XP-G is better XP-E due to 2 factor:
1) larger area
2) more lumens and efficacy by mm^2 at the same current density (or less as you think)
And now we are talking about a graph which would represent the second factor more clear.
Last edited:
The new XP-G: another big announcement from CREE
- Thread starter YuccaPatrol
- Start date
1) larger area
2) more lumens and efficacy by mm^2 at the same current density (or less as you think)
And now we are talking about a graph which would represent the second factor more clear.
Marduke
Flashaholic
Yes, but because they were only rated for 1 watt, there is only a comparrison out to 350mA
TorchBoy
Flashlight Enthusiast
:ohgeez: http://www.philipslumileds.com/pdfs/DS46.PDF is the Luxeon III.
ergotelis
Enlightened
I think it is important to add here, i did some tests with xp-e and carlco and ims reflectors. At 20mm, i could get about 7000 lux and good beam with IMS, xp-e is R2 driven at exactly 1amp. So, for your mods, i guess xp-g+IMs is a win!Though, i found it strange the fact that i could see the lighting emitter outside the reflector! Carlco has not that problems, but it seems to focus a bit less(about 5000+ lux) with a better beam i think.
Wow, if I understand correctly (and I doubt I do), it was just 3 years ago leds could do 70 lumens at 1A, and now we have the S2 which can do 400 at 1A? Amazing. Good time to be a flashaholic.
outersquare
Enlightened
![Pampers Sensitive Baby Wipes, Clinically Proven, Fragrance Free, Unscented, Water Based, Hypoallergenic, Multi-Use, Hand Wipes 336 Wipes Total (6X Flip-Top Packs) [Packaging May Vary]](https://m.media-amazon.com/images/I/41KLBHu9CAL._SL500_.jpg)
Glenn7
Flashlight Enthusiast
Yeah and I keep hearing people on CPF say leds aren't getting much brighter and they are at the end of this technology and cant/wont compete with HID - I say Hmmmm....... we'll see :tinfoil::tinfoil::tinfoil:
baterija
Flashlight Enthusiast
- Joined
- Feb 7, 2008
- Messages
- 1,053
For the comparison Jtr was making thermal management is irrelevant. His computed numbers of higher lumens/current density are all about showing some of the other improvements in the XP-G. His computations simply demonstrate that the changes are not strictly explained by lower current density due to the bigger die. There is something there besides just cutting the dies bigger. One of your early questions was whether Cree is just cutting poor quality wafers into bigger dies to sell. His numbers seem to indicate exactly the opposite.
Now for real world applications like flashlights thermal management is a key design factor. The XP-G will make thermal management easier though. It produces less waste heat versus an XR-E or XP-E whether you compare at constant output, constant drive current, or constant input power thanks to the improved lumens per watt. It also has lower thermal resistance so its easier to move heat away from the emitter. There are already plenty of lights that push high currents, and manage the heat. The XP-G would produce less thermal issues with those same designs.
:hitit:
Marduke
Flashaholic
Now, I must say I was actually surprised when I saw this graph for the first time after I clicked "Finish".
Here is what I choose to do:
I attempted to pick the major LED's that CPF has been lusting since high flux LED's came out. Since Lux I's only go to 350mA and the rest go higher, I choose to go with Lux III's, figuring they are close enough to be representative of the time. Just so you know, I did NOT "cherry pick" these bins to make the graph look as clean as it does. Why did I choose the bins I did? I attempted to choose the bins which are not necessarily the lowest or highest in a model's lifecycle, but rather what was an earlier "popular CPF" bin available in reasonable quantity early in the lifecycle. The "game changers" if you will.
With that in mind, I choose Lux III T, XR-E P4, XP-E R2, and XP-G R5.
The moment of truth:
LED's have indeed came a LONG way in such a short time.
Enjoy...
Here is what I choose to do:
I attempted to pick the major LED's that CPF has been lusting since high flux LED's came out. Since Lux I's only go to 350mA and the rest go higher, I choose to go with Lux III's, figuring they are close enough to be representative of the time. Just so you know, I did NOT "cherry pick" these bins to make the graph look as clean as it does. Why did I choose the bins I did? I attempted to choose the bins which are not necessarily the lowest or highest in a model's lifecycle, but rather what was an earlier "popular CPF" bin available in reasonable quantity early in the lifecycle. The "game changers" if you will.
With that in mind, I choose Lux III T, XR-E P4, XP-E R2, and XP-G R5.
The moment of truth:
LED's have indeed came a LONG way in such a short time.
Enjoy...
4sevens
Flashlight Enthusiast
Great work Marduke! I very clearly remember the jump from Lux III to the new XR-E P4 at the end of 2006 and at the beginning of 2007 with the Fenix P1D. It was a smash hit! That was one of the reasons Fenix caught on so quick (that and my marketing and distribution efforts).Now, I must say I was actually surprised when I saw this graph for the first time after I clicked "Finish".
