Remote Phosphors: Philips LED bulb, Tear-down Part II
We’ve seen that Philips’ dimmable 12.5W 800 lm LED bulb emits light in the same spherical pattern we’ve come to expect from incandescent bulbs. How does it do this?
First off, let’s refer back to the unusual-for-a-white-light-bulb yellow color of the Philips bulb. Philips clearly expected some raised eyebrows from shoppers because the top text on the product package is: “WHITE LIGHT WHEN LIT,” placed right over the yellow bulb top.
[$39.97 at Home Depot.]
Your basic white high-brightness LED is based on an intensely blue LED with a dollop of phosphor directly on top of the blue emitter. The LED’s blue light strikes the phosphor which in turn emits white light, serving as a very directional point-source. This type of LED/phosphor combination is called a primary phosphor.
The yellow plastic of the Philips bulb not just a bulb-like cover over the LEDs, but the phosphor itself. Because it’s located separately from the LEDs it’s called a secondary, or remote, phosphor.
By popping off one of the three yellow plastic sections, we can see that the LEDs – which are blue, not white - are mounted vertically on the interior central column of the bulb.
By using a remote phosphor, the bulb is able to make use of the remote phosphor’s characteristic of emitting light omni-directionally and uniformly, rather than as a point source such as in a standard white LED.
Sure enough, when I power up the bulb with the yellow plastic cover removed (GAAUGH, my retinas…) we can see the intense light of the blue LEDs.
[Although I joke about the intensity of these blue LEDs, I think they are really quite nasty and shouldn’t be looked at directly without protective goggles. Kids, don’t try this at home.]
But let’s keep going and find out What Lies Beneath, although this meant destroying the bulb which caused me great mental anguish.
Here’s the top view of the bulb looking down into the wiring and connectors that take the power to the three LED boards.
I was quite impressed by the use of connectors in this bulb rather than relying on low-cost-labor for hand-soldering which has been used in the manufacturing of CFLs. For example, here’s a view of a pc board from a hand-soldered CFL:
The big solder blobs don’t inspire confidence in manufacturing quality and probably have played a role in shorter-than expected-lifetimes for some CFLs. Philips’ decision to rely on connectors makes for a more repeatable, reliable assembly process and should pay off in long-term bulb reliability.
Back to the Philips bulb: This photo shows the LED pc board and connector, as well as the thermal interface material that helps pull the heat away from the LEDs into the (fairly massive) heat sink.
Popping off the socket to expose the power management pc circuitry:
Whittling away to expose more of the power circuit. My neighbor has a machine shop, but all he needed was a hacksaw and some judicious cutting and out popped the potted circuit:
The rubbery potting compound pulled away quite easily, revealing the side of the pc board with all the magnetics and capacitors:
Drum roll, please. Here’s El Dorado – which dimming IC does it use?
The dimming LED driver IC is the 8-pin IC seen on the left. I couldn’t get a good photo of it, but the lettering looked something like CY[?]0LED which I believe is a Cypress CY8CLED 12W dimmable LED driver. Here is its Reference Design Guide. And yes, when I had it connected it to a triac dimmer, the bulb dimmed beautifully from almost zero to the full 100%.
This wraps it up for the Philips 12.5W bulb. I do have some more very interesting photos of a bulb with a related remote phosphor technology and I’ll get to that in a day or so. Just to belabor the obvious, I am quite impressed by what can be done with remote phosphors, as well as the role of connectors in high-volume electronics.
mediatech commented:
I have some of these Philips A19's and they do look nice in table lamps with their even pattern and soft 2700K color. Another good A19 non-dim "40W" ~430 lumen (9w actual) is the GE62180 LED9A19/830/CD which I prefer in some applications over the Philips. The GE is 3000K. For PAR and MR-16 applications do consider the Sylvania/Osram products. I reviewed and have used just about every type of LED I could purchase at Lowe's and Home Depot. The links to my reviews and residential remodel using LEDs. www.proaudiodesignforumDOTcom/forum/php/viewtopic.php?f=11&t=408
Johndp commented:
One interesting investigation would be the EMC implications of these LED bulbs. Certain varieties show really quite bad emissions both conducted and radiated. In the UK some have been shown to completely stop DAB radio reception...
In addition I believe that there are health concerns about LED lighting.
