Philips LED bulb teardown

[reagle]

The thing conveniently decided to stop working, so naturally I had to take it apart!
What sets this one apart from many others is a use of remote phosphor. The LEDs are blue and then a plastic sphere converts that into warm white light, emitted in all directions. Makes for a very incandescent-like look (when it works )
Edit- a few more pictures added
Full teardown is at http://kuzyatech.com/philips-8e26a60-led-bulb-teardown

[TheWelly888]

Those Philips must be very expensive - the "One Hung Low" brand of LED GU10 lamps I used tripped the lighting circuit breakers after only a short time. I cracked it open and there was only a tiny SM psu with a 105C capacitor and no EMI suppression at all in there. I'll dig the photos out if anyone wants me to.

[reagle]

Oddly enough, I bought it for $21 a year ago. Next gen ones are $17 at Home Depot!

[bilko]

The thing conveniently decided to stop working, so naturally I had to take it apart!

It looks like they are using their Rebel LEDS. How many hours did it run for before it failed ?
They quote 50,000 hours +
I've tested the rebels and not many make 3000 hours.
I think the 50,000 hours is when they are switched off 

[reagle]

Yes these are Blue Rebels. They mostly died from overheating- the bulb was in a light fixture outside and on all night for a year. So that's about 2900 hours.
The remote phospor sphere was semi-melted and LED lenses were all cracked but one. I don't think the LEDs were actually in happy thermal situation, so can't expect them to last much..Even the 130C caps are only 4000 hours rated by Rubycon

[FenderBender]

Those Philips must be very expensive - the "One Hung Low" brand of LED GU10 lamps I used tripped the lighting circuit breakers after only a short time. I cracked it open and there was only a tiny SM psu with a 105C capacitor and no EMI suppression at all in there. I'll dig the photos out if anyone wants me to.

Oh yeaaa sure!! Let's see it.

[FenderBender]

Is that gunk on the board from it dying or is that how it came from the factory.

None the less, good job Phillips. Looks good.

[bilko]

Maybe not so good Philips. They could have incorporated a thermistor to avoid the overheat problem. Poor design, but there again, well done  Philips for designing a product that will expire out of warranty and for the choice of cap that would have made that happened if the led didn't fail first.

I don't get how companies can quote 50,000 hours and then give a 12 month warranty. 5 years would be more appropriate.

[reagle]

The gunk in the driver board is probably from potting. The warranty was actually pretty nice- 6 years based on 4 hours a day .

[tom66]

You have to remember 4000 hours at 130C for an electrolytic is impressive.

At 110C that's approximately 16000 hours. Conventional electrolytic caps manage a peak of 8000 hours at 105C. And at an operating temperature of say 70C, that would easily meet the 50000 hour spec. Rubycon make good caps.

[bilko]

The gunk in the driver board is probably from potting. The warranty was actually pretty nice- 6 years based on 4 hours a day .

So Phillips actually replaced your lamp F.O.C. ?

[reagle]

Better- they  sent me a refund.
I called their support line, described problem, gave them some numbers off the bulb and receipt and had a check in my hands in less than a week.
So I get a new bulb and a teardown target- they did not want this one back!
as far as thermal control- I don't know how many companies actually implement that. Maybe Cree in their systems that correct color temps based on LED temperature.  In this case the driver is an NXP SSL2101T part, that's mainly concerned with driving an LED load and coexisting with dimmers. Didn't see any control loops there for thermals, plus it'd have to sense it based on the diodes's forward drop behavior change- the only signals connected are two lines to the LED string.

So Phillips actually replaced your lamp F.O.C. ?

[FenderBender]

Maybe not so good Philips. They could have incorporated a thermistor to avoid the overheat problem. Poor design, but there again, well done  Philips for designing a product that will expire out of warranty and for the choice of cap that would have made that happened if the led didn't fail first.

I don't get how companies can quote 50,000 hours and then give a 12 month warranty. 5 years would be more appropriate.

Good in comparison to what I've seen. Some of the crap CFL boards and one hung LED drivers. Of course it could've been better, but I guess these bulbs are already expensive to start with. Trying to convince my parents to screw the CFLs.Those thing are like mercury bombs. Real great for the planet, just like all florescents I guess...but still..

