LED vs Compact Flourescents efficiency

[Steelwolf]

Just bought a Philips Essential Genie. These are the compact flourescents designed to be a direct replacement for incandescent globes. They are really compact and fit in to the same space as incandescents, unlike the earlier versions that tended to be bulkier or longer.

The highest rating version currently in the shops is 11W warm white which is supposedly equivalent to a 60W incandescent, including the warm yellow glow. In actual comparison, I found it to deliver the output of a 75W (both brand new).

A blurb on the back states that it is delivering 600 lumens or just under 55 lumens/W. The most efficient LED I know of right now is the Luxeon Star which delivers approximately 15 lumens/W (if you're lucky to get a good one). Someone correct me if there is an LED that is even more efficient.

What would it take to have a LED that would be as efficient as a compact flourescent? IMHO, it would take a new semi-conductor material, and it would be as dramatic a change as the discovery of InGaN, which enabled the first really bright blue and white LEDs. We may have already beaten the InGaN LED to death. I don't think it can get much more efficient, since semi-conductors have a proportional power consumption according to the size of the active area (given constant temperature). If I understand it correctly, the increase in lumens/W of the LS over the Nichia 5mm is due mainly to better phosphorence material.

Forgive the rant. I'm just wondering when that "future in lighting" will actually arrive. It has done wonders for the pocket flashlight, but it still can't compete with the big guns except in the area of robustness and longevity, and it still costs more.

 

[Jonathan]

My understanding is that there are _many_ factors which play into the efficiency of LEDs.

1) Efficient use of materials. 'PeLu' reported that when _underdriven_, an early sample Luxeon 5W produced more than 40 lumen per watt. But this was using lots of semiconductor material to produce not much light. To make LEDs cost effective in terms of up front cost, they are operated below maximum efficiency.

2) Extraction efficiency. My understanding is that the semiconductor materials themselves are quite efficient at producing photons when current flows through them, but that many of these photons never escape the semiconductor material. There is quite a bit of work to be done in solving this issue.

3) Contact and electrode losses. The electrical connection to the semiconductor material is a source of loss, one that again requires quite a bit of research.

4) Luxeons that you buy today can do better than 15 lumen/watt. 'Q' flux category (30-39 lumen at rated current, which means > 20 lumen per watt) devices are out there.

I'm certain that there are many others.

-Jon

 

[Nerd]

5 watt LS are usually around 120 lumens which means average of 24 lumens / watt

 

[snake]

question??
does your bulb power under 110V ? then will it be unfair to compare it with LED run under 3.5V-4.5V (even the operational Voltage of flourescent may not be 110V)? as VA=W so I guest there may have some relation...may be not. Any one have an idea?
Thanks

 

[Steelwolf]

Originally posted by snake:
question??
does your bulb power under 110V ? then will it be unfair to compare it with LED run under 3.5V-4.5V (even the operational Voltage of flourescent may not be 110V)? as VA=W so I guest there may have some relation...may be not. Any one have an idea?
Thanks

<font size="2" face="Verdana, Arial">A little print on the side of the light says 240V 80mA. How this works out to 11W I'm not sure. More like 20W. Maybe the tube is rated 11W and the other 9W goes in to the electronics?

So we have 20-24 lumens/W at the moment. Still a ways to go. BTW, how do we get 'Q' flux LS? Or is it still luck of the draw for small time hobbyists?

 

[Brock]

As a note, most CF's use the wattage rated for the lamp only and don't add the power used by the ballast. I have yet to find a CF that doesn't draw over 130% of what they are calling it, it a 10w CF will pull at least 13w. While a 25w incandescent actually pulls very close to 25w. If you get up to the 28w CF they pull almost 40w. Don’t get me wrong they are still far better then incandescent, but I don’t think it’s right that they can do this.

 

[Jonathan]

It would depend upon the actual power being consumed by the lamp, not the VA being consumed.

If something is called a 10W lamp, but the _required_ ballast is using an additional 9W, then I don't think that it is fair to say that you are only using 10W of power. In this case, you are clearly using 19W.

But with the light in question (80mA at 240V), it is not necessarily the case that 19W is being used. In AC circuits you can have current flowing out of phase with the voltage. The portion of the current flowing at 90 degrees to the voltage delivers no _net_ power to the load; for part of the cycle it delivers power, and for part of the cycle power is returned to the source. Especially with power electronic loads, such as the ballast in a CF lamp, you will see current not in phase with the voltage, leading to higher VA being consumed than actual power being consumed.

