While I am at it, I am in a good mood to calculate theoretical lumens/watt for "trichromatic white" of a couple favorable CCTs (4100 and 3500 Kelvin).
Red primary - 611 nm, the wavelength of the main red-orange/orange-red emission from triphosphor fluorescent lamps.
Green primary - 544 nm, approx. dominant wavelength of composite of green mercury line and emission of a common green phosphor in triphosphor fluorescent lamps.
Blue primary - 450 nm, the multiple of 10 nm that has highest response of human blue receptors.
Dichromatic whites made by mixing 450 nm with a yellow wavelength can achieve a little more still, although color rendering index would be worse than the 20 of an unphosphored high pressure mercury lamp:
4100K:
To nearest 1 nm, use 573 nm as choice to mix with 450 nm.
.739 watt 573 nm (470 lumens) plus .261 watt 450 nm (8.34 lumens) totals 1 watt with 478 lumens.
3500 K:
To nearest 1 nm, use 576 nm as choice to mix with 450 nm.
.788 watt 576 nm (488 lumens) plus .212 watt 450 nm (6.8 lumens) totals 1 watt with nearly 495 lumens.
> I was simply pointing out that these aren't fly by night hucksters as > you seem to imply, or at the very least, I read into your post, who > know nothing about LEDs and their performance.
> Secondly, measuring the output of a Luminaire is significantly > different than measuring the performance of a single LED. The optical > system of the luminaire would significantly impact the performance of > the LED system as a whole with regards to a specific function. Nearly > every LED manufacturer I've dealt with doesn't maintain the type C > goniometer equipment in their labs that would be required to test > physical luminaires.
> I'm not saying whether their claims are true or not, but you used > tongue in cheek language in you original post, implying that they were > making embarassing claims, and subsequently chose to omit other > information about them that might refute your position. If they made > claims to independent data, and if you want to refute that claim you > should provide either evidence to the contrary or at least try to > obtain the data and indicate that it was refused to you to provide some > credence to your implications.
> I'm sure I'm coming across as a a jerk, but I don't intend to. Just > trying to get all the facts into the discussion. Thanks anyway.
We all have plenty to learn.
Two pieces of advice for you
1. keep errors to those of emission 2. when stuck in the bottom of a hole stop digging
Victor Roberts wrote: > I get 189.6 lm/W over the 380 to 760 nm range. It may be 76 > steps, but I am summing 77 values so I divided the sum by 77 > instead of 76. My data is also in 5 nm steps, but I'm not > sure which revision of the CIE data I have.
Victor,
If you're going for a proper account of things, why not weight the end values (at 380 and 760) by half as much? Then divide the sum by 76.
>> I get 189.6 lm/W over the 380 to 760 nm range. It may be 76 >> steps, but I am summing 77 values so I divided the sum by 77 >> instead of 76. My data is also in 5 nm steps, but I'm not >> sure which revision of the CIE data I have.
>Victor,
>If you're going for a proper account of things, why not weight the end >values (at 380 and 760) by half as much? Then divide the sum by 76.
>Regards,
>Mark
First, I don't see why this is more correct. If a sum includes 77 values then the average is the sum divided by 77. In addition, there are so few lumens per watt at 380 and 760 nm that the results using your method would be equivalent to dividing the full sum by 76.
Using your method I get 192.059 lm/W.
Using the 380 to 760 divided by 76 I get 192.060 lm/W.
-- Vic Roberts http://www.RobertsResearchInc.com To reply via e-mail: replace xxx with vdr in the Reply to: address or use e-mail address listed at the Web site.
This information is provided for educational purposes only. It may not be used in any publication or posted on any Web site without written permission.
>Peter Pan wrote: >> Just some numbers: >> incandescent lamps are 10-20 lm/w >> cfl 40-80 lm/w >> leds are 15-25 lm/w
>Just some thoughts on those numbers:
>You need current regulation for an LED, which brings down the overall >efficacy. Not required for incandescents or cfl+ballast.
>If, for example, a 20 lm/W LED becomes 15 lm/W after adding a current >regulator, then it doesn't beat a household incandescent bulb (15 >lm/W).
You're assuming a current regulator with an efficiency of only 75%. There is no reason why a good LED current regulator cannot have an efficiency of 90% or more, which is similar to that of high quality electronic ballasts.