[T4P]

They aren't very expensive as suggested in an early thread ...
BUT
They suck at efficiency still, i remember reading somewhere about 80lm/W ?
Mind you! My CFL's are 68lm/W and is running 24/7 for a year already ...

[reagle]

Agree on CFLs- I've pretty much replaced all of them with LEDs. In many places I get better color rendering, instant on and overall uniform lighting. Kitchen had PAR30 CFLs  in recessed light cans. You'd walk in, turn the lights on and wait till they warm up. And then they'd all vary slightly in color.  Now I have Cree lamps in there- instant on and very much identical throughout. And no mercury- I hate the idea of a flimsy glass thing that if dropped spews that around.

[nukie]

The remote phosphor and the use of 'blue' led and mixed with small ratio of red led help Philips bulbs win the LED bulb challenge. The setup results in not only high lumens output but also high color rendition(RA) which is close to fluoro tubes.

Nichia makes high output, high RA LEDs easily kill Lumileds. Philips manage to pull it off without using competitor's LED in the bulb design.

In terms of LED tech, Philips has fallen behind since the day of Luxeon I

[SeanB]

If they just sent a cheque then they do know that the warranty about life is false, but think that few enough consumers will complain about short life, and have proof of purchase that is legible ( thermal print fades in a year normally) that refunding them is cheaper than redoing the design to actually make it last as long as the marketing claims. Would be interesting to see the class action suit for this one.

[RCMR]

I'm not so much of a fan of Philips lighting any more.

I used to use a 200W planet-warming incandescent bulb in my office when I wanted to shoot video.

It was great -- lots of light and in winter, even kept the place warm.

Unfortunately, the bulb died and when I went to replace it, I discovered that they don't make 200W incandescents any more -- the planet is already too warm apparently.

So I tried some 22W CFLs but they were not very good.

Then I noticed that Philips was pushing its EcoClassic halogen bulb in a 140W size which claimed to have a light output "similar" to a regular 200W incandescent.

Woohoo... the same light for 30% less power *and* I could actually buy one -- so I did.

What a load of BS.

The light output is around the same as a 22W CFL -- which is really about the same as a 100W incandescent.

There is no way on earth that this bulb comes anywhere near close to the light output of a 200W incandescent -- what a rip-off.

In fact, I did some subjective back and forth swapping between a 100W incandescent and this 140W halogen and found that I could *barely* notice the difference.  So it seems I was actually spending 40% *more* power to get about 10% more light than a 100W bulb.

Bugga!

[T4P]

The remote phosphor and the use of 'blue' led and mixed with small ratio of red led help Philips bulbs win the LED bulb challenge. The setup results in not only high lumens output but also high color rendition(RA) which is close to fluoro tubes.

Nichia makes high output, high RA LEDs easily kill Lumileds. Philips manage to pull it off without using competitor's LED in the bulb design.

In terms of LED tech, Philips has fallen behind since the day of Luxeon I

 .... http://www.nichia.co.jp/en/product/led.html?op=cond=type='NSPW510HS-K1' 170lm/w

If they just sent a cheque then they do know that the warranty about life is false, but think that few enough consumers will complain about short life, and have proof of purchase that is legible ( thermal print fades in a year normally) that refunding them is cheaper than redoing the design to actually make it last as long as the marketing claims. Would be interesting to see the class action suit for this one.

And some people actually photocopy every single receipt they have

I'm not so much of a fan of Philips lighting any more.

I used to use a 200W planet-warming incandescent bulb in my office when I wanted to shoot video.

It was great -- lots of light and in winter, even kept the place warm.

Unfortunately, the bulb died and when I went to replace it, I discovered that they don't make 200W incandescents any more -- the planet is already too warm apparently.

So I tried some 22W CFLs but they were not very good.

Then I noticed that Philips was pushing its EcoClassic halogen bulb in a 140W size which claimed to have a light output "similar" to a regular 200W incandescent.

Woohoo... the same light for 30% less power *and* I could actually buy one -- so I did.

What a load of BS.

The light output is around the same as a 22W CFL -- which is really about the same as a 100W incandescent.