The full current being drawn must be reported, as this still results in heating of the supply wire and the blowing of fuses, but the product of voltage and current will in general be different from power.

If the actual system is drawing 10W of power but 19VA total, then the information on the box is correct. Based upon what Brock described, I'd bet the system is drawing 13W of real power, but 19VA total.

-Jon

 

[hotfoot]

Steelwolf,

I'll try answering the bit you asked about
"what it would take". I'm probably taking a waaay too simplistic view and get rapped on my knuckles by Jonathan or Doug S, but its the gist of things.

The next great hope, I thinking, is the perfection of the UV LED. This doesn't necessarily mean the perfection of the high-power UV LED a la the luxeons, but I suspect that once the initial groundwork is laid, the more powerful packages will follow soon after.

Perfection of the UV LED will mean that solid-state devices will be able to generate light basically the same way fluorescents do - by excitation of phosphors via UV. This would appear to suggest that similar levels of efficiency can be had, assuming the electrical portions of the entire equation check out well too.

 

[php_44]

initial thread:


I don't think it can get much more efficient, since semi-conductors have a proportional power consumption according to the size of the active area (given constant temperature). If I understand it correctly, the increase in lumens/W of the LS over the Nichia 5mm is due mainly to better phosphorence material.

<font size="2" face="Verdana, Arial">The efficiency of the LS over Nichia stuff is mostly due to extraction efficiency from the die and the package they put the die in. There is still a lot of room to grow efficiency there. As well, the physics of device operation predict there is still a lot of efficiency to be gained through tweaking the construction of the device and using more perfect crystalline material as a base. They SAY that they may reach 200 lumens per watt for LEDs in the next ten years.

last thread:

Perfection of the UV LED will mean that solid-state devices will be able to generate light basically the same way fluorescents do - by excitation of phosphors via UV. This would appear to suggest that similar levels of efficiency can be had, assuming the electrical portions of the entire equation check out well too.

<font size="2" face="Verdana, Arial">Actually, the only benefit of using UV LEDs for white light would be the ability to use tri-color phosphors and thus produce a light with excellent color rendering characteristics - just like compact fluorescents.

As far as efficiency, it is less efficient to downconvert UV into visible light, than to produce visible light to start with. A blue LED with it's phosphor produces light more efficiently than a UV LED of equal efficiency with a downconvert phosphor mix. But the latter produces "better quality" light.

 

[radellaf]

I've noticed that fluorescent lanterns are always over-rated in their wattage specs. For instance, the Brookstone U-tube 4 D cell light draws about 500mA at 4.8V, a far cry from the supposed "7 watt" rating they give. Similarly, the Energizer folding light with one "4 watt" tube lit, draws about 360mA at 4.8V.

 

[hotfoot]

Thanks, php_44! One question, though:

Why can't tricolor phosphors be used with blue LEDs now, then - is this because these wavelengths are incapable of sufficient excitation?

One of the biggest bugbears of the white LED and general illumination is not just about color-rendering, but color output itself. Many many folks prefer to have warm, incandescent-like white - something that requires LED mixing now. I'm waiting for a high-power, high-efficiency tri-color phosphored LED to become available for this reason.

 

[Albany Tom]

I believe flourescents dim considerably with age, and the efficiency specs for them are at new brightness.

 

[Doug S]

Originally posted by Albany Tom:
I believe flourescents dim considerably with age, and the efficiency specs for them are at new brightness.

<font size="2" face="Verdana, Arial">Yes they do, especially early in life. For this reason, it is the industry standard to rate the output at 100 hours rather than at 0 hours.

 

[shipinretech]

Hopefully most of the serious criminals in the power industry who would try to supress this technology will be dissuaded from doing so by the enforcement attention they are currently receiving. My friend who was responsible for bagging the Enron trader here in Portland told me that he was prosecuted because his actions were the most obvious, not the most damaging. It is possible that the power company people who are criminal conspiritors will keep their hands off this technology development.

It seems to me that building a substrate that is capable of handling the thermal stresses of a high output of light is the stumbling block to light output. Unlike processors or memory, everything on an LED has to be on the surface, not under the hood. Developing a substrate that is not a plane is probably a method that could work, but would take a lot of development. Not anybody's definition of low hanging fruit.

 

[Don Klipstein Jr.]

Steelwolf said:

A little print on the side of the light says 240V 80mA. How this works out to 11W
I'm not sure. More like 20W. Maybe the tube is rated 11W and the other 9W goes
in to the electronics?

