Cree says XM-L (T6) is 40% efficient - how many lumens per watt is that?

WmArnold1

Newly Enlightened
Joined
Dec 24, 2004
Messages
136
Cree's XLamp/XM-L High-Bay reference design document is a facinating thing to read... Indeed; the statement that I found most profound therein was: "The XLamp XM-L LED operates at up to 10 watts of electrical power, depending on the drive current, and requires a heat sink to dissipate this thermal load. About 40% of that energy is converted to radiant flux and the rest to heat..." (ref: here - under Thermal Requirements)

Whoa! Can it be that simple? :devil: Bear with me, this really got me thinking..

Imho, they're probably alluding to a T6 bin emitter w/o lens running at 700 mA and 3.337 Volts, but that is *still* amazing to me because too many people get hung-up trying to convert luminous flux (Lumens) into radiant flux (Watts) using the widely misunderstood 683 Lu/W value from Wikipedia..

Keep-it-simple: 40% of 700 mA * 3.337 Volts is 0.93436 Watts of radiant power emitting from said LED, give or take a little..

At this point I thought; don't get too excited; Flashlight-Wiki states here that that the XM-L (T6) emits between 280 & 300 Lumens when driven at 700 mA; So, we're nowhere near the maximum output yet. Let's just do some math:

290 Lu / 0.93436 W ==> 310 "XM-L Cool White" Lumens per emitted Watt!

Let's consider that number in comparison to Wikipedia's vaulted 683 Lu/W for pure 555 nm "green" light. Countless people have said that our eyes are most sensitive to green light, but I have nearly gone blind reading stuff that never addresses Lu/W for XM-L lights or any white LED that I would recognize in a flashlight... But wait; we just calculated that ourselves! :naughty: And, for CPF technical purists; Cree documents XM-L's relative spectral power distribution here. Further, if the spectral power distribution between led's is fairly similar, then my 310 Lu/W value shouldn't vary much between LED's either.

Of course I don't know the accuracy or details behind the 40% efficiency that someone slipped into one of Cree's application notes, but I don't think I've ever seen it so concisely published before and I hope that people reading this will embrace an approximately 310 Lumens-per-Watt value for XM-L emitters. The main thing to note is that it's only 45% of Wikipedia's 683 value for pure green.

Last, but not least, I assume that radiant efficiency is proportional to luminous efficacy and both obviously degrade as power levels increase from 700 mA. For example, the radiant efficiency of said XM-L (T6) emitter drops to 28% at 3A corresponding with its efficacy drop from 124 to 94 Lu/W - i.e. 72% of said emitter's power has to be dissipated as heat at 3A - Btw, didn't MikeAusC already try to demonstrate 75% dissipation for XM-L here? :)

==> Does anybody see any glaring flaws in my logic or care to post refinements upon the 310 Lu/W value that I derived for "XM-L Cool White"?

Looking forward to your posts, William
 
Last edited:

Kinnza

Enlightened
Joined
Nov 15, 2005
Messages
223
Location
Spain
LER (luminous Efficacy of Radiation), as is called the figure of lm/emitted watt, varies with spectrum, so actually the exact figure depends of the color bin.

All your logic is perfect, based on average values, Cree coolwhite is about 310lm/W. I have measured from 306lm/W (6650K) to 317lm/W (6200K).

I cant find right now the value for the typical coolwhite spectrum on Cree datasheet, but for Neutral White LER is 335lm/W and for Warm White, 315lm/W.

The absolute max LER for white light is below 500lm/W, and that for low color rendering. With current phosphor technology, getting about 400lm/W is the best manufacturers can get, way less for high CRI spectrum.
 

WmArnold1

Newly Enlightened
Joined
Dec 24, 2004
Messages
136
Thanks for weighing-in, @Kinnza! You are obviously an expert regarding LER & CRI!!!

I always hope to see LER published along with nice spectrum graphics because it's a single number and relatively constant, but LER is unduly hard to find.