-- Vic Roberts http://www.RobertsResearchInc.com To reply via e-mail: replace xxx with vdr in the Reply to: address or use e-mail address listed at the Web site.
This information is provided for educational purposes only. It may not be used in any publication or posted on any Web site without written permission.
Victor Roberts wrote: > On 17 Jun 2006 08:22:21 -0700, "redbelly" > <redbell...@yahoo.com> wrote:
> >Peter Pan wrote: > >> Just some numbers: > >> incandescent lamps are 10-20 lm/w > >> cfl 40-80 lm/w > >> leds are 15-25 lm/w
> >Just some thoughts on those numbers:
> >You need current regulation for an LED, which brings down the overall > >efficacy. Not required for incandescents or cfl+ballast.
> >If, for example, a 20 lm/W LED becomes 15 lm/W after adding a current > >regulator, then it doesn't beat a household incandescent bulb (15 > >lm/W).
> You're assuming a current regulator with an efficiency of > only 75%. There is no reason why a good LED current > regulator cannot have an efficiency of 90% or more, which is > similar to that of high quality electronic ballasts.
Vic,
The ones I've seen have roughly a 1V drop. If you run a single white LED (3.5 V) with that, that gives a 78% efficiency. Do you know of regulators with smaller voltage drops, or are you thinking in terms of running strings of LED's on a single regulator?
Victor Roberts wrote: > On 17 Jun 2006 08:08:25 -0700, "redbelly" > <redbell...@yahoo.com> wrote:
> >Victor Roberts wrote:
> >> I get 189.6 lm/W over the 380 to 760 nm range. It may be 76 > >> steps, but I am summing 77 values so I divided the sum by 77 > >> instead of 76. My data is also in 5 nm steps, but I'm not > >> sure which revision of the CIE data I have.
> >Victor,
> >If you're going for a proper account of things, why not weight the end > >values (at 380 and 760) by half as much? Then divide the sum by 76.
> >Regards,
> >Mark
> First, I don't see why this is more correct. If a sum > includes 77 values then the average is the sum divided by > 77. In addition, there are so few lumens per watt at 380 > and 760 nm that the results using your method would be > equivalent to dividing the full sum by 76.
> Using your method I get 192.059 lm/W.
> Using the 380 to 760 divided by 76 I get 192.060 lm/W.
Well, like you said, it's not a big difference in this case. But using 1/2-weighting at the endpoints is equivalent to approximating an integral with the trapezoid rule. See the last equation at this (short) URL:
One of thosevalue that sound fishy, though correct are the ZCR tables where these show higher efficiency than the actual fixture CU. Guess some of the led's crazy values might go with a similar logic.
A led puts out all its flux in a limited solid angle, Then to achieve a similar illumination in the same angle by an incandescent lamp, you need an equivalent incandescent lamp of so many lumens (of course ignoring all the light that doesn't go in the cone of light put out by the led). Therefor those lumens divided by the led power could give those crazy numbers.
The other option is as several have been mentioning, that those led makers are considering all the energy the naked led might be producing throughout its spectrum (with no correction for the eye sensitivity). Guess that if you follow this logic, and incandescent lamp could achieve impressive numbers (though most output is IR!).
In article <nk96929fibrns4hbig8at9mkbfdniva...@4ax.com>, Victor Roberts
<x...@lighting-research.com> wrote: > We also have the small problem that Cree doesn't list any > LEDs with an efficacy even close to 80 lm/W. Since a > fixture for any type of lamp cannot reasonably be expected > to have higher efficacy than the source itself, some > explanation from LLF is certainly in order.
> Don't you think its a bit strange that Cree, who has > apparently gone to great pains to distinguish between the > performance of LEDs that are available for sale and some LED > that might have demonstrated great performance in the their > laboratory, now gives these rare laboratory-grade LEDs to > another company - even if staffed by former employees of > Cree - and lets THEM misrepresent the data to the public?