There is no way on earth that this bulb comes anywhere near close to the light output of a 200W incandescent -- what a rip-off.

In fact, I did some subjective back and forth swapping between a 100W incandescent and this 140W halogen and found that I could *barely* notice the difference.  So it seems I was actually spending 40% *more* power to get about 10% more light than a 100W bulb.

Bugga!

Halogens are also about 10lm/W best case 20lm/W but that also applies to incandescents, but it's interesting to note that halogens have a negative coefficient
so they are best when cold, people still use it for it's colour but it is so hot ... you can stand 2meters away and still feel the heat

Well i have seen 100W CFL's before ... let alone 40W

[SeanB]

If you want a decent video light you buy a LED video panel, and use it. Costs a lot more than the 200W lamps, but lasts essentially forever. If you want a cheap light go to Home Depot and buy a halogen work light for cheap, and a few spare halogen lamps to go with it. Enough light to take photos with a welding glass in front of the camera lens.

[bilko]

If they just sent a cheque then they do know that the warranty about life is false, but think that few enough consumers will complain about short life, and have proof of purchase that is legible ( thermal print fades in a year normally) that refunding them is cheaper than redoing the design to actually make it last as long as the marketing claims. Would be interesting to see the class action suit for this one.

Good on Phillips for sending the refund. I know companies the install architectural lighting on bridges, viaducts e.t.c. They use led lighting because of the quoted 50,000 hours. If they have to organise closing roads, hiring cranes and cherry pickers to replace failed units I'm sure they will be back at the manufacturer's claiming costs.

[tom66]

A warranty penalises a company for producing a product which fails to last. It's good marketing too.

[bilko]

Many years ago I used to build and sell custom PC equipment for industrial automation. We were buying disk drives with 150,000 power on hours quoted (That's over 17 years). They offered 12 months warranty. I mentioned that they mustn't have much confidence in their products if they are only good for 12 months, after much discussion they extended the warranty to three years. 

[SeanB]

I would guess the industrial units are a lot more robust, and most likely well heatsinked. The LED streetlighting is a lot more reliable, there have only been a few total failures since they were installed for the WC in 2010, though there are a fair number of dead modules. Having 3 per fitting does help with the reliability, if you consider a failure as all 3 being dead.

I have LED indicators that i made, that are still running 10 plus years later, being on near 24/7/365. To get that though I used 2 39K 0.5W resistors in series and a reverse diode, and potted the whole lot in epoxy resin.

[bilko]

I would guess the industrial units are a lot more robust, and most likely well heatsinked. The LED streetlighting is a lot more reliable, there have only been a few total failures since they were installed for the WC in 2010, though there are a fair number of dead modules. Having 3 per fitting does help with the reliability, if you consider a failure as all 3 being dead.

I have LED indicators that i made, that are still running 10 plus years later, being on near 24/7/365. To get that though I used 2 39K 0.5W resistors in series and a reverse diode, and potted the whole lot in epoxy resin.

The HDD were standard, but shock mounted.
Which leds did you use and what current were you driving them with ?

[TheWelly888]

The One Hung Low cheap LED lamp jobbie that I mentioned earlier in the thread, here are the pictures:

It has no brand marking whatsoever. My father got them from an ebay seller for only £2 each! The electrolytic caps are rated for 105C but they did not appear to have gone bang. Note the complete absence of any EMF suppression component or potting. The actual LEDs were white and they no longer work.

[SeanB]

I used 2 39K resistors in series to drive one of the then new ultrabright red led's.Current was around 2mA, and it was bright enough at that current to light the opposite wall in low ambient conditions.

[reagle]

re: Remote phosphor. Lighting Science Group also  seems to be using that in their Definity A19 bulb (I call it a flying saucer on top of a cactus model). The weirdly shaped plastic top is what emits light omni-directionally, though I am not sure whose LEDs are in it.
(http://www.amazon.com/Lighting-Science-Group-A19-Non-Dimmable/dp/B004W8EVMO)

[reagle]

The One Hung Low cheap LED lamp jobbie that I mentioned earlier in the thread, here are the pictures:

It has no brand marking whatsoever. My father got them from an ebay seller for only £2 each! The electrolytic caps are rated for 105C but they did not appear to have gone bang. Note the complete absence of any EMF suppression component or potting. The actual LEDs were white and they no longer work.