Jorick commented:
Somewhere I saw a test where different incandescent, fluorescent, and LED lights were tested with a repeated on/off cycle of 2 seconds for two weeks. The incandescents burned out very quickly, the fluorescents burned out a few days later, but the LED lights lasted the entire test without any ill effects.
Frank commented:
I like these too, however there may be unintended consequences due to interaction with intrinsically photosensitive retinal ganglion cells by the blue primary emission, i.e. long term exposure may affect your sleep patterns and circadian rhythms. It'll be a grand experiment!
CEDUP commented:
I think since the L Prize winner PHILIPS is coming out with the newer improved winner of this lamp, no wonder my local Home Depot had these for $14.99 ea, I picked up 4 and they are terrific in my Casablanca fan light with dimmer. Excellent color, BRIGHTER than a 54W long life Krypton filled lamp, great stuff when they get to $5 each they will sell BILLIONS and BILLIONS, PHILIPS produced over 1 BILLION incandescent in one year every year, it's LED now baby
Uma Red commented:
I'd rather buy the one that won the L-Prize at 9.7 watts for the same luminosity. Why are they not selling that yet?
IndustrialLEDs commented:
We sell Replacement LEDs of all types over at Industrialleds.com but I have always wondered what the guts look like. :) Cool!
Andy T commented:
1600
Mouseinhat commented:
Great article! I'm wondering if you know the color temperature of the LEDs themselves, or have a close up shot of the led board? I'm asking as the blue light looks actinic to me if it is, they might be useful .
Thanks.
Guillaume Laurent commented:
Thanks for your article, I actually have one of those in my living room, it works quite well.
I have a more general question regarding LEDs which I haven't been able to find an answer for, I thought may be you would have one : why are LED bulbs made of many small LEDs rather than a few large ones ?
Is there a reason why LEDs are only build at such a small size ? Can't they be made larger ?
Thanks
Mark commented:
How many lumens does a normal (incandescent) 100W bulb emit?
Mark commented:
How much of the price here is for the dimming capability? I want some bulbs like this, but don't want to pay a premium for dimming since I don't need it. Is there a cheaper option with no dimming?
Martine commented:
We tore down 5 A19 LED bulbs (see article in EETimes). Indeed the Philips uses a Cypress CY8CLED. However when we opened it, it was not a Cypress chip inside!
etmax commented:
Very nice tear down, thanks. I think new lighting with LED's should progress to something akin to down lights where the electronics is housed in a small sealed box that sits in the roof cavity to go from mains down to maybe ~36V current limited where the lamp has 3 terminals including a programming resistor that sets the current for the lamp. This would avoid electronics heat being added to the lamp and vice versa yet give the flexibility to select what light intensity you want in a fitting and improve reliability.
James commented:
While I've heard a lot of people talk of problems using CFLs in enclosed fixtures, I haven't personally found it to be an issue. I've been using CFL "bulbs" almost exclusively in my home throughout the past decade with a majority of them in fully enclosed ceiling fixtures and have found those that make it past the first month or so continue on for years.
I can confirm that poor solder joints are responsible for a majority of CFL failures and have repaired quite a few of them just because I can. I had a batch of Feit CFLs that also suffered a high failure rate due to the capacitor tying the cathodes together shorting.
I too have recently been evaluating some Philips LED bulbs. They are very expensive, but the dimming performance is far superior to that of dimmable CFLs, also somewhat expensive and difficult to find.
PlanetaryGear commented:
Throughly enjoyed the teardown photos! These are the best LED lights I’ve ever tested. (and the 40 watt equivalent they also sell for half the price but with a similar power supply and light engine) They work perfectly on older dimmers, including old non-neutral equipped X10 or Z-Wave remote dimmers that choked on CFL’s and other LEDs. An additional savings that isn’t usually considered when talking about the price calculation is the cost to remove the heat generated from regular bulbs. In the summer my AC will not cool the kitchen with several hundred watts of regular bulbs running, but cycles nicely now with LED lighting. That saves real money too.
Thomas commented:
Good benchmark and many good comments!