No I don't find it strange. Several times when developing LED lamps for automotive exterior lighting applications, I have had access to physical samples from the LED manufacturer that were 2 or 3 times the output of the currently available LEDS that were currently released and documented. They gave them to us to help is in our product development cycle knowing that there would be a several month lag between getting first prototypes and actually going into full production. And yes, we would be at the automotive OEMs showing them higher output LEDs than what they could buy on the market. If you have read the papers published in some of the societies such as SPIE, then you've seen published performance of LED headlamp designs using LEDs at levels that were not "documented and released" at the time, so I don't see it as being that uncommon to have someone showing performance for an LED to the public that isn't currently available for purchase from the LED supplier. I don't live around the Detroit area anymore, but I believe the last I heard Visteon was aggressively marketing their LED headlamp capability around the Detroit area using billboards, and my understanding is that some of their performance data was with "non-commercially" available LEDs at the time as well.
It seemed to me that it is a good way for the LED manufacturers to plan their reactor capacity and help move their efficacy along based on a known financial return for making the higher outputs available rather than continuing to try and sell at the current outputs and tying up their reactors at the lower level formulations if there are ready markets for higher outputs.
That's also been one of the issues with using LEDs in automotive is that the 10 year (I think it's 10 years) service cycle required for providing service lamps poses an issue for the LEDs if they obsolete lower output models during that cycle that are needed to keep the lamp legal (lamp has to meet both minimums and maximum specifications).
Peter Pan wrote: > One of thosevalue that sound fishy, though correct are the ZCR tables > where these show higher efficiency than the actual fixture CU. Guess > some of the led's crazy values might go with a similar logic.
> A led puts out all its flux in a limited solid angle, Then to achieve a > similar illumination in the same angle by an incandescent lamp, you > need an equivalent incandescent lamp of so many lumens (of course > ignoring all the light that doesn't go in the cone of light put out by > the led). Therefor those lumens divided by the led power could give > those crazy numbers.
I think you make a good point as to how they might also be making the claim even with the lower output. The specific application here was a downlight, where the LEDs have the advantage of only emitting in the downard direction, and so if you compare to a bulb that could send half of it's flux upward to either be lost, diffusely reflected in no particular direction, or reflected by a designed reflector that absorbs some light with the reflection, then you could get the same output in the downlight application region from the LEDs as a higher output incandescent and make the lumen/watt claim on the equivalent incandescent bulb needed to get the same ouptut pattern. Sneaky....
Boxman <box...@voyager.net> wrote: >That's also been one of the issues with using LEDs in automotive is >that the 10 year (I think it's 10 years) service cycle required for >providing service lamps poses an issue for the LEDs if they obsolete >lower output models during that cycle that are needed to keep the lamp >legal (lamp has to meet both minimums and maximum specifications).
Would it be possible to stay legal by offering as a service part a newer, higher output LED with integral resistor/regulator to reduce its output? Or would that require an expensive recertification of the entire LED- reflector-lens assembly?
>Peter Pan wrote: >> One of thosevalue that sound fishy, though correct are the ZCR tables >> where these show higher efficiency than the actual fixture CU. Guess >> some of the led's crazy values might go with a similar logic.
>> A led puts out all its flux in a limited solid angle, Then to achieve a >> similar illumination in the same angle by an incandescent lamp, you >> need an equivalent incandescent lamp of so many lumens (of course >> ignoring all the light that doesn't go in the cone of light put out by >> the led). Therefor those lumens divided by the led power could give >> those crazy numbers.
>I think you make a good point as to how they might also be making the >claim even with the lower output. The specific application here was a >downlight, where the LEDs have the advantage of only emitting in the >downard direction, and so if you compare to a bulb that could send half >of it's flux upward to either be lost, diffusely reflected in no >particular direction, or reflected by a designed reflector that absorbs >some light with the reflection, then you could get the same output in >the downlight application region from the LEDs as a higher output >incandescent and make the lumen/watt claim on the equivalent >incandescent bulb needed to get the same ouptut pattern. Sneaky....
Since an LED is inherently directional I agree that the efficacy of an LED in a fixture can be equal to the efficacy of the "bare" LED - while the efficacy of a fixture using incandescent, fluorescent or HID sources will always be lower than the efficacy of the bare lamp due to fixture losses.
However, this argument does not explain how an LED in a fixture can have higher efficacy than that same LED when measured without the fixture - unless of course the fixture manufacturer is making up his own definitions of efficacy.
-- Vic Roberts http://www.RobertsResearchInc.com To reply via e-mail: replace xxx with vdr in the Reply to: address or use e-mail address listed at the Web site.