Yikes  what a mess, especially the burned up power leads! Quite a contrast to a clean Philips build I must say. The LEDs might be Chinese as well. While at the last years Lighting Fair I've seen a lot of lights that used their own knock-off LEDs in them.

[SeanB]

I bought the 3W 12V MR16 versions of those. Ripped them apart and remote mounted the nice little ballasts, so I could use them as bench lights. Work nicely, and the ballast now does not get toasted, and as well I added some ferrite beads to reduce EMI. Added a third lead to ground the metal heatsink as well, something that is sadly missing on all of these, and I doubt if they would pass if I was to do insulation testing on them. They do have a nice heatsink though, quite effective if not enclosed.

[reagle]

It appears many SSL fixtures these days are still mainly DIY thing- get it, tweak it and then it may last. Case in point- I use nice looking Utilitech undercabinet lamps. they come with a substantial metal case/heatsink. The LED strips are on metal substrate as well. Sounds good, right? Well, except the strips attach to the heatsink with a few screws and leave most of the contact surface open to a large air gap. Nothing a bit of thermal grease can't fix, but what if I did not open it when purchased??
See http://www.lowes.com/pd_102996-2120-29125_0__  for an example.
I may have to do a blog post on silly things found in SSL at this rate

[SeanB]

Oddd?

Access Denied
You don't have permission to access "http://www.lowes.com/" on this server.

Reference #18.1f4d212e.1344183700.55696f5

[T4P]

The One Hung Low cheap LED lamp jobbie that I mentioned earlier in the thread, here are the pictures:

It has no brand marking whatsoever. My father got them from an ebay seller for only £2 each! The electrolytic caps are rated for 105C but they did not appear to have gone bang. Note the complete absence of any EMF suppression component or potting. The actual LEDs were white and they no longer work.

Yikes  what a mess, especially the burned up power leads! Quite a contrast to a clean Philips build I must say. The LEDs might be Chinese as well. While at the last years Lighting Fair I've seen a lot of lights that used their own knock-off LEDs in them.

The leds are chinese
But as always, pay peanuts and get peanuts

Speaking of LED fixtures my town council decided to retrofit the previous E14 18W globes
with a 40W LED in the shape of a 20W tube ... from optiled (chinese company!)
Remote phosphor or not i don't know but they cost 80$ each and i don't really understand why they want to do so
1) My CFL's i used in the past that i did not pay peanuts for can last 2 years 24/7
2) There might be power savings but electricity isn't a few bucks/kwh ...
3) CFL's that aren't peanuts cost AT MOST $10
and i doubt the driver will be able to last more than a CFL can ... 50,000 hours?
Case in point, because everytime they swapped out the CFL
they are spending 20 bucks plus these LED's are supposedly 2 times stronger
so if they wanted that much lighting in the past they had to spend 40 bucks, 2 times of 1 year half pays for the LED
AND as expected they increased the amount we have to pay as "conservation fee", They said living in the government flats you don't have to pay the town council ... ya right, it's like living in a condominium where you also have to pay a "conservation fee"

[SeanB]

2 years on a CFL. I use PL lamps and magnetic ballasts. They last at least 5 years on 24/7/365. Then you replace the cheap, easy to buy and simple to change lamp, and the expensive wound ballast starts the new lamp with no problems for another 5 years plus. The ones on a daylight switch did over 10 years minimum.

[T4P]

2 years on a CFL. I use PL lamps and magnetic ballasts. They last at least 5 years on 24/7/365. Then you replace the cheap, easy to buy and simple to change lamp, and the expensive wound ballast starts the new lamp with no problems for another 5 years plus. The ones on a daylight switch did over 10 years minimum.

Oh yeah, i don't normally condone CFL for anything permanent, i have a magnetic ballast'd PL on top of my head too
Those things last forever ... i change them 2 years once, when it begins to dim to 80% or so i think, don't have a luxmeter don't know.
But it's worth getting

[FenderBender]

Oddd?

Access Denied
You don't have permission to access "http://www.lowes.com/" on this server.

