Technical evaluation - Cree XR-E Q4 bin

evan9162

Flashlight Enthusiast
Joined
Apr 18, 2002
Messages
2,639
Location
Boise, ID
Technical evaluation - Cree XR-E Q4 bin

The Cree XR-E has been out in the market for almost a year now. It was a bit of a revolution when it came out, even at the 70-ish lm/W level (which is considered "low" these days), it was far above and beyond any offering prior in terms of efficiency; and easily became the brightest single-junction LED and essentially single handedly doubled LED efficiency.

I figured I'd produce a report just as I've done with the 100 lumen rebel and improved K2 (yet unreleased). While I may not be presenting new or amazing information, it'd still be nice to have this all in one place in the LED forum.

The partcular XR-E I'm testing is a Q4 bin, which is supposed to produce 100-107 lumens at 350mA. It's not the absolute brightest XR-E available (the Q5 bin gets that honor), but seems to be easily widely available, and I didn't particularly feel like going out of my way or spending tons of extra for the Q5 bin. Perhaps in the not to distant future, I'll obtain one and present test data on it as well.


Just in case you haven't been around here for the last year, here is what a Cree XR-E looks like:

xre_q4-1.jpg


This is a 4 bond wire version of the XR-E. The P4 I've tested prior only had 3 bond wires. The extra bond wire should help reduce the Vf at higher currents. While extra bond wires do help, there's only so much that packaging can do for the Vf of a device. Ultimately, the dominant factor in the Vf is the LED chip its self, and there is still variation among individual samples. So having extra bond wires still doesn't ensure a low Vf.

The optical system of the XR-E is interesting - there's a thick silver ring, with the optical dome (made of glass). The LED chip sits quite deep down relative to the glass dome, indicating that the beam pattern will be relatively narrow and forward projecting, compared to other power LEDs that we've become accustom to that have nice, wide, lambertian distributions. This could pose a problem for certian reflector configurations.

Here's the backside of the XR-E:

xre_q4-2.jpg


The XR-E is intended to be reflow soldered to a PCB. Because of this, there are power contacts on the bottom of the package, connected to the power contacts on top. This can present a problem if you want to directly mount the XR-E to bare metal for a heatsink, as the heat sink could short the power pads together. The top and bottom contacts are connected with vias in the corners of the package substrate. There are quite a few solutions that are possible to address the shorting issue: clipping the corners of the package, grinding down the bottom contacts, using a thin insulating layer of thermal epoxy, or raising the central portion of the heat sink to isolate the power contacts. I chose to use my dremel and a sanding disk to sand down the contacts in the corners where the vias are present. To me, this is the least destructive solution that allows the simplest mounting, as I can use as thin of a layer of thermal compound/epoxy as possible, and not worry if part of the package makes physical contact with the heat sink.

The central pad is the thermal pad, indended for heat transfer from the package to the heat sink. The thermal pad is fully isolated!!! YAY!! When using multiple XR-E's in various electrical configurations, there is no need to electrically isolate the thermal pads on XR-Es, simplifying mouting and usage.


Here's the XR-E amongst similar power LEDs: the updated K2, and Rebel-100:

xre_q4-3.jpg



The XR-E is about the same size as the K2 (and original Luxeon, Luxeon III, and Luxeon V). The Rebel is substantially smaller (and much more difficult to deal with).


The beam profile of the XR-E is substantially different than other power LEDs. It is significantly more narrow and forward projecting than other power LEDs, and projects much less out the sides.

Here is the XR-E's beam profile, compared to the Rebel-100:

xre_q4-4.png


The relative intensity is a good way to compare different beam profiles, but it doesn't tell the whole story. Another way is to compare the absolute beam intensity at a given current, for two different LEDs with the same brightness. This will show which part of the beam each LED delivers its lumens to, and allows you to estimate how a given optical system will behave with the given beam distribution.

Here's the beam profile for the rebel and XR-E, at 310mA. This is the actual beam brightess in lux, not scaled relative to 0 degrees, like in the previous graph:

xre_q4-5.png


In the central 35 degrees, the XR-E delivers more lumens, while in the part of the beam outside 35 degrees, the rebel delivers more lumens. Thus, the rebel could create a brighter focused beam with a reflector, but the XR-E would produce a brighter beam with optics (such as an aspherical lens). The lack of side-emitted light makes the XR-E less than ideal for reflectors. An especially deep reflector would be needed to be optimal with the XR-E.