This information is provided for educational purposes only. It may not be used in any publication or posted on any Web site without written permission.
> Would it be possible to stay legal by offering as a service part a newer, > higher output LED with integral resistor/regulator to reduce its output? > Or would that require an expensive recertification of the entire LED- > reflector-lens assembly?
> Matt Roberds
It's primarily left to the lamp manufacturer, but if they are a supplier to just about any automobile manufacturer then their quality system requirements would require them to test and recertify the whole lamp if they changed the design. It's a trade off, in some cases a mid stream retrofit that you describe might be cheaper over the long run even if you add in the cost of extra testing as opposed to paying more up front on every part for a supply design that can be tweaked later without a design change.
>Victor Roberts wrote: >> On 17 Jun 2006 08:22:21 -0700, "redbelly" >> <redbell...@yahoo.com> wrote:
>> >Peter Pan wrote: >> >> Just some numbers: >> >> incandescent lamps are 10-20 lm/w >> >> cfl 40-80 lm/w >> >> leds are 15-25 lm/w
>> >Just some thoughts on those numbers:
>> >You need current regulation for an LED, which brings down the overall >> >efficacy. Not required for incandescents or cfl+ballast.
>> >If, for example, a 20 lm/W LED becomes 15 lm/W after adding a current >> >regulator, then it doesn't beat a household incandescent bulb (15 >> >lm/W).
>> You're assuming a current regulator with an efficiency of >> only 75%. There is no reason why a good LED current >> regulator cannot have an efficiency of 90% or more, which is >> similar to that of high quality electronic ballasts.
>Vic,
>The ones I've seen have roughly a 1V drop. If you run a single white >LED (3.5 V) with that, that gives a 78% efficiency. Do you know of >regulators with smaller voltage drops, or are you thinking in terms of >running strings of LED's on a single regulator?
Mark - I agree that many current LED systems use terrible current regulators. I also agree that providing a high efficacy driver for a single LED is a challenge due to the low voltage drop. However, I expect that any commercial LED system would use multiple LEDs wired at least partially in series to increase the load voltage and the driver used would have an efficacy at least as good as current electronic fluorescent lamp ballasts. Large-scale systems can't afford to throw away 25% of the power if they hope to compete as energy efficient light sources.
-- Vic Roberts http://www.RobertsResearchInc.com To reply via e-mail: replace xxx with vdr in the Reply to: address or use e-mail address listed at the Web site.
This information is provided for educational purposes only. It may not be used in any publication or posted on any Web site without written permission.
> However, this argument does not explain how an LED in a > fixture can have higher efficacy than that same LED when > measured without the fixture - unless of course the fixture > manufacturer is making up his own definitions of efficacy.
The only other thing I can think of besides them having higher output dies than commercially avialable is if they are using only the raw chip and using their own extractor designs to get better performance for a specific task. Cree publishes the performance of a full LED package (chip, optical components, and mechanical package), not the stand alone die chip performance.
Some companies specialize in developing non imaging optical devices directly coupled to the chip that can give higher efficacies for specific tasks than the fully packaged LED efficiency because they can use a better matched index gels with the chip and advanced optical design techniques to optimize the efficiency of light extraction and transfer to the usable output field. If I am remebering correctly, this has been discussed in published papers regarding automotive headlamp design concepts with LEDS and showed real improvements over using the LED manufacturers stock package design.
The most succesful implementers of LED technology will not just be buying stock LEDs and throwing them in a box. They'll need to implement the LED technology with well designed thermal management techniques, and well engineered optics to optimize the light extraction and distribution. In my opinion, they are the ones who will eventually lead the field. The exterior automotive lighting equipment field has been doing this for over 15 years with high brightness LEDs, and LEDs are beginning to get bright enough now for people to consider general lighting applications as well. I'm guessing the same cycle of development will need to occur in the general lighting field as well. Optical design and Thermal management are key technologies that aren't necessarily available to everyone manufacturing luminaires, so the larger manufacturers may have some advantage in this respect. But who knows maybe somebody like LLF is putting it all together and getting it right.
> Victor Roberts wrote: > > On 17 Jun 2006 08:22:21 -0700, "redbelly" > > <redbell...@yahoo.com> wrote:
> > >Peter Pan wrote: > > >> Just some numbers: > > >> incandescent lamps are 10-20 lm/w > > >> cfl 40-80 lm/w > > >> leds are 15-25 lm/w
> > >Just some thoughts on those numbers:
> > >You need current regulation for an LED, which brings down the overall > > >efficacy. Not required for incandescents or cfl+ballast.