Respectfully Yours, William
 

slebans

Enlightened
Joined
Mar 1, 2010
Messages
457
Location
Moncton, NB Canada
Cree's XLamp/XM-L High-Bay reference design document is a facinating thing to read... Indeed; the statement that I found most profound therein was: "The XLamp XM-L LED operates at up to 10 watts of electrical power, depending on the drive current, and requires a heat sink to dissipate this thermal load. About 40% of that energy is converted to radiant flux and the rest to heat..." (ref: here - under Thermal Requirements)

==> Does anybody see any glaring flaws in my logic or care to post refinements upon the 310 Lu/W value that I derived for "XM-L Cool White"?

Looking forward to your posts, William

There is no need to estimate as the specified document contains a subset of the LM-79 data. Scroll down to the bottom of page 19 and the top of page 20. Radiant flux is the key value as you can now derive both LER and Wall Plug Efficiency for each of the three luminaire configurations.
I'm on my way out the door but will post the derived values when I return later this morning.

Stephen Lebans
 

Kinnza

Enlightened
Joined
Nov 15, 2005
Messages
223
Location
Spain
There is no need to estimate as the specified document contains a subset of the LM-79 data. Scroll down to the bottom of page 19 and the top of page 20. Radiant flux is the key value as you can now derive both LER and Wall Plug Efficiency for each of the three luminaire configurations.
I'm on my way out the door but will post the derived values when I return later this morning.

Stephen Lebans

The problem with those figures is they refer to whole luminaries measured into IS. So secondary optics affect the result, by having differential transmission at each waveband. The luminary without lens should correspond the closest to the LED's spectra. LER=17110lm/60.85W=281.18lm/W.

I have measured CW with this LER (and lower), but I dont think it is representative of the range, as it is outside ANSI bins, as CCT is over 7000K. Color bin is not cited, but x,y coordinates correspond to a 0C. WPE of the whole luminary very good, at 38.5% (60.85W out vs 158W input) for XML running at 1.4A. I think Cree choose this color bin in order to get the best radiant efficacy, but I dont think it is representative of CW any of us would like to use (the idea that using that tone to match MH is tricky as currently MHs are available on a full rage of CCTs). Indeed, I wouldnt call that tone "white".

LER with optics raises to 303lm/W (Carclo) and 310.8lm/W (Ledil), probably due to reduced transmittance at wavelengths on the extremes (blue mainly) and different current level (1.8A with Carclo).

This report allows to know with accuracy the efficiency of XM-L T6 bin on typical operating conditions (1.4 and 1.8A, good heatsinking and mounting) at 38,5% (1.4A) and 33.2% (1.8A)
 
Last edited:

slebans

Enlightened
Joined
Mar 1, 2010
Messages
457
Location
Moncton, NB Canada
The problem with those figures is they refer to whole luminaries measured into IS. So secondary optics affect the result, by having differential transmission at each waveband. The luminary without lens should correspond the closest to the LED's spectra. LER=17110lm/60.85W=281.18lm/W.

I have measured CW with this LER (and lower), but I dont think it is representative of the range, as it is outside ANSI bins, as CCT is over 7000K. Color bin is not cited, but x,y coordinates correspond to a 0C. WPE of the whole luminary very good, at 38.5% (60.85W out vs 158W input) for XML running at 1.4A. I think Cree choose this color bin in order to get the best radiant efficacy, but I dont think it is representative of CW any of us would like to use (the idea that using that tone to match MH is tricky as currently MHs are available on a full rage of CCTs). Indeed, I wouldnt call that tone "white".

LER with optics raises to 303lm/W (Carclo) and 310.8lm/W (Ledil), probably due to reduced transmittance at wavelengths on the extremes (blue mainly) and different current level (1.8A with Carclo).

This report allows to know with accuracy the efficiency of XM-L T6 bin on typical operating conditions (1.4 and 1.8A, good heatsinking and mounting) at 38,5% (1.4A) and 33.2% (1.8A)

Hi Kinnza,
I was really only interested in the luminaire spec without secondary optics in regards to LER and WPE calculations. It is interesting to see how the addition of a secondary optic affects not only efficiency but LER as well. I always thought about the optic in terms of loss without realizing that not all wavelengths are effected equally.

