LED's for DUMMIES: High CRI ... Efficiency...

jso902

Enlightened
Joined
Feb 5, 2014
Messages
370
Why are high CRI LEDs less efficient than a regular LED? lumen per lumen...
Is it due to an extra tint that scales back an over powering wavelength to allow the less powered spectrum to be 'more' visible?

LEDs typically emit in the blue-UV section.

Different Phosphor tints helps redirect energy to lower wavelengths that help fill in the light spectrum... But in the process, energy is lost to heat and some wavelengths reproduced will land outside of the visible spectrum.
High CRI LEDs use a red phosphor that generates a ton of light that is at >630nm and well past 700nm all where eye sensitivity is low to 0. This is essentially wasted energy.
To clarify, most of the heat is not from the LED, but is made from the driver/electronics behind the LED
LED make mostly LIGHT, and the heat is not from the output, its from the process running the LED [,electronics], behind it. The BEAM is light.


And if that's the case, why don't they just make 3 LEDs onto 1 die so they get the entire RGB spectrum like a computer monitor/tv?

Multiple LEDs may cause shadowing effects that may not reflect well. However, there are multiLED lights that attempt to supplement a red hue to fill in the missing spectrum. But this only works with certain white LEDs.

What role does an LED dome play?

It helps direct the light.

If the above is wrong, please let me know so it can be corrected.
 
Last edited:
W

Wooperson

Guest
Hey jso902,

White LEDs are blue or near-ultraviolet LEDs that have phosphorescent coatings on the dies (region that emits the light). These phosphorescent coatings look yellowish-white. How the coatings work is that they absorb light in the blue/near-UV wavelengths and "shift" those wavelengths to larger wavelengths; this process is called Stokes shift. Larger wavelengths of light have lower energy than smaller wavelengths such as blue/near-UV. Since high energy waves are shifted to low energy, larger wavelengths such as red, yellow, green, etc. that make up white, that energy has to be lost and transferred somewhere else. This is usually heat and other non-light emitting quantum effects. Low-CRI light sources usually have narrower or highly biased spectra that do not represent the smooth spectrum of blackbody (such as the sun (not exactly, but close enough) or incandescent filaments) light sources. This is because they use a phosphor coating that is simple and made of very few components that perform Stokes shift to only those sets of narrow/biased spectra. Now, if you are using more phosphor components to add more colors to the spectrum (higher CRI), you are not only diluting the phosphor that's already there, but adding many other phosphors. The dilution makes the color shifting more inefficient as you get less color intensity. Additionally, addition of more phosphor also increases loss due to more Stokes shifts occurring. The higher CRI lights have more reddish, high wavelength components in the spectrum. Since you are performing a larger Stokes shift, you are dissipating more energy. That's why it's more inefficient.

Making an LED with three dies is an interesting idea, but it has a few problems. Manufacturing is more precise and difficult to do since you are handling three LEDs in the space of one. The second problem is that three light sources will cause three shadows of different tints. Unless your light source is as small as the pixels on the computer screen, you will get some ugly colored beam profiles.

I hope that clears things up for you.

-Woo
 

jso902

Enlightened
Joined
Feb 5, 2014
Messages
370
Holy smokes. I had to read that several times before I understood what you said. And I'm kinda embarrassed to know so little about something I use everyday... Ie flashlights, TV's, laptops,

Having said that, are all xpg leds the same with different domes? (I.e tint/bins)?
If domes determine the color, is that why there is an aura around the hot spot of my light?
If so much energy is lost to heat, why don't we move the tint/dome further away from the emitter? Or use lens filters instead?
And if what you said about led spectrum is true, are modifiers removing the dome exposing a UV hazard?
 
Last edited:

TEEJ

Flashaholic
Joined
Jan 12, 2012
Messages
7,490
Location
NJ
Holy smokes. I had to read that several times before I understood what you said. And I'm kinda embarrassed to know so little about something I use everyday... Ie flashlights, TV's, laptops,

Having said that, are all xpg leds the same with different domes? (I.e tint/bins)?
If domes determine the color, is that why there is an aura around the hot spot of my light?
If so much energy is lost to heat, why don't we move the tint/dome further away from the emitter? Or use lens filters instead?
And if what you said about led spectrum is true, are modifiers removing the dome exposing a UV hazard?


The domes don't dictate the differences, they distribute what IS emitted....and aspects of that distribution impact some of those factors. (IE: You are focusing on domes instead of phosphors, confusing the issues)

The primary differences in tints, etc, are from the phosphors used.

