LEDs and Lumens

[Red Squirrel]

I've been toying with the idea of how neat it would be to setup some pot lights in the soffits of my house with some RGB lights that I can control along with some RGB LED strips for Christmas/seasonal lights that would change based on season.   There does not seem to be such thing other than using RGB wireless light bulbs, which use a proprietary way to control them.  I'd want something that I can use Arduino / RPI / or MCU or anything I can code for using PWM.

So I was thinking RGB LEDs that I would fit in a pot light enclosure with it's other circuitry (mosfets that get pwm signal, temp sensor, etc...).  Upon doing a search I came across these:

http://www.digikey.ca/product-detail/en/XMLCTW-A0-0000-00C3AAAA1/XMLCTW-A0-0000-00C3AAAA1CT-ND/5695391

At 15 bucks a pop I'd imagine this is equivalent to what you'd find in high powered flash lights and only one would be required per light, but looking at the lumens, it does not seem it's that bright.  At 1A I'm looking at 557 lumens with all 3 colours at full brightness.  (I divided the lumens by the test current then times by 1000.  Not sure if this is right)  Of course you probably would not want to drive them at full power even when you just want white light so maybe run them at like 900ma max so you'd get a bit less.   

Now a 100w bulb is over 1000 lumens, so this led at full brightness would only be half as bright as that.    Does this seem right  or am I looking at this wrong?  Do lot of the lumens from a 100w bulb go wasted because they're outside of the visible spectrum perhaps, while the LED lumens will seem brighter?  The datasheet even has a warning about safety for the blue spectrum, basically treating it like a laser, so I imagine these actually ARE very bright, just curious how I know this when looking at LEDs. 

I'm more or less just curious at this point,  as I have no immediate plans to start on such a project.  I would probably go with a flatter LED if I was to do this, as with this one a lot of the light would be wasted inside the pot light can.

[rs20]

http://www.ebay.com/sch/i.html?_nkw=RGB+LED+strip

Lumens are weighted by the human eye's sensitivity to light, so "the lumens from a 100w bulb go wasted because they're outside of the visible spectrum" is wrong, it makes no sense.

Keeping the LEDs cool is important, going beyond the test current may be a bad idea for this reason.

Doubling current does not quite double light output, there's a graph in the datasheet showing this. But it's fine for a rough approximation. At a cursory glance, the lumen figure you stated doesn't match up with the datasheet, so you'll want to double check that.

1600 lumens is a 100W light bulb, so 557 lumens should consume less than 34.8W or else it's less efficient. At 1A it's about 11.8W, so that's about right.

Also consider just separate R, G, B leds... I don't know what "pot lights" are, but it sounds like something where having all the light coming from exactly the same place isn't too useful.

[jitter]

I agree with rs20.

What I must add is that comparing LEDs with bulbs is like comparing apples and pears. The bulb radiates light in more or less a spherical pattern (not entirely true) whereas LEDs radiate light mostly in one direction (with a beam angle of about 120 degrees if there are no optics added).
The bulb sends light in a lot of directions where it's not useful and only a fraction of it is reflected back. When I replaced ceiling lamps with opaque cover that had 800 lm retrofits in them (E27 base) with fixed LED luminaires of 690 lm (also with opaque cover), I found that the latter were subjectively quite a bit brighter.  In the LED-luminaire there was a pcb with the LEDs facing down, sending almost all the light in the right direction. In the old luminaires, I would have had to use a retrofit LED bulb of about 1000 lm to subjectively get the same amount light.

I had to google "soffit" and "pot light", do I get it right that the pot lights are equipped with spot lights? If yes, that's directional light and that is the forté of LED, but you might want to use some optics to focus the light.

In industrial lighting, there are some standards in the making to promote compatibility between brands. For the interface you might want to google "DALI" and for the hardware "Zhaga".

[jitter]

Lumens are weighted by the human eye's sensitivity to light, so "the lumens from a 100w bulb go wasted because they're outside of the visible spectrum" is wrong, it makes no sense.

And a lux meter (lux = lumen per square metre, but you might use foot-candle) will take this bell shaped curve of the human eye's sensitivity into account. We are most sensitive to yellowish green and on both sides of this peak the sensitivity drops off.

Keeping the LEDs cool is important, going beyond the test current may be a bad idea for this reason.

I believe that LEDs in commercial applications are typically driven less hard than they can be to prolong life and make the thermal management easier.

Doubling current does not quite double light output, there's a graph in the datasheet showing this. But it's fine for a rough approximation. At a cursory glance, the lumen figure you stated doesn't match up with the datasheet, so you'll want to double check that.

1600 lumens is a 100W light bulb, so 557 lumens should consume less than 34.8W or else it's less efficient. At 1A it's about 11.8W, so that's about right.

You really have to take into account if the bulb is directional or not. A 100 W bulb cannot be compared to a 100 W spot light. The lumen figure is total light output, but says nothing about the radiatin pattern. That pattern will have different intensities (lux) in different positions from  the bulb. So, if you use incandescent spots now, find its specs like light output, beam angle, radiation pattern graph, etc.