Here is what I choose to do:
I attempted to pick the major LED's that CPF has been lusting since high flux LED's came out. Since Lux I's only go to 350mA and the rest go higher, I choose to go with Lux III's, figuring they are close enough to be representative of the time. Just so you know, I did NOT "cherry pick" these bins to make the graph look as clean as it does. Why did I choose the bins I did? I attempted to choose the bins which are not necessarily the lowest or highest in a model's lifecycle, but rather what was an earlier "popular CPF" bin available in reasonable quantity early in the lifecycle. The "game changers" if you will.
With that in mind, I choose Lux III T, XR-E P4, XP-E R2, and XP-G R5.
The moment of truth:
LED's have indeed came a LONG way in such a short time.
Enjoy...
It's nice to see the jump to XPE's and now XPG's are similar in magnitude.
Additionally, it's astonishing to see how flat the XPG curve is. The "droop" as people call it here is greatly diminished (I don't think it'll ever go away, it's just scale is different - ie. zoom out and you have a similar droop). This indicates some serious technological advancements with lowering the Vf which directly affects the generation of heat.
I have to repeat what so many have said in this thread. What an exciting time to be in with these LED advancements!
Gryloc
Enlightened
Hey there. I have been watching this thread and was unsure how to provide more clarification. This is an exciting time, and thanks, Marduke, for creating those initial graphs. Thank you, jtr1962, for bringing us some experience and for your actual measurements that assisted with some estimates of mine.
Those initial graphs were informative, but I wanted to see more (and I am not going to yell at you, he he). Therefore, I dove back in and created some graphs and charts over the past week. Efficiency comparisons are neat to see, but they do not tell much about performance versus the current input seeing how Cree does not even bin their forward voltage (which can greatly affect the efficiency from reel to reel). So, because I was used to the charts that were created from jtr1962 over the years (again, thank you!), I created my charts like his.
Since the XP-G has a die with an area of 2mm^2, I wanted to estimate the performance of the new emitter by pretending that I attached two dies from the XP-E in parallel and attaching them to the same XP-E package. The results will not be the same as that of measuring the output of an XP-G, but I found that the results are strikingly close! BTW, I am just going for the electrical and physical characteristics of the XP-E with double the die area of the EZ1000 die. I believe that the EZ900 (Gen II) did not exist at the time of jtr1962's testing of the XP-E, so I am not going to touch comparing this new generation of EZxxxx dies.
You can do the same with nearly any LED with more than 1mm^2 die area. The current density decreases with the increase in die area (if current stays constant) because the current is divided somewhat evenly between the entire surface. Since the current density drops, then naturally the Vf and the luminous efficiency will scale rather linearly, and will be lower! Try it out, divide the current and lumen measurements from a P7 emitter or even the large SST-90 emitter, and you will find that they act very similar to multiple 1mm^2 dies wired in parallel.
So, the following is the estimations of the characteristics of the new XP-G emitter:
(click for full-sized view)
Notice that in the above tables, I took the performance of the XP-E R2 bin, and pretended that I connected two in parallel. I doubled the current and luminous flux column as though the two are in parallel, and there was born some similar characteristics of the XP-G (albeit just the R3 bin equivalent). Vf stays the same since the die area is in parallel, so the total power will not change too much (hence the XP-G will run cooler). Notice the 248% increase of brightness at 1000mA compared to at 350mA (like what is stated in the datasheets and press releases). Pretty close. So, by multiplying the flux by an 1.13, the emitter begins to fall into the R5 bin. Notice the 140lm at 350mA and 346lm at 1000mA? Again pretty close to what was released in the press releases.
Next I show both the efficiency- and luminous flux versus forward current comparison in the form of a graph. I compared four bits of data:
-Actual measurements of the Cree XR-E R2 bin,
-A scaled up version of the XR-E curve to fit into the R5 bin (to make a dramatic comparison with the estimated XP-G R5)
-The estimated performance of the XP-G R5 (the efficiency relied on the forward voltage data of the tested XP-E R2)
-The data that I "extracted" or "digitized" manually from Cree's documents. I got the relative output versus forward current data from both the XP-G datasheet, and the forward voltage versus forward current data came from the same May 2009 Announcement document released that Marduke used. I scaled the relative output curve to fit into being a R5 bin (~139lm at 350mA).
Unfortunately, I digitized this data carefully by overlaying a transparent graph in Excel over the original pictures of the graphs, so my raw data is not as exact as Marduke's, nor is the data resolution as high. I believe that they are pretty close, though. Please treat my data as being very close, but not exactly the same as Cree's exact measurements.
