/ Home & Energy

A Nobel Prize for LED light bulbs, but do they get your vote?

LED light bulb

Earlier this week the scientists behind LED light bulbs were awarded the 2014 Nobel prize for physics. Unfortunately, you’ve shared stories of LEDs burning out too soon. Are LED bulbs getting any better?

We lit up your frustrations when we last spoke about LED light bulbs. More than 300 comments were made, most of them about their poor performance and your dissatisfaction with them not living up to their lifespan claims.

I’ve had many problems in the past with LED light bulbs blowing too quickly. Only this morning I noticed another bulb in my kitchen had gone – I’d only replaced it three months ago. Ivan was also blown away by the short lifespan of his bulbs:

‘I bought six LED SES candle light bulbs from Homebase. Two failed within two weeks.’

While Raj had slightly better luck, his bulbs still weren’t up to scratch:

‘I bought 24 off Aurora 6W LED lamps at £11 each. After two years I’ve had to replace four of these. Not

Who turned out the lights?

As LED bulbs continue to improve, there have been ideas as to what else could cause their underwhelming performance. Brian started to wonder whether maybe it was the fittings in his kitchen, rather than a problem with his bulbs:

‘After only six weeks regular use four have failed. Swapping round has got one back working, but only temporarily. Not sure the lamp holders are 100% either as two went off together. Tried all the connections and they seem fine so at a loss as to what is really the problem.’

The light at the end of the tunnel

We do have some good news for you though. Our research shows LED light bulbs are getting better. The failure rate in newer bulbs is much lower than for those produced a year or two before.

We’ve tested 410 LED light bulb samples for 10,000 hours or more, and 75 of those (18%) failed within 10,000 hours, even though they all claim to last much longer. And 69 out of the 185 bulbs (37%) we examined at the 15,000 hour mark had failed by that point. Again, almost all of them claim to last longer than this. So, although there are advances, there’s still room for improvement.

What’s wrong with incandescent bulbs?

Some of our commenters are still longing for the good old days. John Ward pined after the 60s:

‘I’ve never had a better kitchen light than the long fluorescent tube that I had in my first flat in 1968!’

While Derrick said:

‘Glad I stocked up with “old fashioned” bulbs before they disappeared!’

But it’s not all doom and gloom, Nick C has seen the light:

‘I too have replaced almost all the lamps throughout the house (and outside) with LED with great results. I struggle to understand why people are determined to be so negative and backward-looking about any progress that is made’

Are you still having problems with LED light bulbs? Or have you seen similar improvements as in our new test results?


The science behind the blue LED chip that made white LEDs possible is what won the Nobel prize. Totally justified. What people then go on to do is to manufacture products based on this science, and the control circuits that are necessary to operate them This is where the problems have arisen with some manufacturers. You cannot condemn an innovation because of this.

I remember when high pressure sodium lamps (orange-white street lights) were first introduced – good science but containing such a corrosive hot substance in a translucent arc tube with electrical contacts each end proved a very demanding engineering challenge – overcome, but only when products were tried in bulk. Such is progress. Production issues arose, but were solved, with the whiter metal halide lamps. We now have these as established reliable light sources.

The well-known major manufacturers are the better bet for LED lamps and, where needed, matching drivers. Use unknown makes that may be cheaper at your own risk.

I posted an off-topic comment about the Nobel Prize for Physics last week, in the vague hope that it might elicit a new Conversation about LEDs. Thanks Alex. 🙂

Incandescent bulbs generate heat. That did not affect the life of the lamp. When CFL lamps were introduced to save energy, many experienced short lamp life. Although the amount of heat produced by the lamp and the electronics is small compared with an old fashioned bulb, it can mean that the electronic components run at a far higher temperature than is desirable. Not surprisingly, CFLs tend to have a much shorter life than modern fluorescent strip lighting. They work in the same way but the control electronics does not get nearly as hot in strip lighting. The problem with CFLs is that they are designed to be plug-in replacements for old fashioned bulbs.

The problem has repeated itself with LED lamps, where the electronics and LEDs are close together and can overheat. In a downlighter, the electronics is crammed into a tiny space.

The answer is, in my view, to design light fixtures in which neither the LED chips or the electronics are close enough to cause overheating. They could be much more durable. Look at LED lighting in offices and public buildings, in street lights, and elsewhere. Most of what I have seen has been reliable.

Though I must have a dozen or more LED torches (which are fantastic) I think I will wait for LED light fixtures to replace bulb replacements to light my house.

wavechange, commercial and streelighting fixtures have been generally carefully designed, usually with separate electronic drivers and LEDs with good heatsinks, to take full advantage of the efficient light production of LEDs. But, of course, that costs money – but reliable long life is worth paying for. For some reason many do not want to spend what it takes to get decent domestic lighting – probably because they lack the necessary knowledge to tell good from bad.