One must wonder how the beam profile and total output would change if the thick silver ring were absent, and a shorter silicone dome were in the place of the ring/glass dome combination. The output could be significantly higher if the side emitted light were permitted to escape directly. This might be one reason why the Seoul P4 was hitting near 100 lumens last year when the XR-E was at around 80ish, when they both used the same blue die underneath. Granted, there are phosphor differences between the two, but I have a feeling the optics of the packages also have an effect on that.


Here's the raw performance data for this Q4 XR-E.

Code:
Current (mA)	0.1	30	130	310	670	980	1260	1570	1930
------------------------------------------------------------------------------------
Vf		2.53	2.82	3.15	3.37	3.62	3.75	3.86	3.96	4.06
Watts				0.41	1.04	2.43	3.68	4.86	6.22	7.84
Lumens				42	94	178	234	278	312	341
Lumens/W			103	90	73	64	57	50	44



The extra bond wire in the Q4 XR-E's construction should help lower the Vf. This isn't a guarantee, since the Vf is mostly dependent on the individual LED die. Let's see how the Vf of this particular XR-E stacks up:


xre_q4-6.png



This particular XR-E actually has a rather high Vf. This will hurt it in the efficiency department. This doesn't mean that all XR-Es will have such a high Vf, but this particular one does. Even though it is the 4 bond wire version, it still has a higher Vf across all current levels than even my 3 bond wire P4 XR-E. So, when it comes to Vf, your milage may (and will) vary.


Here is how the Q4 stacks up to other power LEDs with overall brightness:

xre_q4-7.png


The Q4 XR-E (100-107 lumens at 350mA) performs almost identically to the Rebel-100 (100-120 lumens at 350mA). In fact, it starts to peform a bit better than the Rebel at very high current levels. The better packaging of the XR-E allows for a better thermal path, and makes heat sinking much easier. The Q5 XR-E should perform even better.


Here's efficiency/efficacy:

xre_q4-8.png


The high Vf of this XR-E hurts it efficiency wise. With a lower Vf, the XR-E will be much more efficient. The efficiency of this XR-E could be about 10% higher if it had the same Vf of the P4 XR-E.


Finally, the infamous "droopyness" graph - an indication of how efficiency scales with drive current:

xre_q4-9.png


The Q4 XR-E scales about the same as all other current power LEDs.


So the Q4 XR-E has about the same performance specs as the Rebel-100, and performs pretty much identically as well. Ultimately the choice as to which one to use depends on a lot of factors. As far as ease of use and mouting, the XR-E wins hands down - it's much easier to deal with than the rebel, even with the bottom electrical contacts - it's also much easier to get good thermal transfer to the heat sink with the XR-E as well. The only gotcha would be with the beam profile, which is much different than other power LEDs that have a more lambertian beam distribution. This could affect performance with certian optical configurations.

So which one to use? Don't ask me! :p I'm still up in the air - with power LEDs coming in different packages, with different levels of performance, and varied aspects of beam profiles, electrical configurations, etc, there are more choices now than ever in the power LED market, but it's less and less clear what the best "go-to" power LED is for general applications.

Given that, the Q4 XR-E is a great performer, and should definitely be seriously considered for any application.
 
Last edited:

Bullzeyebill

Flashaholic
Joined
Feb 21, 2003
Messages
12,164
Location
CA
evan9162 , thanks for the nice thorough eval. I bookmarked it. Neat you are doing this. Sort of what Newbie was doing and important stuff that can help us make decisions for different applications.

Have you looked at recent Seoul P4's developpment to see if their has been some improvement in heat management. Many modders jumped on the Seoul P4 bandwagon and are running them at 1+ amps. I do like the Seoul P4's low vf (H).

Bill
 

evan9162

Flashlight Enthusiast
Joined
Apr 18, 2002
Messages
2,639
Location
Boise, ID
Have you looked at recent Seoul P4's developpment to see if their has been some improvement in heat management. Many modders jumped on the Seoul P4 bandwagon and are running them at 1+ amps. I do like the Seoul P4's low vf (H).

Bill

Vf bin doesn't really tell you very well what the Vf of the device will actually be. Seoul's H Vf bin (3.00-3.25V) is about the same as lumiled's H Vf bin (3.03-3.27V). That's a huge range. Emitters with a Vf of 3.25V at 350mA aren't particularly low, but still considered in the "H" bin. In fact, the Seoul P4 that I have was binned in the H Vf range, but it's on the top end of that bin. So while it appears to have a "good" Vf based on the bin, it is, in fact, only average.

Interestingly, Cree doesn't even bin Vf at all. So you don't even know what range they will be in.