These four series of data were overlaid in a single graph below:
(click for full-sized view)
I was happy to see that the curve for luminous flux versus current sat very closely to that of the simulation of the two paralleled XP-E dies. I understand why the efficiency curve would not overlay because I used actual data of the XP-E R2 sample that jtr1962 owned compared to what Cree released (which compared to any datasheet Vf info from Cree, your results may vary).
Finally, here is the raw "digitized" data that I got from the Cree documents put in the familiar form. I hope it saves others the pain of extracting data from pictures of graphs.
(click for full-sized view)
I hope these estimates can create a better picture as to how the new XP-G could act. Again, because Cree does not bin the Vf, efficiency will vary from part to part as usual.
Finally, do not forget that I used the curve from that of the XP-E, so I am not sure how the tiny emitter will fare with higher currents compared to my estimates. I displayed my data up to 1500mA to allow us to drool more, but I do not recommend operating the XP-G past its maximum rated current unless you know what you are doing. Actual luminous flux throughout the curve will also vary because your R5-binned XP-G will produce 139lm of output minimum at 350mA (so yours may produce more output at 350mA and at higher currents). Heatsinking and forward voltage will affect whether or not you will achieve a 250% increase in output at 1000mA versus 350mA.
Again these are not apple to apple comparisons, but more like apple-shaped-orange to apple comparisons :thinking:. I can add another bin of XP-G or other emitters to the graph later, but I do not know if I have the patience like Marduke to make that many versions of the graph :bow:. I do not plan on making a graph with the same curves, but only one bin lower either :grin2:.
If more info can be shared on the S2 bin, that would be great! I know the flux range at 350mA, but this press release shows how an XP-G of some bin (S2?) can output 400lm at 1000mA, which does not jive with my curves at all!
BTW, I played with the idea of comparing the many bins of emitters using the EZ900, EZ1000, and the estimated "EZ1400" dies based on die area, but I ended up with messy and somewhat conflicting results (due to my limited amount of actual data, or lack of flux figures of the XP-G to use).
Please leave comments and inform me of errors, or if I am mistaken with the concept of treating large dies like multiple 1mm^2 dies. Thanks!
Cheers,
-Tony
EDIT: Sorry for such high res pics (after you keep clicking on them to zoom in). I am too used to DSL and Cable internet. Plus, sorry for the weird color shifting happening to the text in my pictures. Between Excel '07, PSP 8, and image shack, there is something going on with the letters and lines. Please inform me via PM if any "anti-aliasing" or softening features can be turned off in Excel. Thanks for your patience.
Those initial graphs were informative, but I wanted to see more (and I am not going to yell at you, he he). Therefore, I dove back in and created some graphs and charts over the past week. Efficiency comparisons are neat to see, but they do not tell much about performance versus the current input seeing how Cree does not even bin their forward voltage (which can greatly affect the efficiency from reel to reel). So, because I was used to the charts that were created from jtr1962 over the years (again, thank you!), I created my charts like his.
Since the XP-G has a die with an area of 2mm^2, I wanted to estimate the performance of the new emitter by pretending that I attached two dies from the XP-E in parallel and attaching them to the same XP-E package. The results will not be the same as that of measuring the output of an XP-G, but I found that the results are strikingly close! BTW, I am just going for the electrical and physical characteristics of the XP-E with double the die area of the EZ1000 die. I believe that the EZ900 (Gen II) did not exist at the time of jtr1962's testing of the XP-E, so I am not going to touch comparing this new generation of EZxxxx dies.
You can do the same with nearly any LED with more than 1mm^2 die area. The current density decreases with the increase in die area (if current stays constant) because the current is divided somewhat evenly between the entire surface. Since the current density drops, then naturally the Vf and the luminous efficiency will scale rather linearly, and will be lower! Try it out, divide the current and lumen measurements from a P7 emitter or even the large SST-90 emitter, and you will find that they act very similar to multiple 1mm^2 dies wired in parallel.
So, the following is the estimations of the characteristics of the new XP-G emitter:
(click for full-sized view)
Notice that in the above tables, I took the performance of the XP-E R2 bin, and pretended that I connected two in parallel. I doubled the current and luminous flux column as though the two are in parallel, and there was born some similar characteristics of the XP-G (albeit just the R3 bin equivalent). Vf stays the same since the die area is in parallel, so the total power will not change too much (hence the XP-G will run cooler). Notice the 248% increase of brightness at 1000mA compared to at 350mA (like what is stated in the datasheets and press releases). Pretty close. So, by multiplying the flux by an 1.13, the emitter begins to fall into the R5 bin. Notice the 140lm at 350mA and 346lm at 1000mA? Again pretty close to what was released in the press releases.