I have a lot of CFLs around the house in open fixtures – table lamps, ceiling lights – and outside in enclosed fixtures. They arer all GE, Osram or Philips and I have had good life from them. The problem is many do not know how to use them – stick too high a wattage in small enclosed lights and . There should be much more education in how to use them – although many will ignore the instructions that accompany them. Even a 20W quality CFL should not be a heat problem – unless you choose a cheap make with rubbish electronics.

The problem is the market is flooded with poor quality CFLs and LEDs and in responding to correct energy saving initiatives we lack the information to discriminate between good and bad. The safer way, as I have said before, is to go for the well known major manufacturers – maybe not the cheapest, maybe not always perfection, but safer than the alternatives.

Malcolm – We can certainly agree on the importance of avoiding overheating of electronics. I am less convinced that the best known brands of CFLs are much better than supermarket brands these days, though I would not buy unbranded products in case they were a fire risk.

I am fed-up with manufacturers claiming some phenomenal life for LED lamps and then guaranteeing them for a year or two. I see that 5 year warranties are available in the UK and 10 year in the US. That might convince me to buy bulb replacements, but what I would prefer is new light fixtures designed for use with LEDs.

Keen radio listeners who are interested in LED lighting might like to look at a Conversation about interference of DAB and FM radio. Search for: “The energy-saving LED bulb that switched off the radio”

12V LED lighting seems to be a bigger problem than mains-powered LEDs. In a recent test of LED lamps, Which? did not find interference problems with DAB radios, possibly because they have not looked at some of the discount products available on the web.

I purchased 6 led bulbs for my kitchen.After 4 days one had stopped working and another had 2 leds in a second had failed.Also in the same purchase I bought an ES led bulb for hy hall.Half the bulb (banks of leds) went out after 45 mins.Not impressed.

I bulk purchased 45 x GU10 4W replacements some three years ago. 23 were for my son’s flat and the remainder in my home. The reasons for replacing were twofold: 50W GU10 halogens have the lifespan of a mayfly and and the obvious massive energy saving of LEDs.

I purchased these for around £4 each from the Internet – very cheap for the time. I have only had ONE failure and that was one of four that are fitted outdoors in IP66 downlights in the sofits of a new extension.

I also have 4 x RO80, 4 x RO63, 2 x golf ball and three ‘conventional’ shape LEDs. They too are still working perfectly. And I have had no RFi problems with any of our LEDs.

So I’d say that LEDs are fantastic. Reliable, truly energy saving, have a better light output than compact fluorescents and the new types have light outputs that surpass tungsten lamps.

I can only guess that there are some rogue makes out there that are rubbish.

terfar, rogue makes like Philips and Osram?

I also have had good results from Long Life Lamp Company, but not from those two rogue brands above! Yet Which? still has an Osram 10w bulb in its best buys, despite the high temperatures they attain.

Sadly, Which? reports are no longer comprehensive enough, and I rely more on Amazon reviews. Take a look at this, from a former BBC engineer:


Alex, I may have misunderstood your reply . If an LED bulb proves to be running too hot ( and therefore would be short lived) under normal use, or is unsafe, it should be given a zero rating whatever its other features might be. I take it this would be Which?’s view?

I agree with Malcolm r, everything else is irrelevant if the bulb is going to fail prematurely. A weighted average is inappropriate in a case such as this. Giving something a high score when it is sure to fail is just not on, and I am surprised at Which? for doing this.

Imagine a hospital rated well for care, situation, facilities, etc, but with a deadly virus that is going to kill all the patients. Would you go there? I wouldn’t!

Thanks Alex. Just to clarify, do manufacturers of domestic LED bulbs generally give a point on the housing (tc) at which temperature should be measured, and the maximum allowable temperature, and do you test this?

Do halogen bulbs have complex electronics in them, though? I thought they were just filaments in a halogen gas.

2000 hours is less than 3 months. A considerably cheaper CFL could be expected to last far longer than that. We are talking about expensive LED bulbs costing £12 to £25 each. Without a long life, they are pointless as replacements for CFLs. They can’t even match the luminosity of a 25w CFL.

I am not saying all LED bulbs are useless. We have 5 Long Life Lamp Company GU10s in the kitchen, 2 as bedside lights and 3 in the bathroom, and all are lasting for years. The problem is with the larger bulbs that are not even as bright as a 100w bulb but need to fit an existing light fitting that might cost a lot to replace.

If you want longer term tests on the durability of the Osrams, look no further than Richard’s Amazon reviews. He knows what he is talking about. All of his Osram bulbs failed early.

This is why I’m sticking with CFLs, Mike. I keep mine well ventilated and they have been very dependable. I’m sure that LEDs will be a better choice in future but I will wait until they are reliable and cheaper.

And cooler and brighter too. 🙂

It’s interesting to compare the above posts by BJ and Terfar. No doubt there will be many whose experience is between these extremes.

