I live in a new-build small estate of 10 houses. We are all having incidences of short lives of G10 LED downlighter bulbs. They are driven by dimmer circuits made by Deta. The Deta Connect system has been installed to control these dimmer light controls and some of the power outlets. The G10 bulbs are installed in metal pots in the ceiling and the insulation is clear of the pots. The failures appear to be of the small printed circuit inside the bulb. (I broke one bulb open and connected a 50 V DC power supply to prove that the LEDs still worked.) I measured the temperature inside one of the pots after a few hours "on" - the maximum was 42 degrees. I suppose the internal temperature of the bulb will be a few degrees higher. The DETA specification requires a minimum load of 20 Watts. This is marginal as we have some rooms with 4 x 5W bulbs - these sometimes flash. I have measured the mains voltage over one day seeing it between 237 and 243 Volts - so within mains spec. Various makes of dimmable bulbs have been fitted, all made in China, rated for 230 to 240V. I used an oscilloscope to view the voltage applied to the bulbs and did not see any spikes in the waveforms. I wonder if the small circuit produces more heat when operated dim? I suppose most of the heat comes from the LEDs themselves? I have a suspicion that some component in the internal bulb circuit cannot survive for long at the temperature that it is suffering. This suspicion is borne out by the observation that other LED bulbs mounted in unventilated ceiling fittings fail quickly while LED bulbs used outside seem to last well. Surely there must be general concern that short life is spoiling take up of "green" LED lighting? I would welcome comments and suggestions.
There is a general problem that folk expect an LED light to cost about the same as hot wire in low pressure gas, and there are a lot more steps to making the former, so in a bid to squeeze the price right down, the process corners tend to be cut.
Also matching up Dimmers and LED fittings is very much a dark art, in that there is no guarantee that the one maker's idea of a dimmable design is the same as another, and there is no standard set of tests for either dimmable lamps or LED lamp dimmers. There is near universal agreement with waveform slicing dimmers that it is better to cut the power off part way through the cycle and turn it on at the zero crossing, but traditional dimmers for filament lamps using triacs do the reverse, and a few LED fittings are set up to expect that, and some dimmers can be programmed to do either - if yours can, check the right mode is selected.
As you note cooling is a problem - actually I'd say probably the problem for electronics, and light fittings that look like traditional fittings re-purposed are not always the best at heat dispersal.
It should get better as the makers of fittings and lamps wise up, but right now the technology is immature, and everyone is also trying to recoup their outlay on re-tooling, new products etc. You could argue that we should be asking how reliable was the filament lamp in 1880 ? and comparing to that.
So it would seem to be a good idea when buying light fittings to choose ones which have the bulbs pointing up so that the heat does not flow past the electronics, and make sure there is plenty of ventilation. Choose function over style perhaps?
I think the prohibitive costs of prototyping new electronic products are soon outweighed by the massive productive processes usch as the microprocessor chip. The ammount of R&D that goes into that is enormous.
I think many of the failed LED products are eariler editions that need to be cleared off the shelves to make way for later arrivals as well as some shoddy production processes.
If we consider specialist lighting companies like C......w.....d. How many of their products have premmaturely failed?
The cost of bringing something to market is a very variable thing - when I worked for Siemens mobile it was a few million to spin up a new handset model that was generally a new case with innards that were as far as possible making use of the earlier models on a new PCB layout, and a couple of man-years of new software.
Anything for Aerospace or medical use seems to need comparable amounts of dosh just for regulatory approvals.
Set against that backdrop he fact that a wafer of customer chips is a few hundred k of foundry costs and some man-years of design is not such a big deal.
I'm not sure that this scales well for light bulbs which are much more at the 'pile it high sell it cheap' end of the market.
We could do a bomb proof LED light in the same way, but I fear no-one could afford it.
You may find this article that looks inside a number of lamps intersting read
We could do a bomb proof LED light in the same way, but I fear no-one could afford it.
regards Mike
Interesting article, nothing like a consipircy theory.
I agree that an increase in the 'volume' (current), will produce an increase the temperature and therefore decrease the time before failure. So if you want to control the obolesence then make sure the lamp elctronics gets hot
However, there are companies out there starting around £50 a pop, perhaps not the right word, that produce very reliable LED lights so it can't be the technology but the second rate ingredients and the methods of production that lowers the quality.
