Indeed, I was going to say, from my experience high power LEDs are ~20% efficient. You might have 15 watts of power going into your LEDS but only 3 watts of this is actual light, the other 12 you have to provide heat sinking for.
The heat issue with LEDs really only comes into it when you're trying to design a replacement bulb for a standard fitting. If however the lamp is designed from the start as an LED light, then the heat sinking can be easily incorporated into the design (This is of course if we're talking about a standard decorative consumer lamp).
The LED array that digits posted about I've had my eye on. This thing driven at 1 amp would need about a 2 degree/watt heat sink to work within the specified temperature range. At 1 amp the thing draws about 45 watts and used at a CCT that looks nice, will produce around 2500 lumen. For comparison a normal 60 watt incandescent is around 800 lumen, so the LED is roughly 4 times as bright while using 5.3 times less power then the 4 incandescents combined.
LED lighting really isn't at the point where people think it's at. Compared to the best fluorescent and CFLs LEDs are roughly about the same in terms of efficiency. Trouble plagues cheap CFLs, but these same troubles plague cheap LEDs too. Typically, poor quality driver circuits, with crap power factor correction leading to significant power being wasted before it's even reached the bulb. Then the bulbs don't last anywhere near their specified lifespan.
Where a good LED bulb wins vs a good CFL is in the lifespan, but also in the clean up, they are far less of a hazard to the environment. The problem is the price though, LEDs ain't cheap! Ironically, of the good quality LEDs available, Cree happen to make some of the best and some of the most affordable too.
Digikey has a ton of info on this subject in their TechZone area
Lighting Solutions | DigiKey Technology Zone
Lighting Solutions | DigiKey Technology Zone
All the energy ends up as heat. In both cases.
LEDs are currently ~as efficient as sodium street lighting, which is as good as you can get in efficiency terms, the quality of light is better, and they are increasingly being adopted for street lighting. They will get better in future. They offer the possibility of near complete conversion of power to light (in the first instance), which means a cold luminaire.
The other advantage is that their lifetime is considerably longer than any other lighting device.
And anyway, I didn't offer the information solely for your benefit.
w
I've been playing around with high power (1+ watt LEDs since 2002
They make great bike lights and don't blow out like hot wires. From what I gather, the max output--the 100% efficient blue/phospher LED will put out 330 lumens at one watt. My Cree XP-G bin code R5 LED puts out around 140 lumens at one watt but I run it at 2.2 watts for 285 lumens. At the higher drive rate the efficiency goes down (just like a speaker) I'm sitting at around 130 lumens per watt. Not bad, around 40% efficient and 60% heat.
Cree has a LED in their lab that puts out 210 lumens at one watt or 64% efficient--four of those at 4 watts will put out the same amount of light as a 60 watt incandescent bulb but only 1.5 watts of heat. Granted, the "bulb" has to run on DC so you'll lose power through the AC to DC conversion but even at 5 watts--much better than 60 watts!
Technology is getting there, back in 2002 my Luxeon I was pumping 17 lumens per watt and I was amazed! Eventually, your walls will be OLED wallpaper for video displays and room lighting. Adjust the output and tint to your liking then use the other wall for Horn Response.
They make great bike lights and don't blow out like hot wires. From what I gather, the max output--the 100% efficient blue/phospher LED will put out 330 lumens at one watt. My Cree XP-G bin code R5 LED puts out around 140 lumens at one watt but I run it at 2.2 watts for 285 lumens. At the higher drive rate the efficiency goes down (just like a speaker) I'm sitting at around 130 lumens per watt. Not bad, around 40% efficient and 60% heat.
Cree has a LED in their lab that puts out 210 lumens at one watt or 64% efficient--four of those at 4 watts will put out the same amount of light as a 60 watt incandescent bulb but only 1.5 watts of heat. Granted, the "bulb" has to run on DC so you'll lose power through the AC to DC conversion but even at 5 watts--much better than 60 watts!
Technology is getting there, back in 2002 my Luxeon I was pumping 17 lumens per watt and I was amazed! Eventually, your walls will be OLED wallpaper for video displays and room lighting. Adjust the output and tint to your liking then use the other wall for Horn Response.