> > >If, for example, a 20 lm/W LED becomes 15 lm/W after adding a current > > >regulator, then it doesn't beat a household incandescent bulb (15 > > >lm/W).
> > You're assuming a current regulator with an efficiency of > > only 75%. There is no reason why a good LED current > > regulator cannot have an efficiency of 90% or more, which is > > similar to that of high quality electronic ballasts.
> Vic,
> The ones I've seen have roughly a 1V drop. If you run a single white > LED (3.5 V) with that, that gives a 78% efficiency. Do you know of > regulators with smaller voltage drops, or are you thinking in terms of > running strings of LED's on a single regulator?
First, run series strings of LEDs with one regulator. Second, use a switch mode regulation topology. Unlike a resistive or linear regulators (which incurs an energy loss in the series element), these can operate above 90% efficiency easily.
-- Paul Hovnanian mailto:P...@Hovnanian.com ------------------------------------------------------------------ Opinions stated herein are the sole property of the author. Standard disclaimers apply. Celebrity voice impersonated. Batteries not included. Limit one to a customer. Best if used by April 1, 2006. Refrigerate after opening. Void if removed.
In art. <cc7892liomhoib4hg8in07kfq06970e...@4ax.com>, Victor Roberts wrote:
>On 17 Jun 2006 08:08:25 -0700, "redbelly" ><redbell...@yahoo.com> wrote:
>>Victor Roberts wrote:
>>> I get 189.6 lm/W over the 380 to 760 nm range. It may be 76 >>> steps, but I am summing 77 values so I divided the sum by 77 >>> instead of 76. My data is also in 5 nm steps, but I'm not >>> sure which revision of the CIE data I have.
>>If you're going for a proper account of things, why not weight the end >>values (at 380 and 760) by half as much? Then divide the sum by 76.
>First, I don't see why this is more correct. If a sum >includes 77 values then the average is the sum divided by 77.
If dividing by 77, then I would consider that (189.6 lm/W) to be the value represented by 77 5-nm wide bands - which would be for the 377.5 to 762.5 nm stretch.
> In addition, there are so few lumens per watt at 380 >and 760 nm that the results using your method would be >equivalent to dividing the full sum by 76.
>Using your method I get 192.059 lm/W.
>Using the 380 to 760 divided by 76 I get 192.060 lm/W.
I agree with these - both reasonably valid and apparently to me adequately obvious to be approximations and agreeing well with each other.
If you want fun with some hard-to-achieve ideal, how about a radiator that is 100% efficient and equal-energy-per-unit-bandwidth from 422.5 nm to 692.5 nm and radiating nothing outside that range? I get 1931 chromaticity coordinates from that x=.3381, y=.3426, which is maybe 1/8-1/4 of a fine red hair purpler than roughly 5270 Kelvin but to a much greater extent (still very slight) yellow-greenish compared to "equal energy white". And I would guess the CRI of this "ideal" of mine to be about 99. And Y-bar sum (using 5 nm steps) is 21.2889, divide by 55 for average value of .387071, multiply by 683 to get 264.4 lumens/watt by the 1924 photopic function, and I would guess probably between 264.5 and 265 lumens/watt by the 1988 photopic function.
In article <1150557741.341998.241...@g10g2000cwb.googlegroups.com>,
redbelly wrote:
>Peter Pan wrote: >> Just some numbers: >> incandescent lamps are 10-20 lm/w >> cfl 40-80 lm/w >> leds are 15-25 lm/w
>Just some thoughts on those numbers:
>You need current regulation for an LED, which brings down the overall >efficacy. Not required for incandescents or cfl+ballast.
There are inductive switching current regulators that have losses comparable to that of other "ballasts".
>If, for example, a 20 lm/W LED becomes 15 lm/W after adding a current >regulator, then it doesn't beat a household incandescent bulb (15 >lm/W).
I would say derate 20 to 18, 25 to 22.5, and keep in mind that LEDs with overall luminous efficacy in the mid-upper 30's (typically, with minimum at least mid-20's) are now reasonably available and better than this is soon to come or maybe already landing.