Reference #18.1f4d212e.1344183700.55696f5

Lowes is an American home improvement chain. They might only allow people to view their site from the US. I'm not sure why.....

[reagle]

Oddd?

Access Denied
You don't have permission to access "http://www.lowes.com/" on this server.

Reference #18.1f4d212e.1344183700.55696f5

Do they block by country?! Screw them here is the picture

[flolic]

Do they block by country?! Screw them here is the picture

I am in Croatia and can access their server just fine.

[SeanB]

They just do not like ZA, must be because we beat them in swimming........

[T4P]

They just do not like ZA, must be because we beat them in swimming........

Well that's unfair ...
Oh by the way, did china win swimming?
where's my axe

[FenderBender]

They just do not like ZA, must be because we beat them in swimming........

Well that's unfair ...
Oh by the way, did china win swimming?
where's my axe

China win swimming?  Overall?

[WorldPowerLabs]

The thing conveniently decided to stop working, so naturally I had to take it apart!

It looks like they are using their Rebel LEDS. How many hours did it run for before it failed ?
They quote 50,000 hours +
I've tested the rebels and not many make 3000 hours.
I think the 50,000 hours is when they are switched off 

Agreed on the Rebels... they were promising, on paper -- but failed to deliver longevity (even with a very satisfactory heatsink).

The thing that gets me is this:  the alternative-bulb makers (non-incandescent) all promise long life, in exchange for a high initial purchase price.  In my experience, most fail to deliver.  My overpriced CFLs fail at about the same frequency as the incandescents they replaced...

[M. András]

anyone tried the cree plugin led lamp modules or heck one together using bridgelux mega led arrays? 4500lumen light output at 4k color temp while consuming 5x watts datasheet says its around 106lumen/w. i know a forum member did a room lightin using the cree cxa2011 arrays driving them at their test numbers

[FenderBender]

The thing conveniently decided to stop working, so naturally I had to take it apart!

It looks like they are using their Rebel LEDS. How many hours did it run for before it failed ?
They quote 50,000 hours +
I've tested the rebels and not many make 3000 hours.
I think the 50,000 hours is when they are switched off 

Agreed on the Rebels... they were promising, on paper -- but failed to deliver longevity (even with a very satisfactory heatsink).

The thing that gets me is this:  the alternative-bulb makers (non-incandescent) all promise long life, in exchange for a high initial purchase price.  In my experience, most fail to deliver.  My overpriced CFLs fail at about the same frequency as the incandescents they replaced...

Very true. I think initially when CFLs were first coming out, companies might have been producing good quality ones because the technology was still new and had to prove itself...but over the years, they've been cutting costs left and right. Real crap quality today. I don't know if anyone makes a good CFL these days. Maybe some of the big brand American companies like Phillips or Sylvania...but no promises with them either.

[madworm]

Well, in the beginning incandescent bulbs lasted quite long too. But there was a global 'conspiracy' to limit lifetime to about 1000 hours. Bulb manufacturers even had to pay a 'fine' to a consortium if their bulbs lasted for too long.

I think the documentary that covers that is called 'Pyramids of Waste (2010)'.

We'll have to pay attention to the lifetime hours that the LED bulbs claim to last.

[FenderBender]

I agree. Ahhhhh consumer electronics. What a bunch of rubish. Now we can't even buy a good lightbulb for heaven's sake.

[A Hellene]

Well, in the beginning incandescent bulbs lasted quite long too. But there was a global 'conspiracy' to limit lifetime to about 1000 hours. Bulb manufacturers even had to pay a 'fine' to a consortium if their bulbs lasted for too long.

I think the documentary that covers that is called 'Pyramids of Waste (2010)'.

Exactly! The documentary is called The Light Bulb Conspiracy (2010):

There is a fairly new documentary about planned obsolescence, called The Light Bulb Conspiracy (2010) and it can be watched on YouTube:



The documentary begins with an extended reference to the first organised successful attempt of planned obsolescence by the Phoebus Cartel in 1924, whose members, that were manufacturing light bulbs, 'fixed' the life expectancy of their bulbs at less than 1,000 hours.
[...]