I have been following the use of LED lighting by those living off-grid. That’s people who generate their own power for their homes, and users of RVs and boats. They were early adopters because they are reliant on storing power in batteries – low energy lighting was worth paying for, even when when LEDs were much more expensive than they are nowadays. Many LEDs have been killed by voltage spikes and over-voltage, so there are specialist companies that sell LEDs that can cope with such use.

UK mains voltage is reasonably stable but voltage spikes cannot be avoided. Well designed LED lamp drivers and mains voltage lamps will have spike protection (e.g. metal oxide varistors) but that’s not something that is likely to be in the product information. We might be able to learn something from those who live off-grid.

Dirty power may be the cause of high failure rate. I found 50W mains halogen GU10 lamps absolutely terrible for failing which (as I said above) was one reason that persuaded me to invest in LED.

There seem to be many different designs of GU10 LEDs: some have what looks like a single large reflective cone with a single LED in the centre whilst others have a cluster of three or four small reflective cones with smaller LEDs. There also seems to be a discrepancy in the overall size between the various GU10s, possibly that has something to do with cooling.

Of all the LED types that I have installed, it is the RO80s that give by far the best and brightest light. We have four swivel-eye downlights fitted with 6.5W LED RO80s in our kitchen and it really is brilliantly lit. You still need quite a concentration of GU10s to get anywhere near that brightness because none seem to have beam widths wider than 38deg and many are much narrower.

Try these GU10s:


120 degree beam width and very bright.

it is possible that the mains voltage is too high. The voltage in most of Europe is 220v. In the UK it should be 230v but is often higher. The voltage in my house is usually over 240v. This may shorten the life of LEDs in the UK!

The mains voltage in the UK is 230V, with a tolerance of +10% and -6%, so a voltage of 253V would be permissible. When we moved from 240V, the tolerance just became broader, allowing harmonisation of mains voltage in Europe.

Many electrical products are tolerant about mains voltage these days. Incandescent lighting is an exception and the lifetime of lamps can vary significantly. I do not know what voltage range mains LEDs and 12V LEDs will run on, but you could be right, Brian.

There LEDs available from certain suppliers that will operate from 10-30V. These are popular for off-grid use where a significant variation in supply voltage is common.

This could be a factor. Try changing incandescent lamps on a motorcycle (with the aim of brighter warning lights) and the poorly regulated DC will play havoc with the life of LEDs. Not that it’s too good with incandescents!

It could be mains voltage spikes, but that would not be the cause of some bulbs running at far too high a temperature, even when well ventilated. Sudden failure of cool running bulbs, maybe, but not repeated failures of hot ones.

I agree, Mike. If the surface temperature of the casing is too hot to touch, the electronic components inside will be even hotter and even good quality components cannot be expected to be reliable when operated under these conditions.

I wish I knew how Which? carries out product testing. It seems obvious that this should include surface temperatures of electrical goods, which could help identify products that might fail prematurely.

I just measured the external temperature on the Which? recommended 10w Osram and it is over 70 degrees Celsius after being on for about half an hour.

It suggests that there is a discrepancy with the lamps provided or Which? test isn’t testing the lamps the same way. The temperature you are seeing is way too high: I’d be worried about them being a fire hazard if the reach 70 deg after 30 minutes. Theoretically, being just 10 W, the heat should be just a few degrees above ambient. Dissipating that much heat suggests that efficiency must be much poorer than it should be.

There’s possibilities that Osram source their lamps from different manufacturers and they have been slack with QA, or maybe there are fake Osram LEDs in circulation. Something is certainly not right.

Most electronic components are quite happy up to 85 deg C, some more. Which should be using thermocouples on those sensitive components to check whether they are running at an acceptable temperatrure.

There is some information here, indicating a max casing temperature of 55C:


It says the LEDs would be considerably hotter than the case temperature. Also, dimming as the temperature rises is mentioned, which is consistent with my findings. If I leave these Osrams on all evening, the light level is considerably lower at bedtime than at 7pm.

I have bought some Auraglow 12W bulbs which have a substantial heat sink and will measure their temperature later for comparison.

MIke, this link shows the LED (chip) temperature not, I suspect, the operating electronics. A specific Philips LED lamp I found was rated at 45000 hours life for a case temperature of 80 deg C (life not being “death” but when 50% of the lamps have a light output 70% of initial). At a case temperature of 90 deg C the life was derated to 25000 hours.
Which? should, when testing lamps with integral electronics, ascertain the recommended maximum case temperature (a spot should be marked where temperature measurement should take place) and ensure when tested it is within the manufacturers stated value (tc). I wonder how many manufacturers quote (or whether the iffy ones even know) this value? Do Which? test for tc?

Judging by how dim a lot of the larger rated bulbs are, malcolm, I don’t think they could stand even a 70% degradation!

Good post!

