Integrated Reflector Could Lead to Ubiquitous LEDs

Andreas writes "Professor Schubert says he has found a way to raise the efficiency of LEDs to 99%. From an article on Advanced Technology: "Until now, all lighting systems, especially incandescent bulbs, generated more heat than light. But our 99-percent efficient reflectors for LEDs makes them the first candidate for light-bulb replacement that generates more light than heat," said Schubert."

Comments

Not so fast, Sparky!

[rco3]

"Professor Schubert says he has found a way to raise the efficiency of LEDs to 99%.
No, that's not what he says. The reflector is 99% efficient, not the LED. This reflector just means that some of the light emitted by the LED, which otherwise might be absorbed by the LED substrate or other structure and converted to waste heat, is now being reflected back out as usable light.
This new technology does nothing to improve the quantum efficiency of the LED itself. It's an important and useful technology, sure, but it is NOT a 99% efficient LED.

Re:Not so fast, Sparky!

[Pinkoir]

As the man says, a 99% reflector is not a 99% LED but this technology would certainly seem to be a useful advance.

Some current LEDs already use thin flim techniques and reflectors to collect the rear emitted light and throw it out the front but this is the first I have heard of combining the reflector with the wiring. This might also have the potential of moving heat out of the junction more efficiently would would be a real help in a lot of applications.

Still I'm curious about how much this will add to the cost of a white LED. There is no point making a bright LED if the total system cost is still ten to twenty times that of an incandescent source. I'm an illumination engineer in the automotive field and when we look at replacing a bulb with LEDs we have to add in not only the cost of the LEDs themselves but also the board, the heatsink, the drive module and associated cicuitry. So while it is true that more light and less heat would be better the real roadblock in my application is cost and we won't see the widespread replacement of filament sources with LEDs until that issue is resolved.

-Pinkoir

Costs have to be considered in toto

[Engineer-Poet]

Quoth the poster:

There is no point making a bright LED if the total system cost is still ten to twenty times that of an incandescent source.

As a counterexample, consider the compact-fluorescent bulb. Most cost around ten times as much as an incandescent of equivalent brightness, but the savings in replacement costs and power will pay back the difference in as little as a few months.

LED technology has the advantage of longer life than fluorescent. With the increase in efficiency from reflectors, they could cut power costs below fluorescent and become the TCO winners.

Re:Costs have to be considered in toto

[Elledan]

"LED technology has the advantage of longer life than fluorescent. With the increase in efficiency from reflectors, they could cut power costs below fluorescent and become the TCO winners."

Also something to keep in mind is that LEDs are far more robust than fluorescent and incandescent lights. Incandescent lights can't take shocks very well, and the huge temperature delta between an on/off state reduces its lifespan significantly. Fluorescent lights are also relative fragile (ever tried replacing fifty or so of those tubes without shattering at least one of them?), and the ballast used to generate the required high voltage (most types of ballast) create quite a lot of EMI, which is bad for sensitive equipment and cables like Cat-5 etc.

LEDs generate very little heat, require only a very small current (tens of mA!) at equally low voltages, produce no EMI, are unaffected by all but the most severe shocks and last virtually forever (100,000 hours for red, green, etc. with ease).

If LEDs are made brighter, even if this makes them more expensive than other technologies, there are always some (less friendly) places where they would work very well and would be cheaper in the long term.

Re:Costs have to be considered in toto

[Vengeful weenie]

I'd probably add the fact that fluorescent bulbs are environmentally problematic, and can be dangerous to health. While manufacturing LEDs undoubtedly create waste by-products, I would doubt that they would be much worse than when manufacturing fluorescent bulbs.

110/230V AC

[EnglishTim]

Okay, so let's assume all our lightbulbs start being made from LEDs... At some point soon we're going to have to start changing our lighting circuits to 5V, or something like that. It's madness that each lightbulb will have to contain it's own little transformer - it'll make the bulbs vastly more expensive and wasteful.

There are a selection of appliances that work well with 110/230V AC - things that require a lot of power like kettles, hoovers, heaters, washing machines, hobs, tumble driers and the like. However, there's an increasing number of appliances in a modern household that would be much better served by a 12V DC supply.

How long do you think it'll be before we start changing over?

Re:110/230V AC

[Smidge204]

Unfortunately DC power doesn't transmit over any kind of distance very well. AC is much more efficient for that. (Esp. at high voltages... 20,000V+)

Also, all flourecent lights have transformers in them, so suddenly it's not too unreasonable for each light fixture to have its own little transformer in it!
=Smidge=

Re:110/230V AC

[alienw]

Running 5V around the whole house would be much more inefficient than putting a transformer inside each bulb. It would also be expensive, and a fire hazard. Think about it: running 1kW of power through a 5V system would require 200 amps of current! That would require welding cable-sized wires.

