The electronics in the driver are far and away the most likely failure mode. The LEDs themselves have an insanely long lifetime if they are kept cool.
The materials used in the LED itself are generally indium, gallium, and nitrogen. The phosphor is yttrium, aluminum, and oxygen, usually with a trace amount of cerium. The quantities of the rare earths used are tiny.
Whith products sold to consumers, I would think the driver electronics would be likely the first part to fail, especially if there are electrolytic caps in it. But only time will tell for sure.
I maintain the lighting in the lobby and stairwell of our apartment building, and I mainly use consumer grade CFLs. At one point in time I used both the Philips Genie and an identically looking and specced version sold under a private label (but with a Philips Lighting factory symbol on the socket).
In all Genies the ballast outlived the lamp. With the seemingly identical private label, however, it was the other way around.
I opened several of both and found that the Genie had a slightly different ballast with higher grade caps than the private label. In all of the private labels the smoothing cap had blown.
The answer to your question will probably depend on the quality of the components used in the LED driver. I'd definitely stick with reputable brands.
With professional equipment, I'd expect the driver stage to have been designed with as long as possible lifespan in mind.
As an example I take the Philips GentleSpace LED high-bay lighting http://download.p4c.philips.com/l4bt/3/343537/gentlespace_343537_ffs_eng.pdf .
The specified lifetime is unbelievable (75,000 hours at 70% lumen maintenance). They also specify total luminaire failure rate and driver failure rate.
Last edited:
Yes the sapphire substrate is used as a mirror. People think that LEDs are by nature directional, this is true for the light emitted from the completed unit, but the actual emitting device isn't. From what I understand the sapphire substrate is used as a mirror to make sure that most of the light is reflected back in the direction it is wanted. The sapphire part of the construction is apparently one thing that sets a reasonably high price for LEDs. I do know that some manufactures are researching the use of different materials for this job, as a way of reducing price, but as far as I know most LEDs still use sapphire.
This isn't necessarily true. Most of the low voltage LED drivers don't need anything else then a few small ceramics. So in theory any low voltage halogen replacement would never need see an electrolytic cap. Of course the SMPS that steps down the mains voltage probably does have lytics in side but then this can be hidden somewhere out of sight, can be reasonably bulky and can be kept cool. I think the biggest bane of LED lighting is trying to come up with retrofit bulb designs.
I totally agree. Lately I've noticed some wonderful examples of how well LEDs can perform in made-for-LED luminaires in industrial and commercial applications.
But the ugly retrofit LED bulbs intended as a replacement for incandescent lamps? That's not the future of LED lighting, at best it's an intermediate that IMHO should disappear in the next decade and take the old sockets (e.g. E14/E27) with them.
I recently removed under-cabinet lights in my kitchen and fixed instead LED rope there under cabinets. Now it is more reliable and the light is smooth and well diffused.
Speaking of LED bulbs, it is pure business approach, to sell more of goods for existing sockets. And some goods are bad, for example a bulb I bought one month ago has a cap that leaks electrolyte.
Speaking of LED bulbs, it is pure business approach, to sell more of goods for existing sockets. And some goods are bad, for example a bulb I bought one month ago has a cap that leaks electrolyte.
Last edited:
Well I think the biggest problem is the mindset that everyone has attached to lighting in general. This is at least true in the sense of a lot of decorative lighting, where traditionally one buys the lamp body/shade etc to suit their decor and then you fit the bulb and replace it as it breaks.
The trouble with that is that most people already have loads of lamp bodies and as a result the industry tries and makes replacement bulbs. What one needs are entire replacement lamp bodies for complete replacement designs where the body can act as a heatsink. This vastly increases the available surface area for cooling the LEDs and allows an efficient driver to be used in the base of the lamp. Efficient mainly because the size constraints created by needing a tiny LED driver in a replacement bulb drive down component sizes but as a result reduce overall efficiency, they also run hot which reduces efficiency even further, both in terms of the driver and the LEDs themselves. Having a larger cool space means you can use lytics without too many concerns for reliability if decent quality parts are used.
Of course the lamp base design would have to emit light over the surface area of a sphere, but they could be designed with a normal shade fitting built in so one at least can have some extra choice over the decor. They would also be quite expensive, but brightness shouldn't be an issue as the extra heatsink area provided by the base unit would allow the light output to be scaled up so that it easily competes with or outperforms the light output of a standard 60 watt incandescent.
The trouble with that is that most people already have loads of lamp bodies and as a result the industry tries and makes replacement bulbs. What one needs are entire replacement lamp bodies for complete replacement designs where the body can act as a heatsink. This vastly increases the available surface area for cooling the LEDs and allows an efficient driver to be used in the base of the lamp. Efficient mainly because the size constraints created by needing a tiny LED driver in a replacement bulb drive down component sizes but as a result reduce overall efficiency, they also run hot which reduces efficiency even further, both in terms of the driver and the LEDs themselves. Having a larger cool space means you can use lytics without too many concerns for reliability if decent quality parts are used.
Of course the lamp base design would have to emit light over the surface area of a sphere, but they could be designed with a normal shade fitting built in so one at least can have some extra choice over the decor. They would also be quite expensive, but brightness shouldn't be an issue as the extra heatsink area provided by the base unit would allow the light output to be scaled up so that it easily competes with or outperforms the light output of a standard 60 watt incandescent.
Over the last month, I've bought a few different LED globes, The Osram bulbs, though expensive, have much better light than CFL, the cheaper Chinese ones very bright white, interesting to see if they last the distance...