<font size="2" face="Verdana, Arial">Compact fluorescents often have such a thing as non-unity power factor. Current can have a portion 90 degrees out of phase with the voltage or at harmonic frequencies, and that current when multiplied by line voltage is not power consumption but energy circulating back and forth and largely unbilled.

Meanwhile, I do want to add that the wattage rating for a compact fluorescent includes ballast losses if the ballast is built-in. If the ballast is separatable from the bulb or something else in a fixture, then for 18W or smaller compact it usually consumes 3-4 watts extra, and for 20-28W maybe 4-5 watts extra.

- Don Klipstein ([email protected]), http://www.misty.com/~don/index.html

 

[rlhess]

Many power companies want the total power factor to be 100% +/- 5-10%. These devices may have a power factor of 70%. Use enough of them and the power supply will become unhappy. Computer supply power factors used to be very low, they are now improving them (to some degree to meet European codes). If you see a power factor of 70% and you're using a lot of these (don't worry at home), you'll need oversized (typically double) neutrals in a three-phase distribution system.

Cheers,

Richard

 

[php_44]

Originally posted by hotfoot:
Thanks, php_44! One question, though:

Why can't tricolor phosphors be used with blue LEDs now, then - is this because these wavelengths are incapable of sufficient excitation?

I'm waiting for a high-power, high-efficiency tri-color phosphored LED to become available for this reason.

<font size="2" face="Verdana, Arial">Hotfoot -
Lumileds has some great IEEE papers on this. Actually, their favorite white LED now is the Blue LED + yellow phosphors. This is the most efficient phosphor + LED solution they have, and also the most reliable because of the excellent yellow phosphor they have.

They have experimented with a Blue LED along with green and red phosphors to make a white light. The color rendering is as good as a tricolor phosphor, but it has problems. The temperature stability and life are not there yet. (The light changes color with temperature, and as it ages). Also, there is a big efficiency penalty in trading the high energy blue light for the lower energy red light via the phosphor - it's kind of like using a blowtorch to warm some coffee - it works OK but is not very efficient. They are working on it, though.

Likewise, they are working on UV LEDs with tricolor(blue, green & red phosphors). They still have to fix the temperature and lifetime issues with the phosphors. The UV wavelengths produced by the LEDs are longer than those produced by a fluorescent light, and so they can't really use those established phosphors. Also, due to the laws of physics - they will have to make a UV LED that is twice as efficient as the best green LEDs for the UV+phosphor light source to be as efficient an emitter.

They are pursuing all three because even they don't know which will win in the end - but regardless of the outcome - we will benefit from their work with some excellent LED light sources. They project $10 per 1000 lumens in 7-10 years (a 100w tungsten light is ~1600 lumens), and knowing technology it might come sooner.

 

[Doug S]

Here is a link to that paper that php_44 referred to for those that missed it the first time around:
http://www.lumileds.com/pdfs/techpaperspres/IEEEJSTQE-HighPowerPCLEDs.pdf
For those new to the white LED phosphor discussion, an earlier thread may be of interest:
http://www.candlepowerforums.com/cgi-bin/ultimatebb.cgi?ubb=get_topic;f=3;t=002802#000000

 

[Doug S]

Originally posted by Steelwolf:

What would it take to have a LED that would be as efficient as a compact flourescent? IMHO, it would take a new semi-conductor material, and it would be as dramatic a change as the discovery of InGaN, which enabled the first really bright blue and white LEDs. We may have already beaten the InGaN LED to death. I don't think it can get much more efficient, since semi-conductors have a proportional power consumption according to the size of the active area (given constant temperature). If I understand it correctly, the increase in lumens/W of the LS over the Nichia 5mm is due mainly to better phosphorence material.

<font size="2" face="Verdana, Arial">For the answer to "What would it take?" I strongly suggest reading the paper at the following link. Caution: long, somewhat technical, in PDF format, and I noted an error or two. Very much worth a read if you have the stamina.
http://www.lumileds.com/pdfs/techpaperspres/OIDA_28May2002.pdf

 

[Doug S]

Originally posted by hotfoot:
Doug S, interesting. I now begin to wonder if a phosphor blend consisting of elements with broad and/or overlapping peaks is possible and practical. But I truly am now stepping into an area I know absolutely nothing about...

<font size="2" face="Verdana, Arial">You now know everything that I do about this subject.