Next I show both the efficiency- and luminous flux versus forward current comparison in the form of a graph. I compared four bits of data:
-Actual measurements of the Cree XR-E R2 bin,
-A scaled up version of the XR-E curve to fit into the R5 bin (to make a dramatic comparison with the estimated XP-G R5)
-The estimated performance of the XP-G R5 (the efficiency relied on the forward voltage data of the tested XP-E R2)
-The data that I "extracted" or "digitized" manually from Cree's documents. I got the relative output versus forward current data from both the XP-G datasheet, and the forward voltage versus forward current data came from the same May 2009 Announcement document released that Marduke used. I scaled the relative output curve to fit into being a R5 bin (~139lm at 350mA).
Unfortunately, I digitized this data carefully by overlaying a transparent graph in Excel over the original pictures of the graphs, so my raw data is not as exact as Marduke's, nor is the data resolution as high. I believe that they are pretty close, though. Please treat my data as being very close, but not exactly the same as Cree's exact measurements.
These four series of data were overlaid in a single graph below:
(click for full-sized view)
I was happy to see that the curve for luminous flux versus current sat very closely to that of the simulation of the two paralleled XP-E dies. I understand why the efficiency curve would not overlay because I used actual data of the XP-E R2 sample that jtr1962 owned compared to what Cree released (which compared to any datasheet Vf info from Cree, your results may vary).
Finally, here is the raw "digitized" data that I got from the Cree documents put in the familiar form. I hope it saves others the pain of extracting data from pictures of graphs.
(click for full-sized view)
I hope these estimates can create a better picture as to how the new XP-G could act. Again, because Cree does not bin the Vf, efficiency will vary from part to part as usual.
Finally, do not forget that I used the curve from that of the XP-E, so I am not sure how the tiny emitter will fare with higher currents compared to my estimates. I displayed my data up to 1500mA to allow us to drool more, but I do not recommend operating the XP-G past its maximum rated current unless you know what you are doing. Actual luminous flux throughout the curve will also vary because your R5-binned XP-G will produce 139lm of output minimum at 350mA (so yours may produce more output at 350mA and at higher currents). Heatsinking and forward voltage will affect whether or not you will achieve a 250% increase in output at 1000mA versus 350mA.
Again these are not apple to apple comparisons, but more like apple-shaped-orange to apple comparisons :thinking:. I can add another bin of XP-G or other emitters to the graph later, but I do not know if I have the patience like Marduke to make that many versions of the graph :bow:. I do not plan on making a graph with the same curves, but only one bin lower either :grin2:.
If more info can be shared on the S2 bin, that would be great! I know the flux range at 350mA, but this press release shows how an XP-G of some bin (S2?) can output 400lm at 1000mA, which does not jive with my curves at all!
BTW, I played with the idea of comparing the many bins of emitters using the EZ900, EZ1000, and the estimated "EZ1400" dies based on die area, but I ended up with messy and somewhat conflicting results (due to my limited amount of actual data, or lack of flux figures of the XP-G to use).
Please leave comments and inform me of errors, or if I am mistaken with the concept of treating large dies like multiple 1mm^2 dies. Thanks!
Cheers,
-Tony
EDIT: Sorry for such high res pics (after you keep clicking on them to zoom in). I am too used to DSL and Cable internet. Plus, sorry for the weird color shifting happening to the text in my pictures. Between Excel '07, PSP 8, and image shack, there is something going on with the letters and lines. Please inform me via PM if any "anti-aliasing" or softening features can be turned off in Excel. Thanks for your patience.
Last edited:
ICUDoc
Enlightened
Marduke thanks for your (gentlemanly) contributions to this thread. I like the choice of emitters- I think it shows your recollection of CPF pop-culture! I remember "T" bins were highly sought after, and the subsequent jumps you illustrate were all quantum leaps in CPF lumen-lust. Thanks again!
Last edited:
Marduke
Flashaholic
FYI, the efficacy charts above DO take Vf into account, even though the binning structure does not (directly). When you compute the curve for lm/W, the Vf is accounted for in that calculation.
Last edited:
manoloco
Enlightened
Are you saying that the Modoo triple cree XP-G drop-in waitlist will be for S2 ones!?!?... :kiss:
are you trying to get 700+ lumens regulated OTF from a single cell??? :popcorn:
are you trying to get 700+ lumens regulated OTF from a single cell??? :popcorn:
Last edited:
Yoda4561
Flashlight Enthusiast
I'm somehow drawn to the prospect of a 400 lumen fenix LOD sized monster.
Lucky i'm on the waiting list...moddoo, please use S2 if its avail anyway..
TorchBoy
Flashlight Enthusiast
Indeed, I enjoyed reading that very much, scrolling down one line at a time so I could savour every word before the graph was revealed. It's perfect; just what I wanted. Thank you! :twothumbs
When does the XR-G come out?
znomit
Enlightened
TorchBoy
Flashlight Enthusiast
I'm asking about the XR-G in the wrong thread, sorry.
Latest posts
-
Which discontinued flashlights do you wish were still made?- Latest: gottawearshades
-
-
-
-
-
-
-