I spent an hour trying to make the public Cree PCT match the values in the document but without success. The Cree engineers obviously have a more powerful/flexible tool not available to us. There are several issues surrounding power supply efficiencies and drive currents that do not add up. Specifically, I don't see how they state a drive current of 1.4 amps unless they are using ultra low voltage XM-Ls.

My last issue has to do with the reference design for the luminaire without any secondary optics. You could not place this exact design in a warehouse without protecting the LEDs in some fashion. Bugs and dust would soon lower the efficiency of the luminaire, leaving the user without an easy method to remove the accumulated grime. I realize it is simply a reference design, and as you said--Cree needed to produce a design exceeding the efficiency of the comparative MH fixture--but I feel like they are cheating a bit by not including the real world losses from using a protective cover of some type.


Stephen Lebans
 

slebans

Enlightened
Joined
Mar 1, 2010
Messages
457
Location
Moncton, NB Canada
This report allows to know with accuracy the efficiency of XM-L T6 bin on typical operating conditions (1.4 and 1.8A, good heatsinking and mounting) at 38,5% (1.4A) and 33.2% (1.8A)

Kinnza, do you remember last year when we looked at a reference design, driving the LEDs at 700ma or less, and we only ended up with a 30% Wall Plug Efficiency? This Cree spec is a huge leap in one year. With the OSRAM reds in the low 40% range, XM-Ls at lower drive currents coming in at or just above 50% and power supplies at greater than 90% efficiencies, it would be possible to produce a grow light fixture with close to a 45% WPE. My design spec for the lettuce requires 135 PAR watts per square meter therefore I would only need a 300 watt fixture(s) to cover an entire square meter of bench space. This would have a huge impact on the budget for the annual electricity costs!

Stephen Lebans
 

Kinnza

Enlightened
Joined
Nov 15, 2005
Messages
223
Location
Spain
Yeah, this last year has been plenty of efficiency improvements, but just after the large improvement on past years, we noticed it less, as the gain is each times smaller in reference to the previous performance.

As soon as the new improved reds reach the market 40-45% WPE efficiency of grow lights are on sight.

135 PAR W/sq m? Thats a lot. Lettuce dont like more than 400micromols of photons per second and sq meter (uE), if what I read is true (not direct experience growing it). At 5uE/PAR W (reddish spectrum), it does a max of 80 PAR W/m2. I believe lettuce is usually grown under 150-250 uE/m2 irradiances. At 40% of WPE, you would need just about 100W/m2. With stacked stands, multiplying grow area, growth rate is not as important, so still less electricity would do the job, only way to do it profitable (about 5$ on electricity bill per sq meter harvest)
 

slebans

Enlightened
Joined
Mar 1, 2010
Messages
457
Location
Moncton, NB Canada
135 PAR W/sq m? Thats a lot. Lettuce dont like more than 400micromols of photons per second and sq meter (uE), if what I read is true (not direct experience growing it). At 5uE/PAR W (reddish spectrum), it does a max of 80 PAR W/m2. I believe lettuce is usually grown under 150-250 uE/m2 irradiances. At 40% of WPE, you would need just about 100W/m2. With stacked stands, multiplying grow area, growth rate is not as important, so still less electricity would do the job, only way to do it profitable (about 5$ on electricity bill per sq meter harvest)

The growth parameters you quoted are based on two issues with growing lettuce:
1) Lettuce leaves develop tip burn with high light and low humidity levels.
2) Lettuce will bolt(go to seed) at a specific time period.

Based on peer reviewed research, we have been growing lettuce at 600PPF for 22 hours per day at a constant 30 degrees celsius and 70% relative humidity.
For lighting, the target is a Daily Light Integral of 50 moles per meter squared per day. We are currently using just over 650 watts per square meter(Philips WW and LSG Cool White BR30s at around 20% WPE) in dumb purpose built fixtures that are DMX controlled).
Fans blow air directly onto the meristems to eliminate tipburn. The plants are harvested, as baby lettuce leaves, at just under 20 days from planting. CO2 levels become the growth limiting factor and are kept at a target of 1000ppm.