Re:110/230V AC

[bluGill]

Electronics 101: when you connect 2 circuits in series each sees half the total voltage. Connect 24 LEDs in series to a 120 volt line and each sees 5 volts! By definition there is no need to transform the AC into DC - the D in LED stands for diode, which is what you use to turn AC into DC! In the real world you are likely to use 48 LEDs, in two different strings, so that you get light from both sides of the wave.

Re:110/230V AC

[SuperBanana]

It's madness that each lightbulb will have to contain it's own little transformer - it'll make the bulbs vastly more expensive and wasteful.

If you have 2v LEDs, you only need wire about 60 of them in series and you've taken care of the voltage problem. Well, except they'll blink at 60hz, quite strongly...and if one failed, they'd all go out. But in any case, it's hardly rocket science to make use of the higher voltage level, especially since LEDs will tolerate A/C. Incidentally, look at a screw-in fluorescent bulb some time- they've gotten the whole thing down to $10 or so, and that includes a transformer and electronics to raise the voltage. Transformers etc are very cheap.

However, there's an increasing number of appliances in a modern household that would be much better served by a 12V DC supply. How long do you think it'll be before we start changing over?

Never. The whole point behind A/C is that it is very easy to step up/down, and as a result, you can use a higher voltage for transmission and distribution. Higher voltages mean less current flow for the same amount of energy, which means reasonably sized wiring and such.

Even in the short distances involved in a house, losses from wiring can be substantial at such a low voltages as 12v. 48v might be a better choice, but I can't see it ever taking off.

Re:110/230V AC

[falzer]

> Unfortunately DC power doesn't transmit over any kind of distance very well.

Looking at just the wire itself, transmission losses aren't worse for DC. There are a few HVDC transmission lines in operation now. Some are used for 50/60Hz conversion.

The reason AC is used because it's easier/cheaper to efficiently step up (and down) the voltage to useful levels, as per your power transmission example.

Re:110/230V AC

[Peepsalot]

It's madness that each lightbulb will have to contain it's own little transformer - it'll make the bulbs vastly more expensive and wasteful.

Sorry, I don't think you'll see a change(at least not to a lower voltage) in your wall outlet's voltage any time soon. Maybe a transformer in every light bulb seems wasteful, but take into consideration that led's can last roughly 100,000 hrs as opposed to 1,000 for incandescant's.

Still seems wasteful?
Transformers don't have to be huge, especially if they are powering an efficient device that doesn't draw much power. Disposable cameras contain transformers in them that step the voltage from 1.5 to 300 volts, and they are pretty damn small. In fact you could fit a transformer into just the metal base of a normal incandescant bulb. Not to mention the transformer's solid state cousins: voltage regulators, swithcing power supplies, and probably some others I don't know about (IANAElectrician.)

In fact, you could also scrap the idea of stepping down the voltage entirely. How about putting more led's in series. Think about it, led's emit a narrow spectrum of light, so to smoothen out the spectrum, you might want to add a bunch of led's from different parts of the spectrum. Just put 40 little 3v led's on a board and voila, you have a device that accepts 120 volts. Another reason to use a large number of leds: you might want to spread out the surface area where the light is coming from, a single point of light could seriously mess with your vision. Not to mention the fact that I doubt to see any single led in the near future that can output the equivalent of a 60 watt incandescant light bulb.

But the thing the I wonder about is if and when the standard plug model will be changed to a model where the socket and the plug contain coils to transfer the power, like an isolation tranformer that is split into two parts.(Have you ever seen how a sonicare toothbrush charges itself?) The main benefit I see in this is safety; never have any exposed wires to fry yourself with. Actually, if all your wall sockets had coils in them, the plug of your device could determine the voltage that device receives, simply by the nuymber of coils in it. Again, IANAE, so I don't know, maybe it would take the advent of room-temperature superconductors to make this idea feasible and efficient.

Re:110/230V AC

[rco3]

A few points to note, realizing that not many posters around here are EE's:

1) LED's are not voltage mode devices. There IS a typical voltage drop associated with an LED, but it can vary appreciably between devices. One sets the operational point of an LED by controlling the current through it, and allowing the voltage to settle to whatever value it wants. Typically, one would want to see around 30 mA through a normal T1-3/4 LED. Depending on the chemistry of the LED, this could result in a voltage anywhere between 1.5V and 3 or 4V. This, as I said, will vary somewhat between different LED's of the identical type. If you try to set the voltage, you'll get wildly varying currents and a lot of dead LEDs.