So, for example, you would not have all xpg be the same with different domes, you'd have them all have the same domes, and different phosphors.


The energy lost to heat is mostly from the electronics themselves, so you'd be moving the wrong part. (Again, the dome is not the important part) IE: You protect the LED by drawing heat away from the electronics, using heat sinks, fins, or whatever helps remove head from that part of the flashlight. The dome has ZERO to do with the heat management or effects of heat. Remember, the LIGHT from the LED is NOT the hot part, the electronics BEHIND the LED is the hot part.

Primary heat issue concept to get: For an incandescent light bulb, almost ALL it makes is heat, with a small amount of light...so, MOST of the energy put into an incan bulb is emitted as infrared radiation, with a small percentage being visible light. That means its mostly heat not light, and, the beam itself IS the hot part. For an LED, its MOSTLY making LIGHT not heat, so the BEAM is not what's hot, its the electronics behind the LED.



The phosphors are what are shifting the UV to visible light, not the dome (Again, focus on the phosphors, the dome is not important, and, not protecting from UV, as the UV was converted to visible light by, yes, the phosphors...and, so, no, there's no real UV hazard from de-doming)


IE: The color, tint, CRI, are from the phosphors.

Energizing the phosphors uses energy...so, for any juice that flows INTO the LED, LESS LIGHT comes out, because MORE of that juice was used to make more phosphors emit the light at the desired wavelengths. The more you change the CRI, by using more phosphors, the less efficient the LED is at EMITTING light.

So, to get a higher CRI, you lose efficiency, lumens and cd, for any given drive level. (Its more efficient to produce some wavelengths, and less efficient to produce others....)


The "aura" around the hot spot is called the corona, and, its not because of the LED, its because of the way a reflector focuses a beam.

If you've ever played pool, or had to ricochet something, you've possibly noticed that things tend to bounce in complimentary angles...IE: They bounce off at the same angle they impacted at.....straight down gives straight back up (Dribbling a basketball), if you bank a pool shot off a bumper, if you impact the bumper at 15º, the ball bounces off it 15º to the other side, etc....And, when an LED, or bulb, etc, emits light, inside of a reflector bowl, the light is not ALL from the exact same POINT in space, its emitted across the entire surface of the LED.

SOME of the light will be emitted from a point on the LED that ricochets exactly dead center of the beam, and, as you move farther from that perfect spot on the LED, SOME of the light emitted will be LESS perfectly focused, and, NOT be centered in the middle of the beam. The light that's missing the perfect focus, but still gathered by the reflector and sent down range, has the best focused light in the middle, with a donut of less focused light around it. There will also be some light that missed the bowl entirely, and was not focused at all, it just spilled out of the bowl w/o hitting the reflector at all. THAT light is called the spill.

So a normal beam is composed of a hot spot in the middle, surrounded by a corona, with some spill outside of the focused beam.

This is NOT an artifact of the LED per se, in that all LED, bulbs, etc, form this pattern due to the REFLECTOR. (The SMALLER the LED, the easier it is to have MOST of the light it emits come from a point in space closer to perfect focus, and, that's a primary reason a smaller LED can throw farther for the same lumen output....the light is easier to focus. This concept is also referred to as apparent surface brightness, as the reflector "Sees" more light coming from a smaller surface, so, for the same lumens, its going to have higher lux.

Part of a dome's function is to distribute the light emitted by the LED, so it hits the reflector in a way that they thought would appeal to users.

The primary reason that modders REMOVE the dome, is because they want a tighter beam, to throw farther, and NOT the broader beam that the light's maker thought the general public would want. Losing the dome tends to reduce the flashlights' published lumens, which the marketing departments are loathe to do...as the general public has no clue what lux is. Removing the dome increases the apparent surface brightness of the LED, which increases the light emitted closer to perfect focus for the hot spot, but, which reduces the peripheral light that counts towards total lumen output.


The "Hole in the middle of the beam" pattern you see with a poorly focused beam, is typically the shadow of the emitter, as that requires a HOLE in the bowl (The emitter sticks out into the reflector through that hole...). A well focused beam refocuses light to fill in that hole, centering the hot spot in the center of the beam.

:D






I hope this is a starting point for you to get a handle on the concept, and, let go of the dome-based misconceptions.

:D
 
Last edited:

SemiMan

Banned
Joined
Jan 13, 2005
Messages
3,899
Stokes losses are only part of the issue and not the biggest issue. The biggest issue is that the extra phosphors emit in the red to even infrared, wavelengths where the eye has low or no sensitivity.