Now, I have recently purchased for €7.50 an E14 2W LED lamp with 36 5050 SMD LEDs and a promised average lifetime of 40,000 hours, for my kitchen extractor-hood. Well, the LED lamp failed in less than a month's time, with a couple of LEDs having fallen off their PCB(!) and clinking within the lamp bulb! I cracked the lamp open to solder the LEDs back in their position, only to find out that not even one component was properly soldered: All the PCB solder joints were unacceptable in terms of solder quality and quantity, already showing cracks, while the LEDs were only needed to be touched lightly in order for them to fall of the PCB... Of course, a reflow using regular 60/40 alloy solder-wire fixed the problem. I am sorry, though, I did not take any photos of the product and the procedure.

What a waste of the finite earth resources in the name of profiteering...


-George

[nukie]

The Rebels are tough units, how are you testing them and not making 3000 hours. How are they mounted? Are you using the recommended layout and have you included vias in your board for better thermal performance?

I have white Rebels from 2008, they are still working and I over-drive them by 30%. They are not mounted on silly MCPCB, they are mounted on the recommended layout in the datasheet. With the PCB mounted on a passive heatsink with thermal epoxy. Most high power LED failure are due to lack of proper heat dissipation. If the Philips bulb had more fins it would work longer.

The thing conveniently decided to stop working, so naturally I had to take it apart!

It looks like they are using their Rebel LEDS. How many hours did it run for before it failed ?
They quote 50,000 hours +
I've tested the rebels and not many make 3000 hours.
I think the 50,000 hours is when they are switched off 

Agreed on the Rebels... they were promising, on paper -- but failed to deliver longevity (even with a very satisfactory heatsink).

The thing that gets me is this:  the alternative-bulb makers (non-incandescent) all promise long life, in exchange for a high initial purchase price.  In my experience, most fail to deliver.  My overpriced CFLs fail at about the same frequency as the incandescents they replaced...

[T4P]

My LXML-PW11 on a 10C/W under-drived by 10% didn't last long either ... 4000 hours and it just quit working
But my XM-L T6 at 700mA lasted far longer ....

[bilko]

The Rebels are tough units, how are you testing them and not making 3000 hours. How are they mounted? Are you using the recommended layout and have you included vias in your board for better thermal performance?

The leds are bonded onto  aluminium PCB which are then mounted onto aluminium extrusion, all with silver thermal epoxy bonding. They operate at 40 deg C above ambient temperature. Design spec is in excess of manufacturer requirements. All led lamps are manufactured identically using modern pick and place machines and reflow.

So why do some last up to 3000 hours and some fail long before that ?
Is this a led quality control issue ?

All leds were purchased in the UK from Future electronics so I would be extremely surprised if they are fakes.

Show me a rebel led that's survived over 10,000 hours running 24/7/365, ambitious specs, yet to be proven, similar to hard disk drive manufacturers quoting in excess of 150,000 power on hours.

[reagle]

If they were in the kitchen hood, that may be a bit warm environment for them to begin with?

E14 2W LED lamp[/url] with 36 5050 SMD LEDs and a promised average lifetime of 40,000 hours, for my kitchen extractor-hood. Well, the LED lamp failed in less than a month's time, with a couple of LEDs having fallen off their PCB(!) and clinking within the lamp bulb! I cracked the lamp open to solder the LEDs back in their position, only to find out that not even one component was properly soldered: All the PCB solder joints were unacceptable in terms of solder quality and quantity, already showing cracks, while the LEDs were only needed to be touched lightly in order for them to fall of the PCB... Of course, a reflow using regular 60/40 alloy solder-wire fixed the problem. I am sorry, though, I did not take any photos of the product and the procedure.

What a waste of the finite earth resources in the name of profiteering...


-George

[A Hellene]

Not only this; it was also mounted on a slightly vibrating appliance, due to the dual vacuum motor turbines that try to evacuate the cooking vapour and smell.

I understand that I may have used that specific LED lamp in a harsh environment; but its failure in less than a month's time is unacceptable. I neglected to mention that it was RoHS compatible, using the 'hard' tin solder alloy that does not forgive vibrations; now it does, by the removal of the RoHS compatible alloy and the use of the 60/40 one.