2) Stringing together LEDs in series to get something approaching 120V drop is a good idea, but you still have to limit the current. Leaving a few volts between the nominal operational voltage of your LED string and the nominal supply voltage is a good idea, because you can then use an active (or passive) current limitation scheme which operates within that voltage gap. The simplest way is with a single resistor, sized such that R= (Vsupply-VLEDS)/ILEDS. This is subject to variation due to device mismatch, temp variation, etc, and dissipates some power in the resistor. Another way to do this, which allows for the LED to be operated from a much higher voltage than it's rated for, is to use a series capacitor. The determination of proper capacitor size is a bit more tricky, but you can successfully run a single LED from a 120V supply. The indicator in my waterbed heater has run this way for a couple of years now. Nice part is that the capacitor does NOT dissipate any power as heat. Enough of them might screw up your power factor enough to piss off the power company, though :-)

Important messages to take home from this: you can't set LED operating point from the voltage across it, at least not safely and reliably; you can operate LEDs from 120VAC using a capacitor as the gain setting element, which is appreciably cheaper than using a transformer.

Re:110/230V AC

[HaveNoMouth]

The reason AC is used because it's easier/cheaper to efficiently step up (and down) the voltage to useful levels, as per your power transmission example.

Exactly right. The real issue is that transmission lines are not perfect conductors; there is always some small resistance which causes power to be lost in heating up the wire. The heat loss is given by the square of the current times that resistance. Therefore, you want to transmit power with as low a current as possible to minimize the power lost along the wire itself. Lowering the current necessarily requires raising the voltage if you want the same power to come out the other end of the wire, thus power is most efficiently transmitted at high voltages and low currents. This is just as true for DC as it is for AC, but AC allows you to easily step the voltage back down again for use at the load.

If we had superconducting transmission lines--which could become commonplace this century--the whole issue would become moot because we'd be able to efficiently transmit at high current and low voltage and DC would be just as practical as AC (if not more so because of the lack of AC radiation effects mentioned by another poster).

Re:How are narrow-angle traffic-signals done?

[zrail]

These lights use a Fresnel Lens to direct the light from the bulb or LED cluster to only the lane for which it is intended.

wikipedia link

I want a LED lighted DLP projector

[doc modulo]

At the moment, projectors are lighted by expensive, proprietary light bulbs.

Because of the hot bulbs, the projectors are too noisy to enjoy a nice movie night at home and they burn out after a while.

An array of LEDs would be superiour because they'd be more durable (no need for expensive replacements after X hours) and might be cool enough for fanless beamers.

Unfortunately the manufacturers use the projectors like razorblade holders or like inkjet printers. You can only fit the replacement bulb that the manufacturer made themselves and the replacement bulbs are very expensive because of that monopoly.

However, all it takes is ONE monufacturer to produce a good LED beamer to disrupt the current situation. All the others will have to follow if they want customers after LED lightsources take over, the sooner the better.

I made up my mind to ONLY buy a LED beamer because I know it's possible and I know I'll be screwed over by the current beamers if I don't. The less bulb-beamers we as consumers buy, the faster the changeover will happen.

Re:I want a LED lighted DLP projector

[Aidtopia]

Despite the naysayers here, you could be on to something.

A DLP projector uses a "white" bulb and reflects the light through red, green, and blue filters. The filters (ideally) are band pass filters, allowing only a fraction of the light to pass, and absorbing the rest, which must be reradiated as heat.

If our light source was "tuned" to put most of its power out in the bands used by the filters, then a lot less light would wasted. So if you could make your light source from properly tuned and bright red, green, and blue LEDs, it would be much more efficient, requiring less power and creating less heat. I'm sure the details of this make it hard to solve, but in theory it would be a great solution.

[Are LEDs different colors because of the filtering of the packaging or because they're tuned to produce different wavelengths of light?]

DLP projectors have two basic designs. Most reflect the light off the mirror matrix and through a rotating wheel of red, green, and blue filters. Some high end ones start with three light sources (or a single source that's split with prisms). The three light sources are filtered and then bounced off independent mirror matrixes and recombined with prisms.

Both of these could benefit from an LED approach. With a single matrix, you could get rid of the mechanical color wheel by strobing the LEDs instead (switch them on and off in R-G-B sequence). I believe, (though I'm not certain) that LEDs turn on and off very quickly. With three matrixes, you simply start with three different light sources, rather than splitting a single source and losing heat.

I'm sure there are many electrical and optical challenges with this approach. If they can be solved, you could make very robust, quiet, long-life, low-power DLP displays.

colour temperature

[RMH101]

but bear in mind the colour temperature of them seems "weird" to the human eye; lighting a room in them isn't very cozy.

Actually I do...

[Pinkoir]

If you read the article you will see that there is no claim on 99% efficiency of the LED. The claim is 99% efficiency on the reflector. No LED anywhere in the world comes even close to 220 lumens/watt. The best I've seen in the real world is about 80 lumens from the 3 watt Luxeon devices put out by Lumileds. These devices are very hot and need a lot of heat-sinking to avoid destroying themselves in any confined application. You have to remember that LEDs aren't magic. They are just full of inefficiencies and the back reflector issue is only one of them. I work with high intensity white LEDs every day and if you know of some that can give the output you talk about that don't need active cooling for God's sake tell me where I can buy them.

-Pinkoir