(1) Why killing a Golden Dragon with a 25k hour lifespan due to horrid thermal management is 'good' . To be more exact, ifthe LED is being murdered in such a fashion, then why is it assumed that the bulb is going to have it's initial lumen value past half of it's life cycle when it's being cooked to death? JTR1962 might be able to comment on this.
I posted this in another thread but I think it's quite relevant here ( relevant parts in bold ):
The A19 screw-base lamp form factor is limited to
dissipating about 4 watts of heat while still remaining at a temperature suitable for reasonable LED longevity ( i.e. 40,000 hours or more ).
Now here's the fun part. Let's assume the LED ballast has an efficiency of ηb. Further, let's assume that the LED emitter has an efficiency of ηLED. Overall efficiency of the system at converting electricity into light is therefore ηb times ηLED. Let's just call this η for simplicity.
Now let's plug some reasonable numbers in. ηb for a decent ballast could exceed 90%. However, given the size constraints 85% makes more sense ( and this could be as low as 50% for a mass-produced made in China ballast using no name parts ). But let's stick with 85% as we're trying to find the limits of the A19 form factor using state-of-the-art components.
Next up is emitter efficiency ηLED. Unfortunately this is heavily dependent upon drive current. Let's use 700 mA as a compromise between efficiency and number of emitters. The R5 XP-G is currently champ in this department, managing about 120 lm/W by my testing. However, this is under ideal thermal conditions, with the heat sink at close to room temperature. So let's derate this by 10% to 108 lm/W. 108 lm/W equates to an LED wallplug efficiency of roughly 33%, so this is what we'll use for ηLED.
The product of ηLED and ηb is 0.33*0.85 = 0.28 = η. The waste heat is therefore ( 1 - η ) times the AC input power in watts, or 0.72 times input power. This means the maximum AC wattage the lamp can be driven at without exceeding that 4 watt thermal envelope is 4 / 0.72, or 5.56 watts. Of that 5.56 watts, only ηb * 5.56 watts, or 4.72 watts, actually drive the LEDs. The rest are ballast losses. Remember that we earlier calculated LED efficacy to be 108 lm/W at the design drive current of 700 mA.
So this means the lamp will have 4.72 times 108, or 510 emitter lumens. However, you still have optical losses as most bulbs have a diffuser of some type. Assume these losses are around 15%, so we end up with 0.85 * 510, or 433.5 bulb lumens.
This is barely enough to replace a 40 watt incandescent lamp, and this is using the best available emitters and ballast. Now if you want to have better color rendering, figure on reducing your output by roughly another 20%. Now you're down to around 350 lumens. Granted, this is still an efficient bulb in terms of lumens per watt, coming in at around 63 lm/W. But it certainly is nothing to write home about in terms of output.
Now you can end up with
better output and efficiency numbers by moving to
active cooling and/or decreasing the drive current. But both of these things
greatly increase costs. So now you have a great product which you probably can't even sell in numbers high enough to make your investment worthwhile.
In the end once you do the math the conclusion is always the same-
you need to just ditch the screw-base form factor and move to LED fixtures if you want to obtain any reasonable output. In order to get decent outputs rivaling 100 watt incandescents with the current bulb format, you'll probably need to at least double LED efficiencies to around 200 to 225 lm/W. That isn't going to happen anytime soon.
Additional comments:
It's pretty obvious looking at that Osram lamp that the heat sink isn't up to the task of dissipating 8 watts while at the same time keeping the LED cool enough to last more than 25K hours. Note that LED lumen maintenance is
exponentially related to temperature. A mere 10°C increase in junction temperature can reduce life to 50%. Another 10°C on top of that can reduce life to 25%. This is exactly what we're seeing here. The Osram lamp probably operates 20°-25°C warmer than it should. Any well designed LED product should reach at least 100K hours before hitting 70% of initial lumens. Besides the not so great life, efficiency doesn't seem all that great, either. All that being said, I can imagine some niche uses for these. One is where they'll be used outdoors, in mostly cold climate. In that scenario they might actually last over 100K hours. Another might be a hard to change lamp which is frequently cycled. Frequent starts are murder of CFLs, and incandescents have too short of a life to consider using in hard-to-reach areas. Other than those two uses, I'm having a hard time seeing what this lamp does that CFLs don't. It might last 3-4 times as long, but it's way more than 3-4 times the price.
And yes, I know Cree just hit 208 lm/W, but how long until those see mass production? Until emitters with those types of numbers hit the street, LED screw-base bulbs will remain severely limited.