We are using multiple levels(multi shelved production carts). Growth rate, and therefore turnover, are the two keys to profitability.
Electrical is the single largest cost, mainly in the form of lighting but cooling is a significant expense as well. Luckily, this cost will continue to decline each year as LEDs reach new levels of efficiency.

Stephen Lebans
 

blasterman

Flashlight Enthusiast
Joined
Jul 17, 2008
Messages
1,802
XM-Ls at lower drive currents coming in at or just above 50% and power supplies at greater than 90% efficiencies, it would be possible to produce a grow light fixture with close to a 45% WPE

...And would cost a fortune. While the XM-L has remarkable efficiency and I don't dispute the math here what commercial fixture is going to use XM-L's at 700mA and remain cost competitive? Also, high current / low voltage emitters might make for practical driving in battery driven applications but not anything powered by 120/220 VAC.

As for plant growth, I would guess that using XTEs and newer bin 630nm rebels or Crees in a purpose built fixture would increase efficiency by 50-100% and decrease the price over standard BR30s.
 

slebans

Enlightened
Joined
Mar 1, 2010
Messages
457
Location
Moncton, NB Canada
...And would cost a fortune. While the XM-L has remarkable efficiency and I don't dispute the math here what commercial fixture is going to use XM-L's at 700mA and remain cost competitive?

You are absolutely correct, Blasterman. But with the price of LEDs dropping annually by 30% or more, and their efficiencies increasing anually as well, your statement will not be accurate at some point in the next few years. Further, from my specific business perspective, the capital cost of the fixture is secondary to the annual operating costs. As long as the fixture continues to operate during its depreciation schedule, the annual savings in electrical costs will more than make up for the depreciation amount on my income statements.

Also, high current / low voltage emitters might make for practical driving in battery driven applications but not anything powered by 120/220 VAC.

You have forgotten more than I know about building purpose built LED fixtures. But I do not understand your statement as power supplies are readily available with efficiencies exceeding 90%. Finally, I think it was you that had previously mentioned that whole building AC/DC conversion at the mains would be a viable alternative to individual fixture based power conversion.

As for plant growth, I would guess that using XTEs and newer bin 630nm rebels or Crees in a purpose built fixture would increase efficiency by 50-100% and decrease the price over standard BR30s.

Again, in terms of current costs, a purpose built fixture would be both cheaper and more efficient than the BR30s I am currently using. The BR30s currently have a WPE of around 20%. But we are only experimenting at this point and have not reached the stage where I would even contemplate having a fixture designed and built. We would need to purchase hundreds of purpose built fixtures in order for the entire process to be cost effective. Also, since our work is considered R&D we are able to claim roughly 50% of the costs.

Finally, I like the idea of using off the shelf technology. If a bulb dies a quick trip to Home Depot provides a solution. And due to the economies of scale, once LED replacement bulbs become more efficient and mainstream - I am not sure if a purpose built fixture will be able to compete from a first cost basis enough to cover the difference in WPE. But it is a simple accounting decision and I will go whatever way yields the most bottom line profit.

Stephen Lebans
 

Kinnza

Enlightened
Joined
Nov 15, 2005
Messages
223
Location
Spain
The growth parameters you quoted are based on two issues with growing lettuce:
1) Lettuce leaves develop tip burn with high light and low humidity levels.
2) Lettuce will bolt(go to seed) at a specific time period.

Based on peer reviewed research, we have been growing lettuce at 600PPF for 22 hours per day at a constant 30 degrees celsius and 70% relative humidity.
For lighting, the target is a Daily Light Integral of 50 moles per meter squared per day. We are currently using just over 650 watts per square meter(Philips WW and LSG Cool White BR30s at around 20% WPE) in dumb purpose built fixtures that are DMX controlled).
Fans blow air directly onto the meristems to eliminate tipburn. The plants are harvested, as baby lettuce leaves, at just under 20 days from planting. CO2 levels become the growth limiting factor and are kept at a target of 1000ppm.