Chris Benton says:
18 October 2014

I replaced 14 halogen bulbs (50W GU10) a year ago with the LED equivalent. They have always been perfectly reliable and just as bright as the halogen originals.They are in ceiling fittings with ventilation designed to dissipate 50W so the control electronics must be quite well cooled.

Any lamps that I buy from now on will definitely be LED.

Wow, 14 times 50W! That was heating, not illumination!

You didn’t say what make they were. People need that information about reliability. There seem to be large numbers of buyers of obscure and premier brands having troubles, so another positive would be helpful.

I purchased 4 L1 LED light bulbs recently and after just 1 day one of the bulbs had failed. I’m fearful for the reliability of the rest of them and will probably just go back to the original CFL ones.

Hi everyone,

I am the MD of Light Rabbit Limited which is a major UK LED lighting supplier. I regularly go through forums and comments to see what issues and concerns people have with LED lighting and note the above comments which are useful.

We at Light Rabbit are in favour of seeing regulations being brought in to standardise the marketing and testing of LED lighting products. There are wide ranging claims on lifespan and brightness and most suppliers rely upon the information provided by the manufacturers of their lamps for this. Our lighting is sourced from the most trusted factories and we back up our claims with a 45 day no quibble returns policy and 5 year warranty but I still do not believe that this goes far enough.

Ideally, all lighting suppliers should be made to undergo testing with an unbiased testing house. The Energy Saving Trust have a scheme that allows one to place their badge on products that pass their criteria; however, this is voluntary and the EST provide this service really to create an opportunity for brands to market the EST certificate rather than a consumer focussed test for safety and integrity.

Such testing should be standardised so that consumers can make like for like comparisons and know that such results have been independently verified. The Lighting Industry Association (which we are a member of) is attempting to do this but there is little incentive for all companies to participate unless it is made compulsory.

It would be great to have your thoughts on this as it is something I am pushing for.

Welcome David.

I have had a great deal of success with CFLs and having read a great deal of criticism of poor lifetime and radio interference problems caused by LED lighting, I have no intention of taking the risk for the time being. I also believe that domestic LED lighting has an inherent flaw in that the compact design means that electronic components and the LEDs themselves run too hot to ensure reliability. A good warranty might convince me that these products are worth trying and that would have to be at least 50% of the manufacturer’s claimed life.

I particularly applaud your suggestions regarding independent testing. We are seeing failure of lamps manufactured by (or for) well known brands. A warranty offered by an online retailer must, of course, cover return postage costs. We should not have to register our purchases if we have evidence of purchase.

David, there are no doubt good, and rubbish, sources of domestic LED-based lamps. I’d be interested to know your comments on how consumers can decide how to choose a reputable supplier from the many that abound – including the stores such as B&Q, John lewis, Homebase. Do they best stick with the known major brands – Philips, Osram, and so on?

The potential for reliable and well-performing LED lamps is good – street lighting for example appears to be using them successfully in a fairly hostile environment. But then these are mainstream LED makes with separate drivers. Perhaps this is where the domestic market fails, as it can with some CFLs – cheap electronics in an environment that is too hot?

I have the view that the well-known major manufacturers are better placed then many commercial test labs to examine the reliability and performance of their products, and to use the results to aid development. It is the smaller and less knowledgable, or less scrupulous, manufacturers and suppliers who need to have some sort of independent scrutiny – but unless they have a robust quality control system in place at the factory then such scrutiny is of little value.

A couple of examples of reliability information can be found at the following links simply to illustrate the major manufacturers’ approach.



I think we have enough evidence already that major manufacturers are putting failing products out into the market, don’t we? Philips and Osram spring to mind.

If you buy from John Lewis you buy Philips. B&Q sell Osram.

I am currently testing about £150 worth of Auraglow bulbs (12w) on the recommendation of an electronics expert. I have also used Long Life Lamp Company low wattage (4–5w) GU10s for several years with no failures, but then, these stay cool, unlike their more powerful cousins.

I have no affiliation with any company, all I know is that three Philips bulbs I owned failed early and ran extremely hot, one hot enough to crack the translucent cover. I now have three more on test, including one of their latest, and will report back.

I find the Which? recommendations irresponsible. These bulbs are not cheap, and should be backed up with proper n0,000 hours or 10 year warranties. But then, what do Which? care any more about reliability?

Hi Mike,
Any mass produced product is prone to failures. Some of these failures may be due to inherent design flaws and other failures may be due to the environment that the lamps are being used in such as in kitchens where humidity may be high. Manufacturers will typically test products for their typical use – it is impossible to legislate for every scenario.

One thing I have noticed is that B&Q et al seem to be well behind the curve when it comes to LED lighting technology. I notice that they use very old type lamps for their lighting displays (which isn’t great when it comes to improving the public perception of LED lighting) and many of the lamps they sell are outdated. The reason for this is that big stores buy in big quantities and less able to update their stock to better technologies.

In the last year there have been massive improvements in the technology which include better colour render, heat management and tolerance to environment. The latest lamps can produce light which is so similar to halogens that one would find it very difficult to identify an LED from a halogen. These are all the factors which consumers need to know about.

