"3" watt leds

[sleemanj]

Ok, so you know those 1, 3 (and occasionally 5 watt LEDs), they look like this...


So, I have played with the 1 watt variety for a while, they are freaking bright, work well as long as you heatsink them decently.

Anyway I thought I might as well get some 3 watt in my next order from China since I have some space to make up and a lack of ideas for new coolness at the moment.

But, there's something fishy, not just with one vendor, with pretty much all of them, the specs for the 1 watt leds are as you'd expect, about 3 something volts and 300 something mA, that gives you a watt right, no problem.

But the 3 watt ones, all the ones I've seen, have the same 3 something Volts (so dies in parallel obviously), but are rated at only 600 something mA, 750 at the top (so probably ou can say they are all around 650 mA in reality).  According to my maths, that makes for 2 watts. 

The lumen rated output seems to only be about double that of the 1 watt versions also, not 3 times.

In fact if you look closely at a lot of the photos you can see what appear to be two pairs of bond wires, clearly for 2 of the 1 watt die.

Am I missing something here, about luminosity or some funny led business.... or is this just a case of "China-math"?

A few examples from alibaba...
http://www.alibaba.com/product-detail/3-watt-led-high-power-led_1511033816.html?s=p  700mA
http://www.alibaba.com/product-detail/high-power-3-watt-led-chip_812470173.html?s=p 700mA
http://www.alibaba.com/product-detail/High-quality-750mA-high-power-white_1619987056.html?s=p 750mA (you can see the 2 pairs of bond wires clearly)
http://www.alibaba.com/product-detail/High-Efficiency-Cool-White-3-Watt_1715612471.html?s=p 550mA (!)
http://www.alibaba.com/product-detail/660nm-3w-led-1w-3-watt_1645633652.html 700mA

eBay the same. 
Taobao is the same.

I have yet to see an actual 3 watt led in this package.

[T3sl4co1l]

I've ran 'em at 3W before.  They get pretty toasty.  Besides materials quality, they'll probably only last thousands of hours at that rating, and I mean around 25C, let alone in any kind of lighting fixture.

On the upside, they're cheap enough to use at 1W?

48 running at near an ampere beside windows getting afternoon daylight.

Tim

[Harvs]

A couple you linked state forward voltage as 3.6V, so that's about 2.7W for the 750mA ones.

But you're right, they're over rated. 

I once got some LED downlights that were suppose to be 9W from a reputable-ish company, on measuring the power input it was more like 3.5W.  So I kind of complained and got a story about how the whole industry over rated LEDs to consumers and so they were just matching everyone else...

[grumpydoc]

I once got some LED downlights that were suppose to be 9W from a reputable-ish company

I still haven't quite given up on using 9W LED GU10's to replace the halogen spots in various rooms but I am getting increasingly frustrated with the affair.

It's not that they aren't bright enough - to my eye they are as bright as a 50W 240V halogen GU10 so I've never worried too much about whether the LEDS themselves are 3x3W or less.

The rubbish thing about them is that the drivers crap themselves with monotonous regularity - I don't think I've a single one which has lasted more than 18 months.

The LEDs themselves are usually OK so I now have an increasing supply of 3 high brightness LEDs mounted on circular PCBs - just begging for a project but I haven't quite figured out what to do with them yet.

[amyk]

In fact if you look closely at a lot of the photos you can see what appear to be two pairs of bond wires, clearly for 2 of the 1 watt die.

There are probably 3 x 1W ones out there too... but of course they would be sold as 5W or 6W units.

[Corporate666]

Don't trust the specs on ANY LED sold from a Chinese manufacturer, especially if the specs are coming from a reseller, or even worse, are coming from an eBay seller.

LED's are probably the area of electronics with the most bullshit numbers from China, IME.

[Rick Law]

I think regardless of 1W or 3W, it is rather misleading and missing the point.  I know traditional light bulbs are sold that way (by Watts - ie: consumption), and vacuum cleaners are also sold that way (by Amps - ie: consumption.)  In both cases, what one wants is not how much energy it consumes but how much energy it outputs.  Buying it by watts is like getting a bigger engine by finding engines that consumes more gasoline per mile - instead of by how many Horse Power the engine generates.

Hack, I can have a true 10Watt LED easy and cheap.  Just parallel a 3mm 20mA/2.5V LED with a good size cement resistor.  Now you have a real 10W consumption.

Getting it by-the-watts is bad, but on eBay, getting it by lumens/lux is just as bad.  I have some 160-lumens LEDs (G4 connector, 12 5050-SMD) that is no brighter than the 100-lumens from another vendor - and these 160's consume exactly as many watts as the 100's.  One thing good about eBay is, I get the same thrill as going to Las Vegas.  The same rush you have by clicking "buy" as you do putting chips down on the roulette table...