All references I have found about lettuce uses way lower irradiation levels, with daily integrals between 15 and 20 mols of photons. With long photoperiod as 22/2, it means a PPF below 300uE/m2 (about 275uE/m2)for getting 20 mols/day.

Check this reference of NASA, at 1000ppm CO2 aswell: http://docs.lib.purdue.edu/cgi/viewcontent.cgi?article=1000&context=nasatr, tipburn observed at 300uE/m2 (23ºC air and roots all day, 16/8 photoperiod, 17 mols photon/day), density of 40 plants/m2 (initial) to 20-24 (final, harvest at 28d from seed, 20-30cm heads). Productivity in this conditions of 7g/m2*day from seed and 10g/m2*day from nursery (10-12 initial days).

Tipburn is due to low Vapor Pressure Deficit, which produces a reduced uptake of Ca. 30ºC seems too much for lettuce. If you couple it with high PPFD (thus, higher leaves' internal temperature and faster growth rate), risk of tipburn is strongly enhanced. For that NASA study and some others, I got the impression PPF over 400uE/m2 only adds marginal benefits. In other study I read photosynthetic saturation point of lettuce is about 470 uE/m2, and max photosynthesis rate, at 145uE/m2.

I understand that profit key is on shortening the harvest time, and that paying 2-3$ more for Kg may worth. But I think you should try to grow under lower irradiance and check the differential growth rate vs electricity cost and initial lamps cost. Maybe lowering PPF to 400uE/m2 dont get any noticeable lengthening of harvest time, or cost reduction of lowering PPF still more result profitable (if lengthening is just 1-2 days, for example). Reducing PPF would help lowering the cost of air flow aswell (as using LEDs instead of BCR, due reduced IR and photochemical quenching)

This other study found optimal conditions for lettuce (although harvest time parameter was not analyzed, just environmental conditions) as 22ºC (air), 20ºC (root), 1578ppm CO2, 75% RH, 1.3 m/s air speed and 216uE/m2.

The ideal spectrum for lettuce has been studied in detail. Just red+blue works perfect, with blue fraction minimum at 6% PPF, optimal at 8% and up to 12%. Although not studied so much, I believe far red suplementation could be useful on nursery stage to increase ground cover as fast as possible aswell as heads size.
 

slebans

Enlightened
Joined
Mar 1, 2010
Messages
457
Location
Moncton, NB Canada
In other study I read photosynthetic saturation point of lettuce is about 470 uE/m2, and max photosynthesis rate, at 145uE/m2.

Hi Kinnza - thanks for your interest.
I have caused this thread to move off topic but to address your point above, the yield saturation in response to PPF is due to tipburn, which reduces leaf expansion in the latter stages of growth(after canopy closure) when growth rates are maximal. The primary study we used showed a straight line increase in PAR versus yield up to 1000 PPF.

I will EMail you the research paper where we have been able to duplicate their yields, which are significantly above those of the papers referenced in your post.

Stephen Lebans
 

blasterman

Flashlight Enthusiast
Joined
Jul 17, 2008
Messages
1,802
But I do not understand your statement as power supplies are readily available with efficiencies exceeding 90%.

The XM-L is expensive per lumen compared to other packages even though it's near the front in terms of efficiency. You need to drive the things at 2amps to be cost competitive, and that makes your drivers $$$. -1

Most 'on the grid' power circuits are 120/220 volts. Converting from 120volts AC to ~3volts DC at several amps to run XM-Ls just isn't practical in terms of mainstream lighting because you lose efficiency the more you steer away from native voltage (or increase circuit costs). Sure, you can do this with a variety of buck type drivers, but then you need a DC source feeding them. It adds costs and costs efficiency. This is why Cree and Bridgelux are bending over to produce higher voltage packages - makes power supply design simplier and cheaper at higher voltages and lower currents - typically around 350mA or 700mA - both seem to be an unofficial standard. Somewhere Tesla is rolling in his grave with laughter.