I remember getting heated up myself when somebody , an architect, once said to me that he regarded that it was almost unlawful to produce a product that had been, in effect, over engineered!
Unlike their low voltage (DC) cousins we can't just stick a PWM controller on the front of them and cut up the DC. They have to have some form of controller on-board (sometimes to do just that while receiving a terrible supply) and that's why they have to be made more expensive than the non-dimmable versions. It's a hard job to make a delicate circuit that runs off seriously butchered AC waveforms, especially when those AC waveforms can be leading edge or trailing edge cut. Worse still we are dealing with an electronic dimming circuit that needs a load to work properly and more often than not was never intended to be used on suck low current devices. (Most LED dimmers I have seen are incandescent dimmers slightly modified)
Is it your lamps? or is it the struggling dimmer killing the laps? Who knows?
Heat CAN be an issue but many of our GU10's are in quite enclosed spaces - However all the lamps we use have ventilation slots around the sides of the lens to ventilate the rear of the lamp.
We did all this work about 10 years ago.
My specialist field is making studios for music film and TV, you can imagine how "critical" performance of the lighting can be in every aspect (Light output. Quality of light. longevity. Noise levels - physical and electrical) We used GU 10s in many areas including music recording and sound control rooms.
About 10 years ago a study was done on an audio forum (Gearslutz) into performance of dimmable GU10 devices about 10 types available at that time were looked at by various people over a few months and surprisingly the best performing of all were a relatively cheap Chinese brand in light output silence and quality of the dimming.
Worse performers in terms of noise were the top two brands.
We had very different results with the two big name brand lamps in our tests depending upon the dimmer we noted differences in noise and heat from the lamps. As a consequence we decided we cant trust the test results as what people experienced with the lamps may have been related to their dimmers, so we ran a set of tests again.
By the end of it all we decided that the issue was the dimmers as much as the lamps.
As we had a decent budget and plenty of space in the physical constructions we went for the nuclear option and abandoned dimmers in favour of small Variacs. (Well retro cool man!!!)
Using the very silent cheap TCP lamps (specific model no longer made) we have had now zero failures (except a few D.O.A) in 9 years with more than 500 units installed in daily use. (this has shocked us) Almost every installation runs permanently in a dimmed setting.
Additionally We found that a struggling dimmer (too small a load) on a whole load of struggling electronic circuits was a massive source of electro-magnetic interference in the installations. We were getting large fields of pick up in sensitive audio gear that we tracked down to office and corridor dimmed LED systems. (same kind of set-up as the OP has mentioned, dimmer at minimum load lamps "just" making that load and system prone to occasional flashing because of dimmer load issues.
In entertainment lighting the dimmer load issues have been well documented for many decades. When I worked in "sound and lighting" install back in the 90's we would have banks of dummy loads back in the machine rooms.
If you went into any UK Chicago Rock Cafe, Liquid Nightclub, Oceana, or Time/Envy equipment rooms and saw a whole wall full of 100w lightbulbs they were dummy loads we installed for the Mode dimmable electronic neon transformers. it was the same problem, the dimmer could put out a junk signal and ruin the neon transformers or burn the triacs in the dimmers. Neons would flicker like a Hollywood movie neon with the low load issues. In domestic use and for environmental reasons (100w dummy load) this is NOT an option for this issue, it simply illustrates the dimmer problem.
For the Original question.
I know it's a pain in the butt, time consuming, and expensive, but as there is no "standard" LED 230V dimmer / lamp compatibility requirements we are kind of on our own with this until there are.
Our solution was to run tests back at base. Get a few dimmers and a few LED lights rig them up and just run them Look to see if there were issues with load, flickering , and reliability. Our office and workshop lights are actually a long-term test bed.
You really cant rely on having "just" enough lamps to create the required load as this can cause a chain failure should one lamp fail. Everything "just" works OK until a lamp fails then the whole system goes unstable and one by one pop pop pop out go the lights.
Maybe someone who has a particular combination that works OK can tell you and you can go from there.
Sorry but my solution (Variacs) simply won't help in a home (non industrial) environment.
Maybe just as a cheap and cheerful experiment stick one halogen GU10 in the circuit (temporarily) and see if it calms things down. If so then maybe its the dimmer.