[This comment has been edited to align with our community guidelines. Thanks, mods]

I strongly support David Kennedy’s suggestion that LED lamps should be independently tested.

It is worth looking at how Which? conducts its testing. Which? does not accept offers of products for testing, but buys them, as you or I would do. That ensures that the products tested have not been selected or prepared to perform well. Any independent testing house needs to put its own system to ensure that the products that it tests are representative of what we buy online or in the shops.

Like Mike, I am not convinced that well known brands are necessarily of better quality, even if they cost more and claim longer life. Let’s see the longer warranties or long life claims are nothing more than marketing hype.

Hi, Lightrabbit. First, 950 or 1110 lumens are hardly a replacement for a 100w bulb. Second, “Lasts up to 30 years” is meaningless. All it means is that none of the bulbs last more than 30 years. Three Philips LED bulbs (of three; one a replacement directly from Philips) lasted less than 30 years! In fact, several months.

I will certainly look at your site though. What temperature does your “100w replacement” run at?

Two key issues with LEDs relate to their life, which has two forms – failure to operate, or the point at which the light output has fallen to 70% of the original. As life is often quoted from
25-50 000h, determining failure from “death” is likely to be difficult unless it is a very sub-standard LED and not, with decent LEDs, a meaningful measure. For light output as far as I can see accelerated testing still requires up to 6-8000 hours operation to predict the likely outcome. With the volume of lamps on the market such testing is fairly impractical for “independent” test houses. I would think the best we could hope for is to weed out the very-rubbish from the rest. Hence we probably need to rely on the extensive testing that major manufacturers undertake.

I think that Malcolm has addressed Mikes point on the definition of the life of an LED which many people don’t actually know about. It is something that I will discuss with our web content team to make the consumer more aware off.

For the general public, the perception is that a bulb just goes out when its useful life expires but this isn’t the case and should be explained better. The main fear is that it will just last a few months and then die. I think that failure rates should form part of the discussion surrounding the regularisation of testing and that a minimum standard should be achieved. It isn’t in the favour of LED lighting companies to have high failure rates as it diminishes the consumers faith in a brand (and technology) to supply quality products.

What it may prevent is the proliferation of low quality lamps being supplied directly from bad factories and unscrupulous importers which you will see hundreds of on eBay. I understand that this is a big problem in the USA and I understand that there were proposed restrictions placed on LED lighting imports of this type.

Mike: can you let me know which bulb you would like to know the colour temp of and I will obtain the photometric results.

LightRabbit, it wasn’t the colour temperature I was concerned about but the physical temperature, for, say, your highest rated B22 bulb.

If I can add, the case temperature gives an indication but it is the temperature that electronic components and LED chips that really matters. Even where the components can withstand a particular temperature in continuous use, heating and cooling can be very destructive and this is a factor that cannot adequately be modelled with accelerated testing.

Here is some further info on how heat is generated from a LED lamp.http://www.lightrabbit.co.uk/led-hot-vs-halogen-hot
The more efficient a lamp is the less heat it should generate. The internal drivers that convert the voltage from 230v to 12v are an obvious obstruction to efficiency and we are currently looking at a new wave of chips that can run on mains voltage. This is very exciting but a lot of testing to be done before we are satisfied that they are reliable enough.

Malcolm wrote: “I have the view that the well-known major manufacturers are better placed then many commercial test labs to examine the reliability and performance of their products, and to use the results to aid development. It is the smaller and less knowledgable, or less scrupulous, manufacturers and suppliers who need to have some sort of independent scrutiny – but unless they have a robust quality control system in place at the factory then such scrutiny is of little value.”

I expect that the Competition and Markets Authority would regard this as anti-competitive.

wavechange, I don’t follow this argument. The point I was making, having been involved with major manufacturers (and test houses), is that they have expertise, equipment and facilities far better than many test house and their business depends upon reliable products that perform as expected. Many of their customers are commercial and public organisations that do not tolerate poor quality. Independent test houses vary considerably in their expertise, knowledge and facilities – so don’t assume one bad, one good, that is not the case. But key to any mass-produced item is a good quality control procedure – to ensure consistency.

However as has been demonstrated throughout this conversation there are many 3rd rate or worse suppliers cashing in on the domestic energy-saving bandwagon selling to ignorant customers (as in no knowledge of the product).

I don’t see anything anti-competitive in this. 3rd party “validation” of anything and everything seems to be a recurring demand from some contributors – usually wholly impractical.

Malcolm – This Conversation is about domestic LED lighting. I am well aware that different standards apply to products sold for in other sectors, but that is not relevant to the discussion. I accept that independent test houses differ greatly and I wonder just how independent some of them are.