My criticism was not directed towards the lack of any restrictions of the product to be used as a fridge/oven lamp or to be working in a vibrating environment. It simply is a product that, in my opinion, would never reach the promised lifetime of 40,000 hours under any circumstances because of the unacceptable design/assembly quality.


-George

[reagle]

Right, I agree completely

[jahonen]

For me it seems that most lamp manufacturers seem to be just copying the lifetime promised for the LEDs themselves, completely ignoring that the el cheapo electronics will fail well before that. I think that there is no need or legal requirement for the manufacturer to proof that their statement is actually true for realistic operation conditions (near roof, closed armature etc.) I can't stop thinking that we need a special water cooled lightning armature for for those promised lifetimes

Regards,
Janne

[T4P]

I am pretty certain that the LED armature's driver outside my flat will die in 3 years ( It's switched on for 12 hours everyday )

[rollatorwieltje]

Has anybody tried some lower power LED bulbs? I'm looking into replacing some 12V 20W spots. I'm a bit worried about the electronic transformer, those things tend to output pure garbage. It's a 100W model with 4 bulbs connected. I'll probably need 4x 3 watt LED bulbs, that's very close to the minimum required for the transformer to turn on (I'm not sure about the exact value required since it's a pita to get to it).
It wouldn't be very economical to install 4x €17 LED bulbs only to see them die very fast, for the same amount of money I could buy a lifetime supply of halogen bulbs.

[SeanB]

The LED lamps are not happy with the crappy electronic transformer, as they literally can output 100V spikes and call it 12V. Best to replace the transformer with a 20W magnetic unit when you convert, and this will ensure best life from decent LED units, provided you allow them air flow to keep them cool.

[FenderBender]

So an electronic transformer is like a SMPS is it not?

[reagle]

There is also a thing called Acrich 2 LED from Seul Semi. They use two long strings of Leds on a single die, and just need a single resistor to run off AC. No Dc-Dc to break. Has anybody played with them yet?

[FenderBender]

There is also a thing called Acrich 2 LED from Seul Semi. They use two long strings of Leds on a single die, and just need a single resistor to run off AC. No Dc-Dc to break. Has anybody played with them yet?

Hmm they look cool. Transformerless converter?

[A Hellene]

Though I have been a fan of CREE for quite a long time with their XM-L series, I'll have to admit that Bridgelux has also drawn my attention with their high CRI and light output. Unfortunately, I have not yet found the time to play with the latter ones...


-George

[T4P]

There is also a thing called Acrich 2 LED from Seul Semi. They use two long strings of Leds on a single die, and just need a single resistor to run off AC. No Dc-Dc to break. Has anybody played with them yet?

Dangerous and risky for the LED's themselves

[T4P]

The LED lamps are not happy with the crappy electronic transformer, as they literally can output 100V spikes and call it 12V. Best to replace the transformer with a 20W magnetic unit when you convert, and this will ensure best life from decent LED units, provided you allow them air flow to keep them cool.

Magnetic? One does not simply use a big inductor for a LED ...

[reagle]

Actually, no converter. Just two really long strings of LEds in parallel but facing opposite ways so that each one is on during a half cycle, and a resistor to limit current. The whole thing is UL approved and pretty simple. Not sure what the failure modes are though

Hmm they look cool. Transformerless converter?

[nukie]

Do you still have the batch no of the Rebels that failed? There was a recall early 2008.

MCPCB thermal performance varies from different manufacturers. Best is stick to FR4 PCB and lots of vias as a mounting platform for these smd power LEDs. PCB with vias have better thermal performance vs MCPCB.

Finally, dehumifying or ovening is required to drive all moisture from the clear dome before solder reflow, moisture in the dome can cause premature LED failure. Follow the datasheet regarding the reflow instructions. A lot of people choose to ignore the steps and end up with a short lived LED.

The Acrich are high voltage *LED modules*, that includes a high voltage IC controller which makes fitting very easy and straight forward.

The Rebels are tough units, how are you testing them and not making 3000 hours. How are they mounted? Are you using the recommended layout and have you included vias in your board for better thermal performance?