A common technique is to mix a red led with a special white to achieve high CRI. The red led is less efficient than the blue pump led for the white, but the emitted light is at a wavelength of good eye sensitivity. Alternate narrow band red phosphors are also being developed for the same benefit.

Semiman
 

TEEJ

Flashaholic
Joined
Jan 12, 2012
Messages
7,490
Location
NJ
Stokes losses are only part of the issue and not the biggest issue. The biggest issue is that the extra phosphors emit in the red to even infrared, wavelengths where the eye has low or no sensitivity.

A common technique is to mix a red led with a special white to achieve high CRI. The red led is less efficient than the blue pump led for the white, but the emitted light is at a wavelength of good eye sensitivity. Alternate narrow band red phosphors are also being developed for the same benefit.

Semiman

True dat.

:thumbsup:
 
W

Wooperson

Guest
Stokes losses are only part of the issue and not the biggest issue. The biggest issue is that the extra phosphors emit in the red to even infrared, wavelengths where the eye has low or no sensitivity.

A common technique is to mix a red led with a special white to achieve high CRI. The red led is less efficient than the blue pump led for the white, but the emitted light is at a wavelength of good eye sensitivity. Alternate narrow band red phosphors are also being developed for the same benefit.


If the phosphors emit in wavelengths that we can't see, then it's correct that it's another sort of inefficiency in terms of what's visible and what's not. However, the main causes of inefficiency is still Stokes shift since there is a wavelength shift. Phosphors that emit light in invisible wavelengths are largely useless :fail:, but I don't think most LEDs have the problem of emitting in invisible wavelengths.

The red-mixing you talked about is usually a red LED that's supplementing the larger wavelengths which are usually missing from most low CRI LEDs that appear bluish. If red LEDs are used (or any LED without wavelength shifting), it is actually more efficient since there is no wavelength shifting and all produced light is emitted without losses. Our eyes see the full intensity. This is what Philips does with their "L-Prize Award Winning" bulbs. Not sure what you mean by mixing red LEDs with special white, but the closest effect is the anti-Stokes shift that's due to absorption of a lower energy wavelength and emission at a higher wavelength. This process requires the removal of energy from the environment and isn't common (if not unavailable) in LEDs.
 
W

Wooperson

Guest
Holy smokes. I had to read that several times before I understood what you said. And I'm kinda embarrassed to know so little about something I use everyday... Ie flashlights, TV's, laptops,

Having said that, are all xpg leds the same with different domes? (I.e tint/bins)?
If domes determine the color, is that why there is an aura around the hot spot of my light?
If so much energy is lost to heat, why don't we move the tint/dome further away from the emitter? Or use lens filters instead?
And if what you said about led spectrum is true, are modifiers removing the dome exposing a UV hazard?


Haha, sorry if I answered with too much detail. Don't be embarrassed!

TEEJ's response about the dome is correct.
 

SemiMan

Banned
Joined
Jan 13, 2005
Messages
3,899
Sorry but your answer is wrong. Stokes losses are an issue but the predominant loss is not radiometric I.e. stokes losses but photometric I.e. perceived brightness.

As blue LEDs are significantly more efficient than red LEDs especially at higher temps, a blue pumped narrow red phosphor can be more efficient than today's red LEDs ... Hence the research stokes losses and all.

You can't just mix any old white with red for best CRI. You need a special white that is above the black body as the red pulls it down.

Phosphors with QE >1 are possible but rare.
 
W

Wooperson

Guest
Sorry but your answer is wrong. Stokes losses are an issue but the predominant loss is not radiometric I.e. stokes losses but photometric I.e. perceived brightness.

As blue LEDs are significantly more efficient than red LEDs especially at higher temps, a blue pumped narrow red phosphor can be more efficient than today's red LEDs ... Hence the research stokes losses and all.

You can't just mix any old white with red for best CRI. You need a special white that is above the black body as the red pulls it down.

Phosphors with QE >1 are possible but rare.


If the phosphors release light in a limited, visible wavelength range, then the most significant losses must be due to losses before emission from the LED. If you are presuming that the LED emits in an invisible spectrum, then that would lead to more visual inefficiency. Of course, making such an LED wouldn't make much sense in the first place. Perceived brightness matters less if you are getting less light in the first place due to an inefficient phosphor. The perception sensitivity of light follows a Gaussian-looking plot centered around green. The eyes are more sensitive to reddish wavelengths than bluish wavelengths ("Human eye sensitivity and photometric quantities" Schubert of RPI). Therefore, if you are looking at a white light source with very little ultra-blue and infra-red wavelengths, perception intensity shouldn't matter as much since retinal ganglion cells will normalize their firing frequencies so that neighboring regions are not overly stimulated by different wavelengths; therefore you don't get much perception differences as compared to intensity differences due to Stokes losses.