I am seeing increasing evidence that well known manufacturers are producing third rate products and I simply don’t accept that independent testing of products is impractical.

wavechange, I am well aware of the nature of this conversation. You presume, then, that the major manufacturers produce good products for commercial consumers and rubbish for domestic customers?

I think we need solid evidence of who produces 3rd rate products – separating out the conditions under which some might be mis-used. I would start by taking major manufacturers products and looking at two aspects – under normal (acceptable) conditions of use. Early failures and lumen depreciation. Both should take around 6000 hours – 9 months to a year. There are regimes for such accelerated testing. This would need a competent test house, and could well be funded by all European consumer organisations. It may well have been done, or be underway – perhaps Which can tell us? But to independently test every domestic LED lamp from every supplier into the UK (and Europe) would, I maintain, be impractical.

As I commented elsewhere on CFLs, my preferred option is to separate the lamp from the electronics. Commercial CFLs and LEDs generally do this. In the domestic market the lure of simply swapping an incandescent lamp for a CFL or an LED without changing anything else is understandable – it saves capital outlay. But it is not generally the best solution. Buying a dedicated light fitting properly designed for a lamp and separate driver is a far better solution – we simply need to be educated in that. But too late now.

Malcolm – I have not claimed that any large manufacturer is producing rubbish for consumers.

If we focus on manufacturers rather than suppliers then independent testing becomes easier. I would not expect every manufacturer to submit their products for independent testing – which they would have to pay for – but it would be more helpful to the educated consumer than a lot of expensive advertising. I’m familiar with accelerated testing, but it has significant weaknesses. I don’t believe that this is sufficient to support some of the claims being made for longevity of LED lighting.

I have been pointing out the need to separate electronics from the source of heat for a very long time, Malcolm. My view is that reputable manufacturers should have recognised that producing direct replacements for higher power incandescent lamps is not a good idea and devoted their efforts to producing light fixtures that keep both the electronics and the LEDs cool, which could greatly decrease failure rate.

“I’m familiar with accelerated testing, but it has significant weaknesses. I don’t believe that this is sufficient to support some of the claims being made for longevity of LED”
In the lighting industry in particular a good deal of work has been expended on developing accelerated life testing techniques and analysis. Particularly for commercial lamps with ives in excess of 3 years, and now LEDs with lives of 6 years continuous, it is essential. I give a link to one paper that touches on this, but there is a mass of research to ensure reliable predictions – for good quality products of course.


Cree is a reputable manufacturer in both the commercial and domestic markets. The linked document looks at one aspect of LED production – pick and place and the issues that can affect reliability. I just refer to this as an example of the complex nature of the process, the expertise required, and the investment needed to develop production methods – let alone the development of the LED itself. Will you find this commitment with all your LED suppliers?

If we could rely on large manufacturers to produce good quality, reliable products then Which? and similar organisations could close their test labs. Thank goodness they take nothing on trust.

I am prepared to trust independent testing and good warranties, not claims by manufacturers.

Not so – testing is not just about reliability (and despite all the testing this seems a criterion that is rather lacking, otherwise we might have better ammunition to pursue early failure claims). It is also about performance. We buy products partly for how they perform, and will pay more for “better” features and performance. That seems to be the main thrust of many product reports.

Do Which still have test labs? I thought they were closed and it is all sub-contracted. If so, that would be a shame – particular expertise is needed to interpret results in the consumer context. Hopefully I am wrong – perhaps Which? would confirm where they test.

I have a lot of products that are reliable and have done just what the manufacturer said – don’t be too cynical.

Fair enough Malcolm. I agree that testing is not just about reliability and also that it’s something we should be told more about. My point was that Which? or its contractors are trying to test what we buy rather than what a manufacturer would like them to test.

Me a cynic? I’ve also good service from most products I have purchased, but others are not so fortunate. I think we need to keep up the pressure.

Wouldn’t it be good if those of us who are passionate about product quality and reliability could be given an invitation from Which? to see how products are tested. 🙂

Or an invitation to test products? I do this already for Amazon.

My main criterion for performance is reliability. I am not happy to buy a light bulb or a Which? recommended car (eg, a Ford) if it is going to break down constantly, even if in the latter case it has a few minor things going for it.

As for reader feedback, I have filled in a pile of those forms online and every one has missed major points, or has been so superficial as to be worthless, yet readers are expected to base their decisions on an xx% score built on thin air, a waste of time for all who filled in the questionnaire.

Alex – Thanks very much for this information. Many of us would like to know more about how product testing works, beyond what we learn in the magazine. Where your colleagues have provided additional information beyond what is published, that has provided reassurance that testing is more thorough than is obvious from reports.

Sometimes there is a suggestion that testing could be improved. The most relevant example I can think about was relates to CFL testing, where a photo of lamps mounted on a panel appeared on the Which? website and the magazine. That is not representative of domestic use, where the lamps will be enclosed by a lampshade or fixture that could cause them to run much hotter and fail prematurely. While I applaud the fact that Which? coordinates independent testing and understand the need to make use of specialist labs to allow testing of a wider range of products, I believe it would be good to give us the occasional example to prove that testing is carried out with sufficient rigour.