To anyone interested: as far as the LEDs meant for home use, do they exhibit any visible flickering like those LED-based Xmas light strings or the ones used in many automotive tailamps? I hope not.
That's one thing about incandescents I definitely like, and that's their pleasant and "relaxed" light they emit. An example on a much larger scale was on display 2 years ago at a local - and old, about 60 years IIRC - high school gym where I attended a sporting event. I wasn't specifically looking for this effect and we were seated a couple minutes when I realized something was different about this room. Everything was perfectly visible but looked, well, relaxing. Looking up, I realized that the light was being provided by enormous incandescent bulbs, the kind I haven't seen since I was in elementary school in the early 70s (Google tells me these were probably in the "PS35" bulb category). Just a very nice experience, especially compared to the nasty & "sharp" effect that typical modern gym lighting produces.
That's one thing about incandescents I definitely like, and that's their pleasant and "relaxed" light they emit. An example on a much larger scale was on display 2 years ago at a local - and old, about 60 years IIRC - high school gym where I attended a sporting event. I wasn't specifically looking for this effect and we were seated a couple minutes when I realized something was different about this room. Everything was perfectly visible but looked, well, relaxing. Looking up, I realized that the light was being provided by enormous incandescent bulbs, the kind I haven't seen since I was in elementary school in the early 70s (Google tells me these were probably in the "PS35" bulb category). Just a very nice experience, especially compared to the nasty & "sharp" effect that typical modern gym lighting produces.
Depends. Incandescent bulbs have the advantage of the thermal mass of the filament smoothing out the AC and produce a light with minimal flicker. LEDs and CFL's are much faster devices. Some el-cheapo Christmas light strings run half-wave rectified, hence, are off for half the AC cycle. Many people are bothered by the flickering of those. The better light strings run full-wave rectifiers. They still flicker, but they do so at 120 Hz and are only off a tiny fraction of the AC cycle, hence, the flickering is very hard to notice. Of course, all you'd need to eliminate the flicker would be a small cap on the output of the rectifier...
As for LED bulbs, the ones I've tried haven't flickered. In the "old" days with CRT monitors, I would be bothered by 60 Hz refresh flicker so I think I'd notice if an LED was flickering at that frequency. As far as I understand it, the LED driver ICs are essentially switchmode supplies that pulse width modulate the LEDs at some frequency well above the audio range. You shouldn't notice any flicker with those.
My only complaint about LED lighting is the color temperature. The warm glow of an incandescent bulb is nice. But a quality LED bulb should get pretty close to that...
Times change. I'm sure a lot of people were missing the pleasant warm glow of carbon filament bulbs when they went out of fashion and were replaced by the more efficient tungsten filament bulbs.
~Tom
Color temperature
The "relaxed" feel of incandescents is due to their low color temperature, they emit more red and less blue than the sun, the "cool white" fluorescents typical of commercial establishments, and the metal halide typical of other gymnasiums. Blue light suppresses melatonin production, the chemical that induces sleep and relaxation. Reduce the blue light (e.g. at sunset) and melatonin increases and we relax. It's a natural wake up/go-to-bed mechanism. This is the one of the reasons that researchers are saying so many more people are having sleeping problems today, because they have bigger TVs on at night and computer/cellphones practically in their faces late at night that emit lots of blue light. I was at a trade show once where LED PAR cans were demonstrated that had the same exact color temp as incandescent. The guy said that two colors of LED were mixed in the PAR can to make the incandescent-style light: Hot Pink and "BSG" (baby sh!t green). From the outside you would never know. The LED cans had large heat sinks on the back, and the largest ones were fan-cooled.
