/ Home & Energy

Are your LED light bulbs burning out too soon?

LED light bulb

LED light bulbs have a tendency to make grand claims about their lifespan – it’s common to see manufacturers promise bulbs will last 25,000 or even 50,000 hours. But we’ve found many failing well before this.

One of the attractions of LED light bulbs is that they’re supposed to last a long time. And so if you’re shelling out for these bulbs – typically more expensive than other types of light bulb – you’ll want to be sure that they’ll live up to those claims.

But our tests show that not only do many LED light bulbs stop working before the end of their promised lifespan, some don’t even reach the soon-to-be-implemented EU minimum lifespan of 6,000 hours. We discovered bulbs from both Ikea and TCP that failed to reach the 6,000 hour mark for the majority of samples we tested.

Ikea bulb among failures

In the tests – which were carried out by Which? and our European partner organisations – we took five samples each of 46 different bulbs. The bulbs were switched on for two hours and 45 minutes, then switched off for 15 minutes, in a continuous cycle until they burned out.

Five different bulbs stopped working before the 6,000 hour mark for the majority of samples we tested, though the TCP and Ikea bulbs were the only ones which were sold in the UK. Both have since been discontinued.

New EU regulations which will come in from 1 March 2014 say that 90% of any batch of LED light bulbs should last at least 6,000 hours.

Another five bulbs stopped working before the 10,000 hour mark for the majority of samples we tested, despite claiming lifespans of at least 25,000 hours. None of these bulbs were sold in the UK.

In total, 66 of the 230 samples we tested failed before the 10,000 hour mark, though they all claimed they would last at least 15,000 hours.

Has your bulb burned out early?

Ikea said the bulb had passed its own tests and those in a third-party lab. It’s looking into why the bulb failed our test and has removed it from sale in countries where it was still available.

TCP said it was already aware of the problem with this bulb and withdrew it from sale when they discovered the problem. TCP added that it no longer deals with the supplier of that particular bulb and now make their LED bulbs in-house.

We’re in the process of testing the life span of many more LED bulbs, and we’ll update you if we find others that burn out prematurely. But we also want to hear from you – have you bought bulbs that haven’t lasted as long as they should?


A Luxeon data sheet http://www.luxeonstar.com/assets/downloads/rd07.pdf shows the effect on life of increasing junction temperatures – where the LED chip is mounted. In fact it seems that 25 deg C is the normal ambient for giving performance characteristics.

LEDs have life defined in two ways – the point at which a certain % of them fail (usually 10% and 50%) and the point at which the light output has fallen to an “unacceptable” level – usually 70%. This data sheets shows combinations of these two. although for coloured LEDs the principals are for LEDs in general.

Putting your LEDs in a ceiling below a loft seemingly increases the ambient by 10 deg C and as the graphs show this can halve their effective life. However, that means the LED was working right on its temperature limit in the first place in a normal ambient. It illustrates the need to use good quality LEDs that are not overstressed under normal operation and have good thermal management. Nevertheless, even halving the life for a decent LED still gives of the order of
20 – 30000hours (for most of us 20 years life) – so it does not explain the rubbish performance of the others.


My understanding is that the junction temperature of a LED relates to the semiconductor and not the interface with the mounting/heatsink, which will be considerably cooler.

I think it is unhelpful to rate the lifetime of LEDs or any other lighting in terms of years because this will vary according to use. Old fashioned bulbs were generally regarded as having an expected lifetime of 1000 hours, and if anyone wanted to work out what that corresponded to in years they could do that on the basis of their planned use.

Is there evidence that different brands of LEDs perform significantly better than others? I have seen evidence on Which? Conversation and elsewhere that even the well known brands are failing prematurely.


The junction temperature is on the LED chip and cannot easily be measured directly. It is normally calculated by measuring the temperature on the thermal mounting pad and taking account of its thermal resistance. It is the temperature that determines the performance of the light-emitting chip (but not, of course, of the electronics used to operate it which will cause failure if they get too hot).

I’m not sure how many people relate thousands of hours to how long in months or years they might expect a lamp to last in normal use. Generally it is taken that in the home 1000 hours is about a year’s use. But it will depend where it is installed of course. The point about LEDs is tthat 20000 hours say might be 20 years to put it in perspective; but that takes no account of the life of the electronics used to run it – and their life, even with a good LED chip, might be much less..


I appreciate that the junction temperature is difficult to measure, but the lighting industry should have chosen a different term for the temperature of the mounting pad rather than hijacked an established expression. There are other examples of confusion being introduced in this way.

Expression of life of lamps in years became commonplace with CFL lighting, choosing some arbitrary daily use. Ask the public and I’m sure that they would prefer lifetime rated in terms of thousands of hours rather than based on years of operation based on 2.7 or 3.7 hours’ use per day (I cannot remember which).

It seems clear to me that current designs of LED lighting are overheating both the LEDs and the control electronics – hence we are seeing premature failures with both cheap and expensive brands. As we have agreed the best answer is to have separate LEDs and electronics, as in non-domestic lighting. I would have expected the well known brands not to produce direct replacements for existing lighting, but when there is a profit to be made, that takes priority.

Is there any brand of LED lamp (designed to replace halogen or old fashioned incandescent bulb) that can be relied on to achieve the claimed lifetime? I think not – but am prepared to be proved wrong.

Phil - NSW Australia says:
17 January 2016

Linear flouresent tube lamps seem to soldier on in all kinds of heat but are terrible when cold until they warm up , and not as efficient as LED and use nasty heavy metals. We are having major issues with led’s overheating in Australia and failing. I am pushing suppliers for standards to advise consumers and installers the max ambient temperature and airflow recommedned to achieve the quoted lifespan .This allows for fitting design and installation locations to cater better to allow for the temperature rise at the LED itself and the electronics contained within the LED lamp or housing.


With linear fluorescent lamps, the control electronics are not in a confined space and separate from the lamp, so that overheating is not a problem. The amount of mercury is small and the tubes generally have a long life. As you say, fluorescent lamps, including CFLs, don’t do well in low temperatures.

With LED and CFL bulbs it can a significant difference if they are operated cap-down or cap-up, the latter more likely to cook the electronics. I think you are right that we do need to be given advice about how to use lamps to reduce overheating. The real answer is to have the lamp separate from the control electronics, just like fixtures for fluorescent tubes.


Phil -if the Australian external ambient temperature is much higher than countries like the UK and in summer it is . Then exporters should increase the aluminum heat-sinks on the bulbs to cope with this . If it is a case of the driver section overheating -ie – capacitors drying up/ active components overheating then a re-design is required or a higher standard -ie- industrial quality used . I cannot believe that any industrial concern in Australia or anywhere else would put up with low quality lighting requiring frequent replacement . I think we are back to – built to a cost , do you have large imports from China as you are near to it geographically ?


wavechange, the junction temperature is the key temperature in the performance of the LED chip and is therefore used by the chip manufacturers in their data sheets. It is not difficult to measure in the industry. So lighting manufacturers should make this measurement – the chip manufacturer gives the information on how to do this – to ensure under normal operating conditions the maximum junction temperature is not exceeded in their product. They will also need to check other electronic components of course.