The sophisticated technical realization of this lamp clearly shows that - and this message holds for basically all retrofit lamps in LED as well as CFL technology - the basic problem arises from the form factor of the bulb and the related screw based lamp holder system, which is not capable to remove heat. Please note that in a downlight we have a very unfavourable situation, because convective cooling is not possible and the electronic is at the hottest position, i.e. base up. This is why Ph says “suitable for use in open luminaires only”. A thermally relaxed lamp design with a new flat form factor would dramatically lower the effort the handle heat. In addition efficiency will increase by about 0,5%/°C if junction temperature is lowered. Efforts like using a thermal resin to couple the electronics to the housing would no longer be necessary and the life time of the electronics would not negatively affect the system life time. So what we need to do is to design and standardize a new form factor different from the shape of the bulb and to offer new luminaires which can operate those new light engines. The banning of the bulb is under way worldwide, and a phase out plan for using screw based lamp holder systems in new luminaires should be the next consequent step.
SevenGensLabs commented:
Just goes to show how several people can invent the same thing simultaneously with no knowledge of each others work. A good idea to do a tear-down! We published one 2 weeks before yours, with yet even a different way of showing the superior light dispersion, but you tore into the electronics too. You get the prize for that.. We have done several but no electronics yet. I put a link to your dimming video on our website. Good Job Margery!
Clark commented:
Nice Work - I scanned the comments and might have missed it, but I believe that there is a significant UV content in the bare "blue" LEDs, so please be careful with your eyes!
Andy T commented:
www.rochester.edu/news/show.php?id=3385
TBP commented:
I'm not sure I understood Dr Rob's comment. Is Dr Rob saying that a light bulb with a laser sintered filament could be 3-5 x more efficient in lumens/watt than current LEDs? That would truly be worth some investigation. How about a link or pointer to some papers to support this?
Andy M commented:
" big solder blobs don’t inspire confidence in manufacturing quality and probably have played a role in shorter-than expected-lifetimes for some CFLs". What makes you say this? Electrochemical migration or mechanical joint failure... what is the failure mechanism you hve in mind? Looks like these were panellized wavesoldered (not handsoldered) and reliability should not be an issue.
Daniel Xiao commented:
February 17, 2011
In response to: Remote Phosphors: Philips LED bulb, Tear-down Part II
Suded Emmanuel commented:
From previous reading to calculate the efficiency of a light bulb:
Lumens/683=Watts of light so 800/683=1.17 Watts of light.
so efficieny=1.17/12.3=0.095 i.e. efficiency =9.5%.
The incandescent light efficiency=2% so LED is better.
I'm not sure if the above calculation is correct, any comments please?
Suded.
Maybe there is some wrong about below words.
1W = 683 lm is only suitable at the 555nm. So we cant get the result by this data.
Suded commented:
Sorry just ignore my 2nd comment.
Carole,
Thanks for your reply!
Suded Emmanuel.
Suded Emmanuel commented:
From previous reading to calculate the efficiency of a light bulb:
Lumens/683=Watts of light so 800/683=1.17 Watts of light.
so efficieny=1.17/12.3=0.095 i.e. efficiency =9.5%.
The incandescent light efficiency=2% so LED is better.
I'm not sure if the above calculation is correct, any comments please?
Suded.
Carole_S commented:
@Suded
Efficacy isn't normally calculated as light POWER out/Electrical power in. It's more usual to talk in terms of Lumens per watt (Lm/W) which takes into account the response of the eye (there's no point in emitting in the ultra violet, no matter how efficiently!)The maximum possible efficacy, emitting 100% of the input power as light at the eye's best response frequency is 683Lm/W.
LED lamps are now edging ahead of CL's in terms of efficacy, so definately 6+ times as good as standard incandescents.
Suded commented:
From previous reading to calculate the light wattage 800/683=1.17 Watt light output.
input power=12.3 Watt.
Efficacy (efficiency)=1.17/12.3*100=9%.
So the efficiency is 9%,but thats better than incandescent which is about 2%.
I'm not sure if my calculations are right, any comments?
Suded.
Dr Rob commented:
Incandescent technology is still superior to any other lighting tech, especially solid state. When one factors in laser sintering of incandescent filaments that increases their efficacy to 3-5 times even LED's, incandescents absolutely rule when considering their performance and cost of production/recycling/logistics/..ect... I love bandgap technology, but it will be hard pressed to beat Edison's brilliance (it hasn't been done yet, despite articles such as this one).
Margery Conner commented:
--John M: Yes, the package says, "Not for use where exposed to weather." It also says, "Suitable for use in open luminaries." Enclosed downlights have been the cause of death for many CFLs, as Dan Aquinas points out.
--Steven D: Afaik the LEDs are Lumileds. Another thing I should have pointed out about remote phosphors is that because they are removed from the heat-producing emitter, their lifetimes are generally longer than primary phosphors.
