The dimmable bulbs incorporate silicon-based Infineon or ST power transistors. We did not find silicon carbide technology in any of the lamps.
Each lamp uses a different LED driver IC. Pharox uses four ICs, 40 resistors and 16 capacitors, plus transformers and many diodes. All this to dim a 40-W-equivalent light bulb—but only, in some cases, if you purchase a new dimmer, since your old dimmer might not be compatible.
Philips uses a remote phosphor, relying on the bulb envelope material to do the wavelength conversion. Lutetium oxide is used as a fluorescent medium in more than one lamp. We have looked at many LEDs over the last few years and have never seen lutetium used before. More conventional phosphors, such as barium and cerium-doped YAG (yttrium aluminum garnet), are also seen, along with a cocktail of rare earths such as europium and strontium.
Surprisingly, everyone uses blue GaN LEDs—no RGB LEDs in this lot. There are red LEDs in the Pharox, complete with lenses, but they don’t turn on when the lamp is on. They are obviously not for color balance. Maybe the light turns red when the bulb is about to die, or to signal that the lamp is not wired properly.
Philips uses many of its own Luxeon Rebel LEDs in a modular arrangement that could possibly permit the replacement of failed modules. We expected the Sylvania to use an Osram LED, perhaps a Dragon, but that is not the case. Two of the lamps rely on Cree dice.
The LED light-emitting area varies from 4 mm2 to 12 mm2 for the 40-W equivalent bulbs—a variance of a factor of 3—yet their power consumption ranges between 6 and 9 W. Some have four LED dice; others have 48.
Click on image to enlarge.
At what point is such a lamp considered to have reached the end of its operable life, given its dozens of individual LEDs? Nowhere on the fancy packaging, with all its warnings, disclaimers and warranties, is there a definition of the point of final failure for the lamp. Will it go out with a flash? Will it still be usable for its intended purpose until the last second, or will there be a point at which the light has become so faint that the user decides to replace the bulb?
Lots of questions, and we still haven’t answered the first one: Why are A19 LED bulbs so expensive? At 50 to 60 lumens/W, the LED bulbs have the same efficacy as 40-W-equivalent compact fluorescent lights, which typically offer 560 lumens at 10 W, can be purchased for a few dollars and should last, or so the packaging claims, up to eight years. For the same savings in electricity as you’d get with a single LED bulb, you would need to purchase two or three additional CFL bulbs over the next 25 years. Luckily they are available in packs of two or four.
Light bulbs are a commodity item, no matter how long they last. For all the claims of CFL longevity, I have yet to see them reach their expected lifetime. Usually the point of CFL failure is the electronics in the neck, and the same will likely hold true for LED bulbs.
With CFLs, however, we have traded energy consumption for mercury contamination. The attraction of LED bulbs is twofold: They are instant-on, and they are mercury free. The toxicity of rare earths is low, and very little of the material is used in the bulbs anyway.
Over time, LED bulbs will drop in price and will provide a cleaner, safer alternative to CFLs—presumably with a longer life expectancy as well. We can also expect new features; the door is fully open on that front.
I've several years experience in drivers for LED lamps, and did investigate visible and non-visible flicker. Next link provides my input for Energystar guidelines on LEDlamps and light modulation.
Electrolytics are available which survive 12,000 hours at 105C. Unless you are cooking your LED's then you can get extremely long lifetimes and my own patented 900 lumen, 90+lumens/w 60W equivalent dimmable design exceeds 98% efficiency so almost no heat from the driver. And all that with 0.993 PF excellent coliur rendering and NO EMI!
Good idea - there are CFL fittings that work like this, the 2-pin lamp has the tube and the starter and the ballast is separate. There are also 4-pin types for electronic ballasts. So same idea should be good with LEDs - and reduce the ongoing cost a bit maybe.
Hi I am from India and developed LED lamp that works out from the regular incandecent bulb sockets.All led bulbs require Electronics to step down the voltage as they work on 3.3 to 4 volts only.As the current requirements are more heat will be developed and cannot be avoided.The led may last more but the electronics is bound to fail soon than claimed.
Those would be directly emmitting diodes then. These diodes shine onto phosphors to make white light. The phosphors continue to glow for brief period after the diode stops emmiting and might be able to mask the flicker. We ran into a problem with white LEDs on a system that was trying to blink at a 50% duty cycle only to have the phosphor glow distort it.
I guess Incandescents have a good bit of thermal inertia and don't dim much if at all in between half cycles.
TV's would also have a bit of persistence on the phosphor...and CFLs also usually have an electrolytic in them. Whereas LEDs would definitely be on-off devices.
Memo to self...when studying engineering textbooks, do not move head.... ;-)
An interesting question as to why need electrolytic capacitors. A portion of the problem is indeed avoiding flicker, although I am not convinced that a small PP capacitor right behind the line rectifier would not be adequate. Of course one other challenge is the emphasis on absolute minimum BOM cost, notwithstanding the quality problems that result. One dollar more to assure the 50K hour lifetime would be quite reasonable to me.
What I think to be very interesting would be a cost breakdown on the various devices, since I doubt that any of them has over $5 in electronic parts, and I am talking Digikey prices. Knowing just where the cost lies would be a very good start.
I have also wondered about why nobody markets an LED device that uses the rectified line voltage and a current limiting resistor, and perhaps 80 or a hundred "fairly bright" surface mount, or even through hole, LEDs. No filtering, just leds on the full wave rectified line, with 120HZ flicker, which I don't think we could see. Does anyone see a fundamental problem with that?
I bought a LED desk lamp recently which runs off a switching "wall wart" type 20W supply. The separate PSU is probably good for 50000 hours. The way forward with LED lamps is probably to decouple the bulb from the power supply - the low voltage DC supply should be part of the lamp socket, not the lamp itself.
Join our online Radio Show on Friday 11th July starting at 2:00pm Eastern, when EETimes editor of all things fun and interesting, Max Maxfield, and embedded systems expert, Jack Ganssle, will debate as to just what is, and is not, and embedded system.