The leds are bonded onto  aluminium PCB which are then mounted onto aluminium extrusion, all with silver thermal epoxy bonding. They operate at 40 deg C above ambient temperature. Design spec is in excess of manufacturer requirements. All led lamps are manufactured identically using modern pick and place machines and reflow.

So why do some last up to 3000 hours and some fail long before that ?
Is this a led quality control issue ?

All leds were purchased in the UK from Future electronics so I would be extremely surprised if they are fakes.

Show me a rebel led that's survived over 10,000 hours running 24/7/365, ambitious specs, yet to be proven, similar to hard disk drive manufacturers quoting in excess of 150,000 power on hours.

tapatalk

[reagle]

Another cool company is http://www.luminus.com
Had their rep in, and he showed some really huge LEDs. Some are used in projectors, some in cars and even in UV curing.
http://www.luminus.com/products/CSM-360_11_1694759754.pdfl 40W input power , 6000 Lumen
These are mostly designed to be very concentrated light source, but pretty cool nevertheless

[T4P]

6000lm ? Wow .... WOW
150lm/W 

but my favorite XM-L U2 (which is nowhere to be found ...) does 129.8lm/W @700mA
Or the T6 which does 121.2lm/W @700mA

[SeanB]

An electronic transformer is a SMPS without feedback, so the output voltage is dependant on both load and input voltage. Most output AC direct from the secondary, and rely on the lamp being a resistive load with high thermal inertia to integrate the pulse waveform applied into a RMS approximation of 12V. Unloaded or low load they can have very high voltage spikes, and this will not be liked by a LED lamp that has a bridge rectifier and capacitor on the input, which will be charged up to the peaks. They are designed for an input after the bridge that is from 11 to 16V DC for the non dimmable ones, the dimmable ones work from 5-16VDC. This is often not shown on the package, but is often in fine print in the catalogue.

[nukie]

No where to be found??? There's plenty of U2 bin available since their release never a shortage! I just bought 7pcs like 1 hour ago!!!

6000lm ? Wow .... WOW
150lm/W 

but my favorite XM-L U2 (which is nowhere to be found ...) does 129.8lm/W @700mA
Or the T6 which does 121.2lm/W @700mA

[T4P]

At least on major suppliers 

[M. András]

Another cool company is http://www.luminus.com
Had their rep in, and he showed some really huge LEDs. Some are used in projectors, some in cars and even in UV curing.
http://www.luminus.com/products/CSM-360_11_1694759754.pdfl 40W input power , 6000 Lumen
These are mostly designed to be very concentrated light source, but pretty cool nevertheless

their sst50/sst90 leds are used in flashlights suchs as the olight sr50/sr90/91 series pretty nice beast

[reagle]

their sst50/sst90 leds are used in flashlights suchs as the olight sr50/sr90/91 series pretty nice beast

Wow, they are more of a weapon at that light intensity levels.. 2,250 Lumens 10-30W.

[T4P]

their sst50/sst90 leds are used in flashlights suchs as the olight sr50/sr90/91 series pretty nice beast

Wow, they are more of a weapon at that light intensity levels.. 2,250 Lumens 10-30W.

The flashlights they are in usually have a assault crown ... They are INDEED a weapon
But the LED alone is quite expensive ... about 20-30 bucks

[M. András]

i have an sr50 here with lithium-ion cells inside it, its pretty powerful for that 700 lumen, the big brother lumen value is only for the led itself what comes out of it around 1800 but still pretty high. for that 400bucks it must deliver it all the time btw however its thin aluminium on most of parts its pretty strong and looks nice

[reagle]

Just saw a news release about 3M making a new cool lamp:
http://ledsmagazine.com/news/9/8/21?cmpid=EnliifSeptember62012
The driver is in the dome, and the light is is shaped by the reflective light guides.

[TheWelly888]

^^^
Looks great but the use of the electrolytic smoothing caps worry me - suppose some corporate procurement peon decide to buy a bunch of One Hung Low caps for 10 cents each instead of Reputable Brand for 25 cents each? These blighters will end up with a life of 5 years with a merry ending!

[SeanB]

Thay already do, so what is new. Never had one last the rated life with anything close to the rated lumens either.