[Corporate666]

Getting it by-the-watts is bad, but on eBay, getting it by lumens/lux is just as bad.  I have some 160-lumens LEDs (G4 connector, 12 5050-SMD) that is no brighter than the 100-lumens from another vendor - and these 160's consume exactly as many watts as the 100's.  One thing good about eBay is, I get the same thrill as going to Las Vegas.  The same rush you have by clicking "buy" as you do putting chips down on the roulette table...

The above is why I don't trust anything unless it comes from the manufacturer (never the seller, especially eBay!) - there are just too many variables for light output to judge based on a wattage, as you said.  However, it's highly likely you would not notice the difference between 100 lumens and 160 lumens anyway.  The eye does not respond linearly, so it takes about 4 times the brightness for a light source to appear twice as bright to an observer... and it takes double the lumens for an observer to judge one source to be "noticeably brighter" than another.  And about 30-50% more light output to even detect a difference.  So there could be a 60 lumen difference without you noticing much of anything by eye.

[sleemanj]

Yes the light emissivity given by pretty much all led vendors is of course largely meaningless in the real world (which is why when I sell LEDs I tell people the only way to know if it suits, is to try it), and power consumption isn't useful to judge light output, but I was just wondering if this was something along the lines of "3W" doesn't mean "3 Watt" or something in the "led trade".  Seems it's just hokey marketing.

One of these days I'm going to get around to building a test rig to see if I can make quantitative relative measurements of various LEDs.  I'm thinking a large box painted white inside, DUT in the center with the point-source (or best equivalent) of each DUT put at about the same level.  Then a few light sensors such as http://www.adafruit.com/products/439 placed in fixed locations looking at different areas of the box.

[Fsck]

Yes the light emissivity given by pretty much all led vendors is of course largely meaningless in the real world (which is why when I sell LEDs I tell people the only way to know if it suits, is to try it), and power consumption isn't useful to judge light output, but I was just wondering if this was something along the lines of "3W" doesn't mean "3 Watt" or something in the "led trade".  Seems it's just hokey marketing.

One of these days I'm going to get around to building a test rig to see if I can make quantitative relative measurements of various LEDs.  I'm thinking a large box painted white inside, DUT in the center with the point-source (or best equivalent) of each DUT put at about the same level.  Then a few light sensors such as http://www.adafruit.com/products/439 placed in fixed locations looking at different areas of the box.

could always DIY an integrating sphere.

[Rick Law]

Yes the light emissivity given by pretty much all led vendors is of course largely meaningless in the real world (which is why when I sell LEDs I tell people the only way to know if it suits, is to try it), and power consumption isn't useful to judge light output, but I was just wondering if this was something along the lines of "3W" doesn't mean "3 Watt" or something in the "led trade".  Seems it's just hokey marketing.

One of these days I'm going to get around to building a test rig to see if I can make quantitative relative measurements of various LEDs.  I'm thinking a large box painted white inside, DUT in the center with the point-source (or best equivalent) of each DUT put at about the same level.  Then a few light sensors such as http://www.adafruit.com/products/439 placed in fixed locations looking at different areas of the box.

Visit a camera store and look at their photometers.  That may give you some ideas.

I went as far as testing (my 160's vs 100's) using my old film SLR - which has a photometer.  Measuring the relative reflected output of light shine onto a standard 8.5x11 white "copier paper" at a distance.  Keeping the same exposure setting and adjusting the distant of LED to reflecting paper gives the relative light output.  (In square - ie: 2x the distance cuts the light to 1/4).   But, I could not easily measure the difference in "light cone angle" as one LED may spread a wider beam, nor could I easily account for possible differences in color temperature.  So I gave up on that venture.

[georges80]

Seriously - alibaba, ebay, china etc and you wonder if their specs are credible... 

In the US it's easy enough to order Cree or other reputable LEDs from places like Digikey/Mouser etc. Yes, you won't get a bag of LEDs for $1 but you will be able to order a product that has a datasheet that is accurate and also be able to choose lumens/watt and colour tint etc AND receive that product.

Chinese drivers (the cheapie $1 for a bucket type)  yes, and you wonder why they may only last minutes/weeks/months etc. Another product that the spec can be very optimistic and often those drivers are poorly built (bad choice of caps and/or diode ratings and/or resistor ratings). Often with large unshielded inductors that sit on top of the board held only by their wires. Typically very poor or non-existent thermal path. Poor documentation. Mostly these are the same drivers that are used in the generic and equally poor quality chinese torches/flashlights (the dollar store variety).

I've been in the LED driver business for quite a few years now so write this with some experience...

cheers,
george.