--Steve A: The light color is 2700K - a warm white.
--Neil: Refer to the study done by Osram and Siemens on life-cycle assemssment of LED bulbs: the study found that over 25,000 hours, the total energy required — including manufacturing, use, and end-of-life disposal/recycling — for a 8W LED lamp is about 700kWh. The total energy requirement for incandescent lighting is about 3,300 kWH. Read more here: edn.com/blog/PowerSource/36030-UPDATED_Life_cycle_study_examines_total_energy_used_in_LEDs_CFLs_and_incandescents.php
--I should have been clearer: I was referring to the poor reliability of *hand-soldered* joints, which is what you see a lot of in CFLs. Yes, a nice wave-soldered-joint is relatively high-quality, but a good connector - which is what's used in these bulbs - is much better than a crappy hand-soldered joint.
--I agree with Carole S that a properly de-rated e-cap should not be a weak link.
...aaand I should probably put this comment up in the article as an update.
As always, I appreciate your comments as to where I should be clearer, as well as your own insights as to design and technology selection decisions.
Lejaune commented:
I can see someone will carry this "remote phosphors" concept one step further and paint the ceilings or walls or even driveways with the phosphor paint and shine the bare blue LEDs on to the paint.
John M. commented:
I don't see any seal for the plastic cover to heat sink joint -- so it must only be indoor use?
B. commented:
Agreed with Alan Re: robustness of controller/inverter; agreed with desire to see frequency spectrum output(of fully assembled device); agreed with skepticism in lifetime of capacitors/non-LED components. Even at $40, I suspect there is a design trade-off to get to that cheap a price.
Overall, I'd like to see a commercial/residential solution that's sold in two pieces: the controller and the LEDs. That way you can replace them separately as they go bad. If that also means buying new fixtures to replace old edison format, I will phase them out.
To the person who wants to wait until prices go down -- the lower price denotes risk of skimping on the controller/inverter components and thus shortened overall lifetime of "bulb". Despite creative advertising of "LEDs last 20 years", you may as well read: "controller lasts 1 year". The cheap end ones will give the LED bulb market a bad name if they start shipping crappy controller/inverter components. CFL bulbs did this exact thing.
Carole_S commented:
Electrolytic capacitors will have been chosen to match the expected degredation performance of the LED's so shouldn't determine the lifetime of the product. The Cypress solution isn't tremendous in controlling input harmonics, but it does lift the power factor and provide good controllability by phase-cut dimmers. Domestic applications usually have a lifetime that represents a 50% degredation in light output having been operated at full output in a 25C ambient. Commercial applications usually have a 30% degredation (70% of full output remaining) level.
This type of circuit produces a good correspondance between power in and light out, not like incandescents or CFL's where you can wind up using 40% power for 5% light out.
Radiated EMI is generally less than CFL's because the PSU output is DC not high frequency AC over a fairly long path.
Laserdude Phil commented:
H8 2 S8. This thing weighs like two pounds (re: heatsinking question.) but I don't care, I'm going out to buy another 4 B4 they run out again. :)
Mel commented:
Before you add power conditioning. Remove low cost photo switches a big sorce of RFI (Radio Frequnicy Interferance. Next take out all neon lighting and arc welders and heaters. If yout building has a steel frame or metal studs you may need a 2 remote antennas, one within line of sight of cell tower and one connected inside your shelded area. Additional ground rods around the building will reduce noise.
K L commented:
IS THERE ANY GALLIUM IN THE PHOSPHORS ?
THANKS KL
Neil commented:
I don't agree with a previous comment that we have come a long way in lighting.
Seems that we have replaced a simple easily manufactured item with a complex, expensive one.
It would interesting to do an energy audit of the total manufacturing process ( from mining, refining, manufacture of components to final assembly) and compare it to a similar audit for an incandescent globe.
I would hazard to guess that over the life of the globe (see ref. to caps) they consume more energy than an incandesent.
I'd love to be proved wrong.
Steve B commented:
Very interesting design indeed. It would be good to see a plot of the radiation pattern around the periphery and across the top of the bulb. I also wonder what the dispersion angle for the phosphor is; particularly, how much of the light emitted from the phosphors is "lost" into the phosphor layer or radiated back toward the LED?
