but it is usual to assume that any failed lamps will be replacedand so they are not included in the design calculation. In streetlighting, forexample, if 10 per cent of the lanterns fail, the average light leveldoesn’t fall by 10 per cent. Ten per cent of the installation will not be illuminatedat all. So multiplying light depreciation by abrupt failures doesn’t have anypractical use. However, to complete the picture we can introduce a furtherdefinition called the combined failure value which is the percentage of LEDproducts that have failed because of light loss and abrupt failure at therated median life.I have therefore introduced five definitions:
[*]· Useful life (LxBy) The time after which a given light outputremains for a given percentage of a batch
[*]· Time to abrupt failure (Cy) The time after which a givenpercentage of products in a batch fail
[*]· Median useful life (Lx) The time after which a given light outputremains for half of a batch
[*]· Abruptfailurevalue(AFV) The percentage of a batch that has failedcompletely by the end of the useful life
[*]· Combined failure value (CFV) The percentage of a batch that failsthrough either decline in light output or total failure.
The first two terms are general and can have any value of y,whereas the last three are specific conditions of the first two. These threeare designed to be the most commonly used life definitions.Who is responsible for providing this data? In the past the lampmanufacturer would measure production samples over many years and produce dataon light loss and abrupt failures. Generally these were unaffected when thelamp was fitted into a luminaire. Today we know that the performance of an LEDmodule can change when installed in a LED luminaire. And each luminaire designcan affect the module performance in a different way. It is not sufficient forthe luminaire designer to simply quote the module performance as representativeof the way a luminaire will perform. The final supplier of the equipment has tobe responsible for supplying life information for its product.In summary, compared with the way things used to be, LED life isdetermined from light loss not physical failure, and light loss is a result ofthe complete system not just the light source.LED performance standards from the IEC are now in the final stagesof drafting and should be published in 2014. With unambiguous definitions theywill enable better understanding of how to specify LED performance in future.