The "relaxed" feel of incandescents is due to their low color temperature, they emit more red and less blue than the sun, the "cool white" fluorescents typical of commercial establishments, and the metal halide typical of other gymnasiums. Blue light suppresses melatonin production, the chemical that induces sleep and relaxation. Reduce the blue light (e.g. at sunset) and melatonin increases and we relax. It's a natural wake up/go-to-bed mechanism. This is the one of the reasons that researchers are saying so many more people are having sleeping problems today, because they have bigger TVs on at night and computer/cellphones practically in their faces late at night that emit lots of blue light. I was at a trade show once where LED PAR cans were demonstrated that had the same exact color temp as incandescent. The guy said that two colors of LED were mixed in the PAR can to make the incandescent-style light: Hot Pink and "BSG" (baby sh!t green). From the outside you would never know. The LED cans had large heat sinks on the back, and the largest ones were fan-cooled.
As Tom states, it all depends on the design
Flickering drives me nuts so I avoid CFLs for reading lamps. My high powered lights used on my bicycle are current regulated so don't use PWM. As long as the PWM designs use a really high pulse rate, I don't notice it. One of my flashlights uses variable PWM but has a switching speed of 1,500 Hz and I've never noticed the flicker.
One way for you to test it is by turning the light on, splay your fingers out in your hand and wave it quickly back and forth. If you get a strobe effect, it is PWM.
The "cold white" of LEDs are the most efficient, the "warm white" versions trade the warm yellowish incandescent color for less efficiency. At least the technology is getting advanced enough to have options. I have a warm white version, I notice it works better outside when viewing greens and browns.
Flickering drives me nuts so I avoid CFLs for reading lamps. My high powered lights used on my bicycle are current regulated so don't use PWM. As long as the PWM designs use a really high pulse rate, I don't notice it. One of my flashlights uses variable PWM but has a switching speed of 1,500 Hz and I've never noticed the flicker.
One way for you to test it is by turning the light on, splay your fingers out in your hand and wave it quickly back and forth. If you get a strobe effect, it is PWM.
The "cold white" of LEDs are the most efficient, the "warm white" versions trade the warm yellowish incandescent color for less efficiency. At least the technology is getting advanced enough to have options. I have a warm white version, I notice it works better outside when viewing greens and browns.
This isn't quite correct I don't believe. The theoretical max for an LED is 300 lm/W, but the theoretical max for a light source is more like 680 lm/W. An LED is still going to chuck out 55% of the power as heat even at it's most efficient.
The Cree XP-G is around 20% efficient but this is in a CCT that would look downright depressing in a domestic environment. The 140 lm/W falls to 107 lm/W for the attractive light source. I used 4 in a lamp @ 1 amp that gives out about 1100lm. A 100 watt incandescent gives out around 1400lm which is coincidentally exactly the same amount of light as 5 of those XP-Gs. Mine run at 3.34volts for 1amp of current yielding a total of 16.7 watts being dissipated by the LEDs. Linear technology make an offline single chip mains driven LED driver that operates at 80%+ efficiency, so if designed as a replacement bulb you'd be looking at a total system power of ~21 watts. This is considerably less then a lot of CFLs, ones we've got draw up to ~40 watts. Of course the LED bulb should last a lot longer too.
With regards to heat, the last lamp I built was designed around 6 Cree XPE-HEW LEDs. These operate at 770mA with a total forward voltage of 18.7volts or 14.4 watts dissipated by the LEDs. The LED driver is a MAX16832 and due to space constraints a highish switching frequency has been used ~1 Mghz. this reduces the size of the inductor, but reduces the efficiency too, either way, it's measured efficiency is 93%. Naturally this needs a source of DC power and I used an old LCD monitor switch mode. If a modern efficient switch mode were used I reckon you'd be looking at an end efficiency of 85% from AC power to total dissipated LED power.
Here is a picture of the lamp with one of the frosted plastic domes removed, showing where one LED and the driver for all 6 is located. The construction is 5mm aluminium, the aluminium bar is 25cm long and 3/4" in diameter, the plastic domes are 5cm in diameter. The surface area of the aluminium is reasonably large, yet it reached 47 degrees in the 22 ambient, for an increase of 25 degrees. This is hot to the touch, but not quite hot enough to make you want to remove your hand. The eggs are for a size comparison!
Here is also an image of the lamp turned on, note that it might look 'darker' then the first image but this is in fact just a trick of the light. The surroundings obviously got brighter re the first image, only with the lamp on the exposure was shortened otherwise the image would saturate.