Red LEDs are significantly more efficient. Cree cites their minimum red output as 45.7 lm/350 mA compared to a minimum of 30.6 lm/350 ma for blue LEDs. Maxima show the same trend with 73.9 lm/350 mA for red and 39.8 lm/350 mA for blue. (Cree XP-E data sheet, Cree, 2013). Lumileds also states that their efficacy is 72 lm/W for red LEDs and 37 lm/W for blue leds (Lumileds "All in 1 LED lighting solutions guide"). A blue pumped red phosphor is still going through a conversion of wavelengths so it cannot possibly be more efficient than a die that only releases in a certain wavelength. More conversion usually results in more intersystem crossing and other quantum losses.

Our eyes cannot detect light phase shifts, but can only detect the final superposition of light. That's why we can't differentiate different color waves combined into one unless there is a shadow. The emission spectrum can't be "pulled-down" by a red LED, the red can only supplement and fulfill the larger wavelength emission spectrum. Shifting spectra only happens if there is some sort of absorption/emission phenomena. Pumping blue light to get red wavelengths is counterintuitive for higher efficiency since you have to shift and therefore dissipate the lost energy (as stated above). Therefore, you can actually mix red-emitting phosphors for a higher CRI since our eyes can't tell the difference. That's why many high CRI bulbs do that and include both red and blue LEDs. The blue usually takes advantage of the phosphor layer. Other LEDs just include a red phosphor because it's simpler to manufacture, but there is a reason that the multi-color LED designs usually win efficiency prizes.
 

TEEJ

Flashaholic
Joined
Jan 12, 2012
Messages
7,490
Location
NJ
Um, fellas, look at the OP.

I think a more basic discussion - for this thread at least, would help him better.

The poor guy asked what time it was, and you guys are going into the metalurgy of clock springs.

:D
 
W

Wooperson

Guest
LOL yeah, sorry jso902. Good point, TEEJ. I realized I'm not answering OP's question and rather doing literature search. I'm gonna go do something else now... :whistle: I think the material in this thread answers OP's questions pretty well and will continue answering for eternity.
 

TEEJ

Flashaholic
Joined
Jan 12, 2012
Messages
7,490
Location
NJ
LOL yeah, sorry jso902. Good point, TEEJ. I realized I'm not answering OP's question and rather doing literature search. I'm gonna go do something else now... :whistle: I think the material in this thread answers OP's questions pretty well and will continue answering for eternity.

I just noticed you're in Philly. I'm over the river in Lawrenceville.

Are you going to the Photon Fest?
 
W

Wooperson

Guest
I'm over at the University of Pennsylvania and I might be busy, but I'm definitely interested. Do you know where I could learn more about it?
 

TEEJ

Flashaholic
Joined
Jan 12, 2012
Messages
7,490
Location
NJ

SemiMan

Banned
Joined
Jan 13, 2005
Messages
3,899
If the phosphors release light in a limited, visible wavelength range, then the most significant losses must be due to losses before emission from the LED. If you are presuming that the LED emits in an invisible spectrum, then that would lead to more visual inefficiency. Of course, making such an LED wouldn't make much sense in the first place. Perceived brightness matters less if you are getting less light in the first place due to an inefficient phosphor. The perception sensitivity of light follows a Gaussian-looking plot centered around green. The eyes are more sensitive to reddish wavelengths than bluish wavelengths ("Human eye sensitivity and photometric quantities" Schubert of RPI). Therefore, if you are looking at a white light source with very little ultra-blue and infra-red wavelengths, perception intensity shouldn't matter as much since retinal ganglion cells will normalize their firing frequencies so that neighboring regions are not overly stimulated by different wavelengths; therefore you don't get much perception differences as compared to intensity differences due to Stokes losses.

Red LEDs are significantly more efficient. Cree cites their minimum red output as 45.7 lm/350 mA compared to a minimum of 30.6 lm/350 ma for blue LEDs. Maxima show the same trend with 73.9 lm/350 mA for red and 39.8 lm/350 mA for blue. (Cree XP-E data sheet, Cree, 2013). Lumileds also states that their efficacy is 72 lm/W for red LEDs and 37 lm/W for blue leds (Lumileds "All in 1 LED lighting solutions guide"). A blue pumped red phosphor is still going through a conversion of wavelengths so it cannot possibly be more efficient than a die that only releases in a certain wavelength. More conversion usually results in more intersystem crossing and other quantum losses.