[tggzzz]

I've been toying with the idea of how neat it would be to setup some pot lights in the soffits of my house...

I thought pot lights were placed inside lofts or basements, where law enforcement is less likely to spot them

[Red Squirrel]

Ohh So is lumens based on the TOTAL output and not a given area?  I figured there was a standard, like you measure a certain amount of square cm from a certain distance.  So when a 100w bulb says it is 1600 lumens, that is the total light going in almost all directions, while a LED that says is 500 lumens they are more concentrated into a certain direction so it would seem like more light?  Of course the one I posted may be a bad example as it has a decent spread, but still not as much as a light bulb.  So technically could one of those LEDs appear as bright as a 100w bulb?  I probably would end up going with many cheaper LEDs though, 10-15 bucks for something I can accidentally destroy while screwing up soldering is a lot of money. 

Interestingly I also found these through hole LEDs that appear to be insanely bright:

http://www.cree.com/LED-Components-and-Modules/Products/High-Brightness/4mm-Oval-P2/Screen-Master-4mm-Oval-RGB

Some are like 1500 lumens at 20ma, does that make sense?  I must be reading that data wrong. Maybe the typical value means when there are a lot of them together?

[IanB]

Lumens are weighted by the human eye's sensitivity to light, so "the lumens from a 100w bulb go wasted because they're outside of the visible spectrum" is wrong, it makes no sense.

Of course it's not wrong, it's just an accidental slip in the choice of words. Try it this way: "many of the photons from a 100 W bulb go wasted because they're outside the visible spectrum". The infra-red photons get weighted by zero so they don't contribute to the visible lumens.

To the OP: consider power input and visible output. A 100 W bulb is consuming 100 W (obviously). If you take a typical white LED with a forward voltage of 3.1 V and drive it with a (typical) current of 350 mA that is about 1 W. It's 1% of the power input, therefore it is not surprising if it is not as bright as a 100 W bulb. But if you compare lumens per watt the LED does much better. And it's still pretty bright if used for decorative lighting or spot illumination.

[rs20]

Of course it's not wrong, it's just an accidental slip in the choice of words. Try it this way: "many of the photons from a 100 W bulb go wasted because they're outside the visible spectrum". The infra-red photons get weighted by zero so they don't contribute to the visible lumens.

  He's comparing lumen values and indeed wondering if there's such a thing as an "invisible, infrared lumen", which by definition there is not. Look at the context:

Now a 100w bulb is over 1000 lumens, so this led at full brightness would only be half as bright as that.    Does this seem right  or am I looking at this wrong?  Do lot of the lumens from a 100w bulb go wasted because they're outside of the visible spectrum perhaps, while the LED lumens will seem brighter?

[Red Squirrel]

I just wanted to know if whether or not lumens count the light that is outside of the spectrum too, that's all I was wondering, as incandescent bulbs throw off quite a wide spectrum compared to LEDs so was wondering if the high value of a incadescent could not be really compared to LEDs due to that.  For example a TV remote emits a certain amount of light, but you can't see it.  So I thought that maybe they count that spectrum too when they advertise a bulb.   But that was answered so now I know it does not count the other spectrums.

[IanB]

At 1A I'm looking at 557 lumens with all 3 colours at full brightness.  (I divided the lumens by the test current then times by 1000.  Not sure if this is right)  Of course you probably would not want to drive them at full power even when you just want white light so maybe run them at like 900ma max so you'd get a bit less.

The light output from LEDs is not proportional to the current. The efficiency drops way off at higher currents. When they give the test current at 350 mA that is a coded way of saying "we recommend you drive this LED at 350 mA".

You still need to consider the power input. If you drive all four dies of that RGBW LED at 350 mA that's approximately 1 W per die for 4 W total. It's far short of 100 W. Even with the greater luminous efficacy of the LED they are hardly comparable. (That's why LED replacement light bulbs usually have lots of LEDs inside them.)

[Red Squirrel]

Guess what I was just surprised it is how low the lumens per price is.  I have a flash light that is 2000 lumens and uses a single LED module, for example, and I did not pay all that much for it and it will run for hours on a single lithium ion cell so I don't imagine it takes more than a couple watts at most.  Though I guess the difference is these are single color LEDs while RGB is 3 separate modules in the die so guess that will be more expensive too.  But even looking at individual modules they're in 10's of lumens.  I would need like 200 of them together to compare to the flash light I have, for example and it would cost more than what I paid for the flashlight. So I was just wondering if maybe I was looking at it wrong but guess these companies get big discounts by buying in the millions too, so probably how they can sell the flashlights for that cheap.   Or maybe they miss advertise the lumens?   Basically I'm just trying to compare lumens to known sources so I have a good idea of how bright leds I'd need.

Then there's these:

http://www.cree.com/LED-Components-and-Modules/Products/High-Brightness/4mm-Oval-P2/Screen-Master-4mm-Oval-RGB

How are they getting thousands of lumens at 20ma or so?  I must be reading that wrong.  Look at the 5th entry green one.

Edit: NM If I read more closely it looks like that's another unit of measurement and not lumens.