An interesting problem with a 4.5W Osram bulb in a downlight fitting - it would flash, with the power turned OFF - apparently the capacitance between live & neutral is enough to charge the starter circuit to the point where it will discharge. Swapping the bulb with a different model fixed the problem, but apparently a high value snubber resistor will also fix it.
An interesting problem with a 4.5W Osram bulb in a downlight fitting - it would flash, with the power turned OFF - apparently the capacitance between live & neutral is enough to charge the starter circuit to the point where it will discharge. Swapping the bulb with a different model fixed the problem, but apparently a high value snubber resistor will also fix it.
This is a common problem with CLFs too and happens in most of the ceiling hung lights we have in our house
Sapphire is used to get the heat out, CREE uses SiC and is the only source of starting material hence everyone else using sapphire (or almost so). Diamond starting material is available but I suspect it is not cost effective.
CREE is doing a good job publicity wise, I bought some flood lights (Costco) that said CREE inside or something to that effect. There is a theoretical max efficiency for Lumens/Watt and I think they are about 1/2 way or so there unlike solar panels.
Last edited:
Interesting I can't remember what it was I was reading about LEDs and the reason behind Sapphire, but the must have got it completely wrong. Either that or I skim read it and got the wrong impressions.
Well I do wonder if we will ever actually reach the maximum, perhaps, it depends on the operating conditions.
For example you get far more for your input power if an LED is driven with less current. The theoretical maximum I think is something like ~300lm/w. So say CREE come up with a 300lm/w LED when it's driven at 500mA, naturally when you turn it up to 1amp this will deteriorate, but if you turn it down, if 500mA is already at the ceiling limit, then surely it cannot get any brighter. Or is this theoretical maximum derived from some specific operating conditions?
CREE do have the marketing handled well, but also their products do shine brighter then the competition so they've got themselves a bit of a win-win situation, imo.
Remember white LED's have a largish forward drop. Digikey has individual units like the ones in the floods. The specs are 6V @ 2A, that's 12W and those die are not that big. I suspect the heatsink also helps with code rules for mounting in enclosed spaces. I wonder how many towns have updated their building codes, i.e. only flourecents in closets, etc.
In my brief reading the other day on the sapphire substrate, it noted a particular serendipitous characteristic of the sapphire lattice structure. To quote Wikipedia, "In a remarkable phenomenon of nature, the r-plane of sapphire has oxygen atoms spaced at a distance that is close to the spacing of the atoms in the (100) plane of a silicon crystal. The oxygen atoms on the r-plane have a square symmetry that also mirrors the symmetry of the (100) plane of silicon. The (100) plane of silicon is the same crystal plane that is used in all CMOS electronics." Something wholly new to me. I also read that light directivity is a major hurdle in designing LEDs.
LEDs emit light around the "edges" of the die. Imagine a LED die as a jam sandwich, with the P and N layers being the bread, and the junction layer being the "jam" in between. The light is emitted from the edges of the "jam" layer. If you take a careful look in a clear LED, you'll see the die sits in a small reflector unit that reflects the emitted light through 90 degrees to form a unidirectional beam.
Laser diodes are somewhat diffeent, they emit their light from a small region on one edge. The light emerges in a broad, oval shaped beam. A specially shaped ("aspheric") lens is normally added to focus the beam into a round spot.
Last edited:
Sapphire is actually transparent, this is a nice article on white LED technology and how they are trying to make it more cost effective.
http://www.mikewoodconsulting.com/articles/Protocol Winter 2011 - White LEDs.pdf
http://www.mikewoodconsulting.com/articles/Protocol Winter 2011 - White LEDs.pdf
I think I share most of the authors concerns with regards to LED lighting too. My main worry is the quality of light. My guess is that manufactures might push efficacy whilst maintaining an average CRI of around 80, but then once certain limits have been reached they will then perhaps push to increase the CRI.
Scott, thank you for the link to the interesting article.
I don't think we'll see a much higher CRI than in the 80% range for standard products. The phosphors used in white LEDs are "closely related" to those used in fluorescent tubes and they typically have a CRI in the 80% range for standard quality tubes and CFLs (e.g. the very common colours 827, 830 and 840). Since we're already used to this kind of CRI, I don't think there's much incentive for manufacturers to go higher, at least not in standard products.
Technically it should not be a problem to make high CRI LEDs as the technology already exists for high CRI - in the 90% range - fluorescent tubes (e.g. 930 and 940). But as is the case with these tubes, I'd expect similar quality LEDs to also become a specialty product, especially as the high CRI seems to be less efficient (e.g. a fluorescent tube 36W/840: 93 lm/W and 36W/940: 77.8 lm/W).
I don't think we'll see a much higher CRI than in the 80% range for standard products. The phosphors used in white LEDs are "closely related" to those used in fluorescent tubes and they typically have a CRI in the 80% range for standard quality tubes and CFLs (e.g. the very common colours 827, 830 and 840). Since we're already used to this kind of CRI, I don't think there's much incentive for manufacturers to go higher, at least not in standard products.
Technically it should not be a problem to make high CRI LEDs as the technology already exists for high CRI - in the 90% range - fluorescent tubes (e.g. 930 and 940). But as is the case with these tubes, I'd expect similar quality LEDs to also become a specialty product, especially as the high CRI seems to be less efficient (e.g. a fluorescent tube 36W/840: 93 lm/W and 36W/940: 77.8 lm/W).
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.