The above lamp puts out about 1100 lumens.
As can be seen the heat isn't too much of a concern if the end product is designed as an LED light from the start. The amount of light that LEDs give out is currently fine for the effective creation of lamps etc, any brighter and the point sources of light might end up being harmful or distracting. What needs to happen is for similar brightness LEDs to be released that operate at lower total power.
PWM control can create flicker, but it depends on the frequency used. I've got a couple of PWM controlled LED lamps and I found I needed a PWM frequency of around 120hz for it to remain unnoticeable.
Typical LED drivers don't operate on PWM unless you tell them too, they operate on constant current. So if the PWM duty cycle is 100% the LEDs are on 100% of the time, ergo no switching. You generally keep the LED current at a fixed amount as this keeps the CCT constant, LEDs exhibit a shift in the light temperature as the current changes. This is similar to normal incandescent bulbs, the lower the current the cooler the filament runs and the warmer the light. Besides the CCT, the LED drivers are typically more efficient at higher currents too, so to vary the light intensity PWM is used. Essentially for a 50% duty cycle the LEDs are off for half the time, reducing the apparent intensity. Speeding up the PWM frequency reduces visible flicker, but it comes at a price, the inductors used in the LED drivers tend to whine at the switching frequency and it is VERY annoying. The lower the frequency the less you hear it 120hz is a nice compromise. 1khz though drives you insane!
Typical LED drivers don't operate on PWM unless you tell them too, they operate on constant current. So if the PWM duty cycle is 100% the LEDs are on 100% of the time, ergo no switching. You generally keep the LED current at a fixed amount as this keeps the CCT constant, LEDs exhibit a shift in the light temperature as the current changes. This is similar to normal incandescent bulbs, the lower the current the cooler the filament runs and the warmer the light. Besides the CCT, the LED drivers are typically more efficient at higher currents too, so to vary the light intensity PWM is used. Essentially for a 50% duty cycle the LEDs are off for half the time, reducing the apparent intensity. Speeding up the PWM frequency reduces visible flicker, but it comes at a price, the inductors used in the LED drivers tend to whine at the switching frequency and it is VERY annoying. The lower the frequency the less you hear it 120hz is a nice compromise. 1khz though drives you insane!
What's your basis for believing that? That's 44%. I understand Cree have a blue production chip (EZR) @ 47% (30mW for 64mW in). I'm not aware of a theoretical limit other than 100% efficiency (radiant). 'Course the whole discussion is complicated by considerations of efficacy versus efficiency, but apparently a guy at NIST (Yoshi Ohno) has shown a theoretical max RGB efficacy of over 400 lumens/watt, which means that 200 lumens/watt (apparently a stated goal of the US Department of Energy for SS lighting) is achievable with 50% radiant efficiency.
LEDs Magazine - Metrics for solid-state lighting
w
You should check out AL9910 driver chips, it comes in tiny So8 package, runs off universal mains with only a small dil rectifier, a mosfet, a diode and inductor,some resistors and 3 cheap capacitors... oh and it as a TTL level PWMD input pin that can take inputs put to about 1k. Unit can be totaly suspended to 1mW or something by pulling PWMD pin to ground. It also shuts down with no load with no damage and has thermal protection built in.
The figure was admittedly from wikipedia, but it usually contains fact when it comes to empirical science.
Luminous efficacy - Wikipedia, the free encyclopedia
44% max efficiency as you stated, or (as I stated) about 55% requiring a heat sink. If this isn't true then all the better, doesn't change the fact that currently most white LEDs fall between 10-20%.
That AL9910 looks like a great find! I wonder how many watts it can safely operate at. They say over 1amp of drive current, but I wonder if it can handle that at 80 volts too.
Well it switches a mosfet on and off, so its not realy connected to the load.... it takes it power directly from whatever power you have at the mains socket, through a 3k resistor.
If an LED has 15% light output and an ordinary incandecent 5%, it still makes the LED a 300% more efficient light producer.
If an LED has 15% light output and an ordinary incandecent 5%, it still makes the LED a 300% more efficient light producer.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.