Our eyes cannot detect light phase shifts, but can only detect the final superposition of light. That's why we can't differentiate different color waves combined into one unless there is a shadow. The emission spectrum can't be "pulled-down" by a red LED, the red can only supplement and fulfill the larger wavelength emission spectrum. Shifting spectra only happens if there is some sort of absorption/emission phenomena. Pumping blue light to get red wavelengths is counterintuitive for higher efficiency since you have to shift and therefore dissipate the lost energy (as stated above). Therefore, you can actually mix red-emitting phosphors for a higher CRI since our eyes can't tell the difference. That's why many high CRI bulbs do that and include both red and blue LEDs. The blue usually takes advantage of the phosphor layer. Other LEDs just include a red phosphor because it's simpler to manufacture, but there is a reason that the multi-color LED designs usually win efficiency prizes.

At first I thought you knew what you were talking about but rereading what you wrote shows a lack of understanding. Quoting and compiling from Wikipedia and other sources does not make you knowledgeable.

You stuck your foot in your mouth when you used lumens to discuss efficiency of red and blue LEDs. Lumens is useless for that discussion. Blue LEDs are far more efficient than 620-630nm LEDs and so yes blue pumping a narrow red phosphor makes a ton of sense and hence why many companies are working on narrow red phosphors (and green).

You changed feet when the tried to explain how adding in red would not pull down the white point. Seems you are not familiar with blackbody curves and 2d color spaces.

So ... As previously stated. High CRI LEDs use a red phosphor that generates a ton of light that is at >630nm and well past 700nm all where eye sensitivity is low to 0. This is essentially wasted energy. Red LEDs in higher CRI bulbs are narrow spectrum emitters and emit where the eye us fairly sensitive. They are not very efficient radiometrically especially when hot, but the spectrum is all "useful", no wasted energy.
 

SemiMan

Banned
Joined
Jan 13, 2005
Messages
3,899
Um, fellas, look at the OP.

I think a more basic discussion - for this thread at least, would help him better.

The poor guy asked what time it was, and you guys are going into the metalurgy of clock springs.

:D

Agreed but unfortunately an answer provided while useful to the argument was wrong in its conclusion :)
 

jso902

Enlightened
Joined
Feb 5, 2014
Messages
370
Woah! LoL...I don't mind the extra info...
Let's see if I understand this correctly...
LEDs classically emit in the blue-UV section.
Different Phosphorus tint helps redirect energy to lower wavelengths that help fill in the light spectrum... But in the process, energy is lost to heat and some of the wavelengths overshoot into the IR spectrum.

So... What's the point of the dome? And why do I hear people talk about tint shift when a person removes it?
When you say distribute it, is that directing the light? I.e flooding the LED?

The blue / red confuses me a little.
1. I'm surprised LEDs are more efficient to the higher light spectrum. Is this because phosphorus's valence electron shell is higher? ( i.e. shorter fall from one valence level and producing higher frequencies?)

2. My understanding of red vs blue perception is that red was a lower frequency and required less energy to increase intensity while blue required substantially more energy to be more visible.
 
W

Wooperson

Guest
Yes, the first part is correct! However, overshoot is one thing manufacturers try to avoid since overshooting into the invisible wavelengths is pointless for illumination.

The dome acts as a lens that directs the light in a preferable pattern. For example, surface mounted LEDs are more useful if light is emitted in one direction and away from the electronic circuit components nearby. The dome focuses the light so that it projects forwards and away from the board which the LED is mounted on. If you remove the dome, you don't get this focus so light spreads out in all directions that the LED die faces.

Removing the dome essentially removes another material the light has to pass through. The tint shift occurs due to a few reasons, but the major reasons I can see are that different wavelengths are focused differently by the dome (some color waves are bent more so than others as light moves from one medium to another) and some wavelengths reflect at different angles within the dome. When you remove the dome, you change how different wavelengths of light are focused so you get some perceived changes in tint in the beam. What the exact effects are, I'm not totally sure.

Also:
1. Lower wavelength light such as blue are higher in energy and requires electrons to be raised to a higher energy level. This requires more energy input to do per electron. Red is lower energy so electrons don't have to be raised to such a high energy level as blue. Therefore, you can raise more electrons to a level that emits red light than blue light with the same amount of electrical energy.

2. Yes, that's true. Our eyes are more sensitive to reddish wavelengths than bluish wavelengths. Therefore, you need more blue to be as visible as red.
 
Top