I am using R-30 LED bulbs in a hallway at my home and must say that so far I am quite pleased with them. I replaced 2ea 28W CFLs with 2ea 10.5W LEDs, glare is eliminated and the light comes on full brightness with the switch. I understand that CREE provides the LEDs for this fixture as well.
OJBIII commented:
Where do you get 10 2500hr 60 W bulbs for four bucks?
kokhon.ng commented:
Nice n detail explaination and tear-down. Great work!
John commented:
Wow, these teardown analysis showed how much these LEDs can deliver even with the plastic covering! By the way, can I have a closer look at the LED engine? the assembly portion with the connector?
Steven D commented:
Margery -- Thanks for the continued good articles on incandescent replacement bulbs. One piece of information not given: Who manufacturers the LED? From your past articles, it seems that CREE is the only one with a high quality LED that will last (and not loose half its output in a short time)
William commented:
What is the watts to lumens efficiency of the bulb at 100% and lets say 20% brightness? How does it comepare to other bulbs?
Matthew S commented:
Great teardown. I agree that solder connections should be more reliable than connectors. Lifetime is probably mostly limited by the electrolytic cap. The one thing I think is really clever here is the LED heatsinking, which you didn't mention. It looks like the U-shaped metal pieces act as heatsinks, keeping the LED temperature down. This is one of the major problems making LED bulbs that fit into incandescent sockets.
Bob Downs commented:
I love the LED concept and plan to use them in
my home....after the price comes down to realistic levels. Current prices for them just
make no sense for almost all applications.
ecdy commented:
Connectors don't make sense for a nonrepairable commodity like a light bulb. They're too expensive and in fact less reliable than good solder connections. If Philips is using them it must be for a purpose--such as dealing with yield problems on LED boards. The CFL board you showed was more probably wave-soldered than hand-soldered although somebody might have had to touch up a connection or two.
Deo Shenai commented:
Nice tear-down and really illuminating reverse engineering. We have have come a long way from Edison's light bulb (patented in 1880) to more powerful LED bulb of today. The photos and clear explanation makes it easy to understand how LED bulb works. Thanks.
Steve A commented:
One important point was missed, I think. What is the color temperature of the light it emits? Anything much hotter than the 2700K or so from a standard incandescent bulb is annoying to many people.
GM commented:
I would have thought that a soldered connection, if done properly, should be definitely more reliable than a connector!!!
Dan Aquinas commented:
Interesting article!
Personal experience says CFLs are good in open air/free standing installations, bad (i.e, don't do it) in enclosed boxes (think ceiling lights). Any guidance from Philips on installation restrictions? I could not find any such info listed at the Home Depot site you referenced. I did see that bulb is free of Hg and Pb, and has a rated life of 25,000 hours. So it is only 10x more expensive than buying a pack of 10 incandescent 60W bulbs at menards.com (assuming a 2500 hour life).
Kyle B commented:
Ahhh Too bad they chose to use EL caps. That'll probably be the achilies heel for lifetime of this design.
GKaye commented:
Margery,
Thanks for doing the reverse engineering on these gizmos. That looks like a ton of fun.
How about doing an analysis of RFI that these things give out? AM and FM radio is still here and many of us still use it.
I live in a building full of LEDs and CFL's and I can't get a radio station, AM or FM on my clock radio and my Ibook G4 laptop system crashes in there too, I assume from the rf hash. It is the only location I get the dreaded Apple grey screen that says "you must restart your computer". I'm guessing the wifi card is getting swamped.
I think it would be useful to do a spectrum analysis and a practical use analysis with a standard $29 radio, to see how far you have to get away from the lamp to get decent reception.
Also, how much does this pollute the power line with harmonic distortion and RF?
Thanks,
George
Zappy commented:
I second that. Great explanation and photos!! Nice to also know that there is still some quality out there too! Phillips should send you some extra bulbs for exposing their great work!!
gus in denver 99gus commented:
Now that you have wrecked the bulb...... may I have it ? heh heh
Alan commented:
Interesting dissection. I do have one semi-cranky question about all of these LED bulbs. The stated savings require the full rated life of the bulb. However, having just seen a couple of different electronic gizmos fried by power interruptions/surges, does any of us have sufficient confidence in the controller chips buried in these LED bulbs to believe they'll actually last as long as needed to break even on the cost? I, for one, am a skeptic unless I'm going to start adding a whole bunch of power conditioning stuff all over my house.
Andy T commented:
Nice work Margery....thanks!