Customer feedback questionnaires are extremely shallow, though, so perhaps we should ignore those scores? Are these done in-house, or outsourced to the GCSE exam question setting teams at AQA/Edexcel/OCR?

I still have no faith in the testing. 40 years ago it was rigorous and well documented, if not totally comprehensive. Now it seems to be superficial, undocumented and usually misses major brands, making the results inconclusive and worthless. Shame on you, Which?

Alex, thanks.

As a Which? member I would be very keen to get access to the reports with full information that lie (not as in liar!) behind the published article. I might, for example, have different criteria in selecting a product from those used by Which?

I may also be interested in the testing methodology, particularly for products where I might have some expertise or knowledge. Would there be an objection to making them available on-line?

Alex – The photo in the article you have linked to is the one I referred to yesterday. I really hope this is not the way that Which? has CFLs tested. In the real world, people use them in unventilated lampshades and even in totally enclosed light fixtures, where they are likely to run much hotter and fail prematurely. I would like to see Which? advising us that CFLs are not suitable for use in this way. Some lamps are sold with a warning about this problem, but only in small print.

Alex, thanks for the reply. I had looked at this page but I was looking for more than this gives. For example, I would like to see, for each lamp type tested, their actual lives, early failures, how much the light falls off over the test period. i appreciate most people will not want this, hence access to the raw data for those few who might be interested. Stars are limited in their usefulness!

One testing issue I would raise is durability – here it is given a 30% weighting. To my mind, for most products, whilst I want good performance that is no good unless the product lasts a decent time. If I have to replace it, or repair it, within an unacceptable time then the product should never be given a best buy or even a recommended buy status. This does mean benchmarking acceptable trouble free lives for products – something I believe Which? should be focusing on.

wavechange, CFLs sould be capable of operating in enclosed lights – I have a number that have operated perfeclty well and have had long lives. It will, of course, depend upon their wattage and their quality – as we have said before, rubbish electronics will cause early failure. Lights designed for CFLs will have the maximum watts on a label; older lights for incandescent will not (or they will have the GLS rating which is not appropriate).

I don’t know what the EN testing regime is for CFLs – I hope Which? follow that. But sometimes additional testing is worthwhile to reflect what consumers might do with the product – so testing in enclosed lights at the appropriate wattage should be included.

Malcolm – I am looking at the box for a 950 lumen Tesco spiral CFL and the information on the box includes: “Do not use in enclosed fixtures. Usage in recessed fixtures could result in reduced product life.” I have been studying the issue of overheating electronics in CFL lamps since we started to discuss them on Which? Conversation, looking at the extent of browning of plastic lamp bases and evidence of heat damage to the electronic circuitry inside.

From what I have found, I believe that Tesco was right to put this warning on its lamps. Electronic components – even better quality ones – are more likely to fail if they get hot. That’s why keeping electronics away from the sources of heat is well established as a way of achieving reliability.

Plug-in replacement LED lamps tend to be smaller than comparable CFLs and how we use them probably has less influence on their operating temperature, but I have no practical experience so far.

wavechange, it’s probably just a “get -out” by Tesco. CFLs are widely used successfully in enclosed and recessed lights. It is a case of ensuring the power is suitable for the size of light. A large enclosed light will clearly be quite capable of handling a CFL, but a tiny one will not, or may only handle a very low wattage.

Keeping electronics within their maximum temperature is key – but good quality components will operate at quite high temperatures, hence my comment on quality of the CFL.

As I said, I have used them very successfully – years of operation. Buy quality and use common sense. I would urge Which? to include testing in enclosed lights – keeping to the maximum recommended wattage of course.

Malcolm – On the other hand, Tesco might be doing a good job in providing helpful information. I have seen other examples of various supermarkets providing excellent information.

Obviously there is less problem with lower power lamps and larger enclosed fixtures, but don’t forget that we have both made positive comments about non-domestic lighting in with the electronics is separate from the lamps themselves.

Can we get back to the domestic replacement bulbs that (a) Which? have tested, and (b) are generally available in abundance all over the internet and towns? Saying that we should separate this from that is irrelevant to the Which? test results. We have a very specific problem here, namely plug-in LED replacements for standard sockets, of which the UK must have a billion.

Replacement, LED dedicated, light fixtures, unless they are lamp stands or table lights, would only be a very long term option. Most of us also have LED torches, electronic equipment containing LEDs, etc, but they are NOT plug-in light bulbs!

Mike, the best solution is dedicated LED products. You will find all these from reputable lighting manufacturers on the web – luminaires made specifically for them with decent electronics. They are mainly sold into the commercial market because the domestic consumer has not been educated to look for them, to appreciate their advantages, they are not readily available on the high street and, importantly to many, they are not cheap! So in many ways we get what we have paid for. Some plug-in LED replacements are good, others not. But there are so many out there how do you weed out the good from the bad, and they change all the time. It might be anti-competitive, but picking first-off a small number of supposedly reputable manufacturers and having consumer associations sponsor rigorous testing on reliability and light output would be a start.