>The LED does beat a few-Volt flashlight incandescent (around 10 lm/W), >with fewer/zero bulb changes thrown in as a bonus.
Along with efficacy not decreasing significantly as the battery weakens, compared to incandescent flashlight lamps having luminous efficacy roughly proportional to battery voltage raised to a power of at least 1.5, maybe closer to 2.
In <1150576553.151718.138...@r2g2000cwb.googlegroups.com>, redbelly wrote:
>Victor Roberts wrote: >> On 17 Jun 2006 08:22:21 -0700, "redbelly" >> <redbell...@yahoo.com> wrote:
>> >You need current regulation for an LED, which brings down the overall >> >efficacy. Not required for incandescents or cfl+ballast.
>> >If, for example, a 20 lm/W LED becomes 15 lm/W after adding a current >> >regulator, then it doesn't beat a household incandescent bulb (15 >> >lm/W).
>> You're assuming a current regulator with an efficiency of >> only 75%. There is no reason why a good LED current >> regulator cannot have an efficiency of 90% or more, which is >> similar to that of high quality electronic ballasts.
>The ones I've seen have roughly a 1V drop. If you run a single white >LED (3.5 V) with that, that gives a 78% efficiency. Do you know of >regulators with smaller voltage drops, or are you thinking in terms of >running strings of LED's on a single regulator?
I know of such circuits having voltage drop more like half a volt. Consider inductive switching ones using power MOSFETs and "catch diodes" being Schottky diodes. Along with the obvious design parameter of higher battery voltages.
I would say that 85% efficiency of "an LED ballast" is low end of someone trying to work at that and shaving costs by so severely as 1/10ths of a cent, and 90% is fairly easily achievable without great cost.
In <1150582606.289557.245...@c74g2000cwc.googlegroups.com>, Boxman wrote:
>Peter Pan wrote: >> One of thosevalue that sound fishy, though correct are the ZCR tables >> where these show higher efficiency than the actual fixture CU. Guess >> some of the led's crazy values might go with a similar logic.
>> A led puts out all its flux in a limited solid angle, Then to achieve a >> similar illumination in the same angle by an incandescent lamp, you >> need an equivalent incandescent lamp of so many lumens (of course >> ignoring all the light that doesn't go in the cone of light put out by >> the led). Therefor those lumens divided by the led power could give >> those crazy numbers.
>I think you make a good point as to how they might also be making the >claim even with the lower output. The specific application here was a >downlight, where the LEDs have the advantage of only emitting in the >downard direction, and so if you compare to a bulb that could send half >of it's flux upward to either be lost, diffusely reflected in no >particular direction, or reflected by a designed reflector that absorbs >some light with the reflection, then you could get the same output in >the downlight application region from the LEDs as a higher output >incandescent and make the lumen/watt claim on the equivalent >incandescent bulb needed to get the same ouptut pattern. Sneaky....
As efficient as they make downlight optics designed for incandescents (whether in the fixture or in the lamp [bulb]), the really outrageous LED efficacy claims in comparison to incandescents are ignoring highly commonly available incandescent downlights that are reasonably designed for such duty!
In <1150587171.679105.260...@f6g2000cwb.googlegroups.com>, Boxman wrote:
>> However, this argument does not explain how an LED in a >> fixture can have higher efficacy than that same LED when >> measured without the fixture - unless of course the fixture >> manufacturer is making up his own definitions of efficacy.
>The only other thing I can think of besides them having higher output >dies than commercially avialable is if they are using only the raw chip >and using their own extractor designs to get better performance for a >specific task. Cree publishes the performance of a full LED package >(chip, optical components, and mechanical package), not the stand alone >die chip performance.
> As efficient as they make downlight optics designed for incandescents > (whether in the fixture or in the lamp [bulb]), the really outrageous LED > efficacy claims in comparison to incandescents are ignoring highly > commonly available incandescent downlights that are reasonably designed > for such duty!
I'm not sure what you meant on your other reply to my other post. If you wanted to see papers on using extractors to increase the efficiency of a chip on board design as opposed to using the standard LED packaging, there is one in the SPIE proceedings on Illumination design from last August if I remember correctly. I beleive it is this one but I could be wrong:
In the paper they were developing an LED headlamp optic and the effciency of the light extractor (not the whole system) relative to the published LED performance was over 100% because they put the optic and index matching gel directly to the chip and improved the extraction for the beam pattern they were attempting to form as opposed to putting optics onto the standard package which already has a pre-formed lens on it. Whether it's practical in mass production, I don't know.
