LED bulbs keeps dying

[ericmark]

All my LED bulbs are cheap, I bought expensive Philips CFL bulbs and within a year out of 16 bulbs half had failed, they were replaced to start with using cheap Home Bargains CFL bulbs, now Lidi and B&M Bargains LED bulbs. I have learnt a lesson, buying quality bulbs does not mean they last longer, so now cheap bulbs every time.

 

[ban-all-sheds]

Plastic housing (unacceptable)

In what way?

 

[aptsys]

In what way?

Heatsinking

 

[Iggifer]

Heatsinking

Why can they not have a heat sink shrouded in plastic?

Osrams LED lamps are all externally plastic. Hardly a cheap/low quality brand

 

[aptsys]

Why can they not have a heat sink shrouded in plastic?

Osrams LED lamps are all externally plastic. Hardly a cheap/low quality brand

They make both cheap and expensive lamps. The plastic ones are designed for lower duty cycles - you do not get low thermal resistances between heatsink and air if you enclose in plastic, so once the system reaches equilibrium the LEDs are running at really quite a high temperature. Most of the time, this is OK since the lamp will be designed for 2 or 3000 hours life.

 

[Iggifer]

https://www.osram-lamps.com/ecatalo...assic-bulbs/led-superstar-classic-a/index.jsp

25,000h rated life, 100,000 switching cycles and plastic shrouded heatsink. First lamp I checked.

I wouldn’t call that lower duty

 

[JohnW2]

https://www.osram-lamps.com/ecatalo...assic-bulbs/led-superstar-classic-a/index.jsp25,000h rated life ...

I wonder how they arrive at these specs. Do you really believe that they run them for nearly 3 years before selling any of them with that spec? There isn't really any way I can think of that one can do an 'accelerated' test on something like an LED lamp.

Kind Regards, John

 

[aptsys]

https://www.osram-lamps.com/ecatalo...assic-bulbs/led-superstar-classic-a/index.jsp

25,000h rated life, 100,000 switching cycles and plastic shrouded heatsink. First lamp I checked.

I wouldn’t call that lower duty

Interesting specs, but it could be suggested that they mean the lamp may only be designed to be turned on for 15 minutes at a time.

Find me an electrolytic capacitor rated for 25000 hours at elevated temperatures please.

 

[Iggifer]

I wonder how they arrive at these specs. Do you really believe that they run them for nearly 3 years before selling any of them with that spec? There isn't really any way I can think of that one can do an 'accelerated' test on something like an LED lamp.

Kind Regards, John

Yea I don’t know. Switching cycles is obviously easy to test in a relatively short space of time.

At this point in the life cycle of LEDs generally I think they probably have a reasonable knowledge of how long their products last. They presumably did a lot of R&D in the early days and now know that X driver or X chip lasts X hours and they can come to a compromised guesstimate - as that’s all they ever are as to how long a lamp is gonna last

 

[bernardgreen]

Find me an electrolytic capacitor rated for 25000 hours at elevated temperatures please.

It need not be an electrolytic, given the small amount of power is being stored a dry capacitor could be used, especially as long life time electrolytics are expensive

 

[aptsys]

It need not be an electrolytic, given the small amount of power is being stored a dry capacitor could be used, especially as long life time electrolytics are expensive

I can confirm these Osram lamps have a standard switcher inside with an electrolytic capacitor after the bridge rectifier.

 

[JohnW2]

They presumably did a lot of R&D in the early days and now know that X driver or X chip lasts X hours and they can come to a compromised guesstimate - as that’s all they ever are as to how long a lamp is gonna last

I wouldn't say quite that. The figures quoted are obviously probabilistic, so mean nothing in terms of an individual lamp, but if full-life testing is undertaken on a substantial sample of products one will get a decent estimate of the average time to failure (or, which is what they usually quote, the time taken for 50% to fail).

Kind Regards, John

 

[aptsys]

Normally they'd run a large sample in parallel and apply some statistics to the failures that occur, but under what conditions? Ambient temperature, orientation, airflow etc. Almost impossible to provide a representative number let along the fact that the rated lifetime exceeds the lifetime of some of the components in the device. The rated lifetime is probably marketing crap suggesting the lifetime of the individual LED dies.

 

[ban-all-sheds]

Interesting specs, but it could be suggested that they mean the lamp may only be designed to be turned on for 15 minutes at a time.

I would read it as 25,000h or 100,000 switching cycles, whichever comes first.

Maybe it means that each switching cycle knocks 15 minutes off the life, i.e. the only way you'd get to 25,000 hours is if you never turn it off.

If they rate switching in the same way as life, AIUI that means that by 12,500h/50,000 switchings half of the lights will have failed.

 

[JohnW2]

If they rate switching in the same way as life, AIUI that means that by 12,500h/50,000 switchings half of the lights will have failed.

As far as I can make out, that is the usual way that the express life expectancy. It's probably easier to interpret than MTBF, since the latter can be biased downwards by very early failures (usually 'quite common') or biased upwards by a subset of the sample which refuse to ever die!

Kind Regards, John