Having looked at Luminaire, I might consider fitting one of their fixtures in a porch, but in a lounge or bedroom? No way! I assume this is the site?


Mike – I don’t see why LED light fixtures are a very long term option. Look at how many people have installed any number of downlighters in place of central lighting, wall lights, chandeliers, fluorescent strip lights and a bizarre collection of energy-guzzling decorative fixtures with handfuls of halogen capsules. There is a very good reason to opt for properly designed LED lighting fixtures in which neither the electronics or the LEDs run hot.

I hope Which? sees the light and starts testing some fixtures that are designed to make best use of LED lighting.

I do accept that this Conversation is about LED lamps that are plug-in replacements for bulbs, but the sooner that we recognise that they are inherently a poor design the better. The large manufacturers with their test facilities should have managed to work out that for themselves.

Mike – no! Luminaire is the posh name for a light fitting! So I meant light fittings in general, not this site. You could put “LED luminaire suppliers or manufacturers” into Google, but I’ve no recommendations. I’m sticking with CFLs generally – and LED strips under the kitchen cupboards.

If LED’s were just a bit dearer than CFL’s I would take a chance and install them everywhere, but at £7 – £20 each for the popular types, their endurance very unpredictable, and possibly going to pop far too soon, I shall stick with what I have. Eventually, after many trials and disappointments, I have found CFL’s that do the job satisfactorily and have plenty of life left in them. After all, it’s not as though lighting is the biggest problem in energy consumption; most of the time they just hang there unused and are only on for a short time anyway.I would replace halogen spotlights though as they give out too much heat and use too much power for a very unsatisfactory illumination effect [a cone of projected light that does nothing to enhance the appearance of rooms or the people in them].

Very true, It’s like the campaign to say goodbye to standby, when standby these days costs about 10p a year per appliance, unless it is the much pushed and promoted Youview and , when the cost can be between £4–12 a year. Given that some people have to heat their house with electricity, even that is insignificant.

I’ve been watching the prices in shops, and CFLs seem to be rising in price. On the other hand the price of LEDs is falling, particularly the lower brightness ones.

I sticking with my CFLs, and have enough to keep me going for years, including a bunch of Philips lamps that I bought for 10p each and unwanted CFLs donated by friends and family who received them free by their energy suppliers (sadly not clever enough to ask which style of cap would be useful).

John, agreed that CFLs are a good general-purpose replacement for incandescents. However, no good where you want accent lighting – such as tungsten halogen reflector lamps. I’ve a bar of those in the kitchen that are widish beam and properly aimed – no glare and effective. They’ve been there for 2 years and I’ve no intention of changing them to LEDs – not worth the cost and I like the colour.

Agreed Malcolm. Your home sounds very stylish. I liked the effect you could once get with top-silvered bulbs reflecting back into a brushed reflector fitting: halfway between a spot and a flood. I left some behind in a previous house and have often regretted it. I don’t think you can get top-silvered lamps these days however, nor the reflectorised fittings.

With a couple of dozen GU10 halogens that failed frequently and guzzled electricity, I tried two mid-priced LED replacements. I avoided the cheapest and bought Aurora and a mid-priced unbranded lamp that claimed a Samsung LED inside. Three out of 8 Aurora have failed within 18 months, two of the others. The fittings are all well-ventilated.
I have also seen two bulbs fail simultaneously in the same fitting. This also happened with halogen bulbs and I suspect they are insufficiently protected from the sudden variation in voltage/current when a neighbouring bulb fails.
We really need a new Which? review of these bulbs, both on reliability and light quality.The colour temperature is often quoted, but says nothing about the effect on colour rendering. I have found the LEDs generally better for colour than CFL or strip fluorescents, but they seem much less bright than claimed. Most GU10 LEDs give uneven, harsh light. I have found some that have an even beam or an effective diffuser, but only by buying several hopefuls by mail-order and trying them one by one. Most retailers stock only one type and have no idea whether it is any good. I would like to know whether premium brands deliver, or are just charging for their name.
Despite the very poor lifespan, I will stick with LEDs and try to fine better bulbs – halogens are similarly unreliable and better at heating than lighting!

I suspect that the simultaneous failure of two LED lamps will be due to voltage spikes rather than one lamp causing failure of the other.

It might be worth looking at suppliers who specialise in selling LEDs for off-grid use. These can cope with over-voltage and voltage spikes, which are common problems for these users. I cannot recommend any supplier.

Unfortunately we don’t know much about how manufacturers or Which? tests LEDs but I hope tolerance to voltage spikes is included. We need to get to the bottom of why some people are having problems with LEDs but others have no problem.