The reason LEDs might offer an advantage over standard incandescent's in terms of an optical efficiency in a downlighting situation is if you consider downlights that have spot angles in the range of 10 degrees to say 60 degrees wide. Then the souce size of the LED provides a significant advantage in terms of being able to transfer the flux into a usable beam. The law of conservation of etendue describes the fundament limit on how much flux you can transfer when you have a given source size and a given solid angle that you are projecting into. The LED source size for a single chip approaches a 1 mm square in some cases. Most standard incandescent filaments are very large compared to that. In some cases the incandescent filament is so large that even at reasonable lamp openings the etendue condition restricts the transfer efficiency into the beam spot.
The reason you see most narrow angle T-5 LEDs having a beam angle of around 4 degrees is that etendue doesn't allow it to be any smaller than that. If you run the etendue calculations with the lens size of the T-5 package and the chip size in the package, you'll find this limit. When you have an extended source and a specific lens size, you can't do anything to the shape that will get you anything better than what the conservation of etendue specifies.
Secondly, the LED can use direct coupling and total internal reflection devices to avoid the reflectivity losses that the incandescent sources will undergo when they use reflectors. A typical metallized glass bulb reflector loses around 15% at the reflection interface. A TIR device has 100% reflectivity because total internal reflection is 100% efficient.
Thus it is conceivable that an LED downlight will use less lumens to get the same resulting output in the beam when compared to an incandescent.
It also depends on the load power, for it does tend to be more difficult to obtain high efficiencies for small power designs.
It is true that working at low voltages cause added complications. However, most of the circuit could be working at a much higher voltage, and the output be at the required voltage (remember 5V computer ps). Depending on the topology they could have a flyback or buck design. Nonetheless, it wouldn't be uncommon for efficiencies greater than 90%.
In article <slrne99mgm.ahm....@manx.misty.com>, Don Klipstein
<d...@manx.misty.com> wrote: > In <1150576553.151718.138...@r2g2000cwb.googlegroups.com>, redbelly wrote:
> >Victor Roberts wrote: > >> On 17 Jun 2006 08:22:21 -0700, "redbelly" > >> <redbell...@yahoo.com> wrote:
> >> >You need current regulation for an LED, which brings down the overall > >> >efficacy. Not required for incandescents or cfl+ballast.
> >> >If, for example, a 20 lm/W LED becomes 15 lm/W after adding a current > >> >regulator, then it doesn't beat a household incandescent bulb (15 > >> >lm/W).
> >> You're assuming a current regulator with an efficiency of > >> only 75%. There is no reason why a good LED current > >> regulator cannot have an efficiency of 90% or more, which is > >> similar to that of high quality electronic ballasts.
> >The ones I've seen have roughly a 1V drop. If you run a single white > >LED (3.5 V) with that, that gives a 78% efficiency. Do you know of > >regulators with smaller voltage drops, or are you thinking in terms of > >running strings of LED's on a single regulator?
> I know of such circuits having voltage drop more like half a volt. > Consider inductive switching ones using power MOSFETs and "catch diodes" > being Schottky diodes. Along with the obvious design parameter of higher > battery voltages.
> I would say that 85% efficiency of "an LED ballast" is low end of > someone trying to work at that and shaving costs by so severely as 1/10ths > of a cent, and 90% is fairly easily achievable without great cost.
On 18 Jun 2006 00:16:51 -0700, "Boxman" <box...@voyager.net> wrote:
>Thus it is conceivable that an LED downlight will use less lumens to >get the same resulting output in the beam when compared to an >incandescent.
This last part is correct (if you mean "source" lumens) for any small source and will certainly be true for LEDs, but it does not explain how the total lumen output of an LED-based fixture can be higher than the output of the LEDs themselves, measured in lumens.
-- Vic Roberts http://www.RobertsResearchInc.com To reply via e-mail: replace xxx with vdr in the Reply to: address or use e-mail address listed at the Web site.
This information is provided for educational purposes only. It may not be used in any publication or posted on any Web site without written permission.
|