[IanB]

I have a flash light that is 2000 lumens

That's a dubious claim. It may be marketing wankery. Have you measured it?

Then there's these:

http://www.cree.com/LED-Components-and-Modules/Products/High-Brightness/4mm-Oval-P2/Screen-Master-4mm-Oval-RGB

How are they getting thousands of lumens at 20ma or so?  I must be reading that wrong.  Look at the 5th entry green one.

Edit: NM If I read more closely it looks like that's another unit of measurement and not lumens.

Yes, that unit is mcd, or millicandelas. That is a unit of brightness per unit solid angle, or intensity. When tightly focused even a small light source can appear very bright.

[rs20]

More specifically, 1 candela = 1 lumen per steradian. The solid angle of a complete sphere is 4pi steradians. A very tightly focussed beam is a very small number of steradians, which is how a modest number of lumens can become a large number of candelas.

[Red Squirrel]

I have a flash light that is 2000 lumens

That's a dubious claim. It may be marketing wankery. Have you measured it?

I imagine you need very expensive equipment to do so, so it's safe to say no.  But if lumens counts total light and not a given area, then it's more focused so maybe it just SEEMS brighter than say, a standard 100w bulb.

This is the flashlight I'm comparing to:

http://www.amazon.ca/gp/product/B00NL45PEI?keywords=cvlife%20flashlight&qid=1448159198&ref_=sr_1_1&sr=8-1

Did a bit more digging and none of the xlamp  XM-L modules even come close to 2000 lumens, so are most flash light manufacturers just outright lying?  Most LED flash light seem to be in the thousand lumens range.

[jitter]

Ohh So is lumens based on the TOTAL output and not a given area?

Yep (luminous flux), google "integrating sphere" and you'll find the device used for measuring the total light output of a light source. As the term "integrating" implies, it's total output (remember integrals in maths, loooong ago?).

So when a 100w bulb says it is 1600 lumens, that is the total light going in almost all directions, while a LED that says is 500 lumens they are more concentrated into a certain direction so it would seem like more light? 

Basically, that's it.
As an example: I have a luminaire hanging over my table with a E27 socket facing down, so the lamp base will be vertical and pointed towards the ceiling.
I tried a good old incandescent of 630 lm and compared it with a retrofit LED bulb which has the LEDs on a flat pcb facing down. That retrofit LED bulb was specced at 470 lm.
In this application, the LED bulb clearly threw more light on the table, even though it was the weaker one. The light was directed at the table.

That's why I don't like retrofit LED bulbs very much. By trying to emulate an incandescent bulb, they waste a lot of available light.
IMHO, the "classic" LED (i.e. the square or round chips) are best used for directional lighting. If you do need unidirectional from a LED bulb, then a LED filament bulb does a better job. At first I was suspicious obout their life span, but now big brands (e.g. Osram and Philips) deem them good enough to sell them under their own names.

Interestingly I also found these through hole LEDs that appear to be insanely bright:

http://www.cree.com/LED-Components-and-Modules/Products/High-Brightness/4mm-Oval-P2/Screen-Master-4mm-Oval-RGB

Some are like 1500 lumens at 20ma, does that make sense?  I must be reading that data wrong. Maybe the typical value means when there are a lot of them together?

No, those are not lumens, they're mcd, millicandela and as rs20 explained, they're different units. mcd is light intensity. And yes, that's a rather useless unit for comparison.

From wikipedia: "1 lx = 1 lm/m² = 1 cd·sr/m²".

Light intensity of two different light sources can only be compared when they've got the exact same specs for radiation angle and measuring distance from the source.
Example: two light sources have the same amount of luminous flux but one of them as a a radiation angle that's narrow, the other wide. When measured from the same distance, and in the centre of the beam, the narrow beam will result in a higher reading than the wide beam. But then, (i.e.) 45 degrees off centre, the narrow beam may be reading less than the wide beam...

So yes, a narrow beam of less total light output may indeed seem brighter (in the lit area) than a wide beam light. Apples and pears...
It's probably why EU regulators first made regulations for unidirectional light sources and left the directional light sources out. Now the directional light sources must also be specified by the manufacturers in lumen and they must specify only the light in a cone of 90 degrees (there are exceptions). All light that falls outside that cone must not be specified.
All of a sudded we found that a 50 W halogen spot light had less light output than a bog standard 40 W bulb? What? Apples and pears..., you can't compare them...

Long story short... it all boils down to choosing the right beam angle for the application and then choosing the right amount of light output for the job.

Did a bit more digging and none of the xlamp  XM-L modules even come close to 2000 lumens, so are most flash light manufacturers just outright lying?  Most LED flash light seem to be in the thousand lumens range.

2000 lm is a hell of a lot of light, it looks like they used the wrong unit in the flash light ad. If they meant lx, then it says nothing without knowing how far from the light source it was measured. I can easily make my lux meter "think" it's seeing direct sunlight by sticking the flashlight up close to the sensor. It will then read O.L. as it's over it's limit of 40,000 lx (direct sunlight is over 100,000 lx).