As for plants, I've noted on many agricultural sites that to grow the more light hungry crops like tomatoes, 25watts per ^foot in terms of white halide is a common number and has been for years. That's around 250watts per square meter by my math. That may seem like a lot, but we're getting 2:1 efficiency advantages over halides in reefing forums, and that's with chinese based fixtures. Dedicated Cree or high bin Rebel fixtures are prolly closer to 3:1. 300watts of XT-E or 448nm top bin Rebels concentrating in a ^meter area would likely render the ground sterile.

Probably wouldn't take that many XT-Es to provide maximum vegetative growth in a ^ meter area. Usually it's red where you need such high energy levels (??) In any respect, they are easy to build and royal blue Crees are very cheap right now.
 
Last edited:

Kinnza

Enlightened
Joined
Nov 15, 2005
Messages
223
Location
Spain
Hi Kinnza,
I spent an hour trying to make the public Cree PCT match the values in the document but without success. The Cree engineers obviously have a more powerful/flexible tool not available to us. There are several issues surrounding power supply efficiencies and drive currents that do not add up. Specifically, I don't see how they state a drive current of 1.4 amps unless they are using ultra low voltage XM-Ls.

Coming back to topic, document dont state at all the current of each lamp. I just made a gross estimation. The no optics lamp emits 17110lm with 40LEDs, so it gives 428lm/LED, which is roughly the emission of a T6 at 1.4A. After your post, I wanted to check it in deep. XML at 1.4A emits 1.8x the flux of binning at 700mA, and thermal loss is about 12% at the temps specified. So 280lm(T6)*1.8*0.88=443lm at 1.4A. I would say that maybe actual temps on the final lamp were a little higher, so it explains the difference, but maybe the lamp were actually running a little below 1.4A.

Power drawn will clear the question: 158W at a driver efficiency of ~95%, so 150W, for 40LEDs, 3.75W/LED, which indeed means the no optic lamp was running below 1.4A. For the average Vf of the XML, about 1.25A. Checking again theoretical lm emission at 1.25A (~1.63x), 280*1.63*0.88=402lm. For average emission T6s (290lm@700mA), 290*1.63*0.88=416lm. This reverse engineering means that lamp was running at about 1.3A on low Vf XM-Ls.

For the carclo lamp, 17800lm net at an optical efficiency of 92% correspond to 19348 gross lm. Between 32 LEDs, 605lm/LED. I thought it correspond to 1,8A just by eye, but a look to the datasheet shows actual current was higher. At 1.9A (~2.37x), 280lm*2.37*0,88=584lm- For an average T6 (290lm), 290*2.37*0.88=605lm. Good match, so first bet is a current of 1.9A in this version. It match with low Vf XM-L aswell, for a power of about 5.3W/LED.
 

slebans

Enlightened
Joined
Mar 1, 2010
Messages
457
Location
Moncton, NB Canada
300watts of XT-E or 448nm top bin Rebels concentrating in a ^meter area would likely render the ground sterile.

That is an interesting thought.
Royal Blue is nominally 50% efficient. Let's forget about power supply losses and say that 300 watts of input would yield 150 watts of output.

For Solar irradiation, generally an average of 1000 watts per square meter is an accepted value. If we restrict the irradiation to the wavelengths between 400nm and 500nm we would derive a total value of 140 watts per square meter.

If I was a betting man, and I am, I would bet that 300 watts of Royal Blue XT-E would not render the ground sterile.
;-)

Stephen Lebans
 

WmArnold1

Newly Enlightened
Joined
Dec 24, 2004
Messages
136
I started this thread saying that Cree's XLamp/XM-L High-Bay reference design document is a facinating thing to read...

But everyone's posts here have been absolutely awesome and a great thing to study in their own right! Kudos to Kinnza, Slebans, and Blasterman :clap:

The thing I hope to remember from this thread is that Luminous Efficacy of Radiation (LER) is relatively constant for a particular emitter and family of emitters because it is a function of the output power spectrum. And, using LER, we can convert lumens into luminious Watts for all kinds of thermal considerations.

Hopefully, LER values will be published more often...

Thanks again for posting, William
 
Top