The "new" V-bin Seoul P4 has an output range of 118-154 lumens, again a huge range. Without parts in a higher bin, we must expect that the parts will tend to cluster at the low end of the bin range.

Given the ratio of performance between the XR-E and Seoul P4, I'd expect the new Seoul to be performing in the low 120s initially. The P4's beam profile is much more like Luxeons and other nicely lambertian distributions, but there's still that pesky positively-connected heat sink slug to worry about.
 
Last edited:

mahoney

Enlightened
Joined
Jan 7, 2002
Messages
603
Thanks for all your work on this.

FWIW, I just got a few Q5 Crees., It's a tiny sample size, and I use a crude testing method, but compared to the 8 or so P4s I've had, the Vf for the Q5s seems higher. Direct drive from a 3 volt battery pack, the P4s would pass 50 to 80 ma, the Q5s were all between 33 and 37 ma
 

VidPro

Flashlight Enthusiast
Joined
Apr 7, 2004
Messages
5,441
Location
Lost In Space
wow Evan, you are are a meticulous writer/illustrator in the post, i understood almost all of that . oh my what is happening to me :eek:oo:
 

evan9162

Flashlight Enthusiast
Joined
Apr 18, 2002
Messages
2,639
Location
Boise, ID
When I get around to obtaining one. I only really got this Q4 because I needed thermal epoxy, and the sandwich shoppe happened to have Q4 Crees in stock.

Right now, I have a parts drawer with over 60 various power LEDs, mostly luxeons. I really should stop collecting LEDs and start using them.
 

1 what

Enlightened
Joined
Jul 6, 2007
Messages
617
Location
Australia
Loved the write-up and photos. Thanks.
One question: :confused: how obvious is the connection between the top and bottom electrical contacts?
I'm just about to receive some Q5's and want to direct mount to a heatsink. Is it obvious how much of the corners to remove ? do you need to remove all 4 corners or just 2 of them ? Do you have a photo of a "doctored" Q4 ?
Thanks again.
 

evan9162

Flashlight Enthusiast
Joined
Apr 18, 2002
Messages
2,639
Location
Boise, ID
My photo above is of a "fixed" XR-E. I've removed the minimum amount of the bottom contacts that need to be removed in order to ensure that the bottom contacts are isolated.

There is one electrical via in each corner. On the bottom side, you must grind down the copper over each of the vias until you reach the ceramic substrate. Then you will have removed the electrical contact between the top and bottom power pads.

After you've ground things down, check that the bottom pads are isolated. Use a DMM in either continuity check, or resistance measurement mode. Place one contact on the bottom power pad, and another on the corresponding top power pad. If there is continuity, or your meter reads around 0 ohms, then you must grind down a bit more material. You only need to remove material on each of the corners the bottom of the package, since those are where the electrical vias are. I would leave the top contacts alone, since that's what you'll be soldering to, and they are small enough already.

[edit]

Here are close-ups of the 4 ground away areas of the bottom power pads, showing the vias underneath (starting at the top left corner from the original image, going clockwise):

xre_via1.jpg


xre_via2.jpg


xre_via3.jpg


xre_via4.jpg
 
Last edited:

45/70

Flashlight Enthusiast
Joined
Oct 9, 2005
Messages
2,800
Location
Rural Ohio
Vf bin doesn't really tell you very well what the Vf of the device will actually be. Seoul's H Vf bin (3.00-3.25V) is about the same as lumiled's H Vf bin (3.03-3.27V). That's a huge range. Emitters with a Vf of 3.25V at 350mA aren't particularly low, but still considered in the "H" bin. In fact, the Seoul P4 that I have was binned in the H Vf range, but it's on the top end of that bin. So while it appears to have a "good" Vf based on the bin, it is, in fact, only average.
A question I've been meaning to ask, aren't SSC's Vf binned at 350mA, and Luxeon's at 700mA? If so, there is little comparison between an SSC H bin and a Lux H bin. Then again, maybe I'm wrong about that. :)

Dave
 

chimo

Flashlight Enthusiast
Joined
Sep 16, 2004
Messages
1,905
Location
Ottawa, Canada
Thanks for posting your data! I know that takes a whole lot of time and effort. :goodjob::thanks:

Paul
 

evan9162

Flashlight Enthusiast
Joined
Apr 18, 2002
Messages
2,639
Location
Boise, ID
A question I've been meaning to ask, aren't SSC's Vf binned at 350mA, and Luxeon's at 700mA? If so, there is little comparison between an SSC H bin and a Lux H bin. Then again, maybe I'm wrong about that. :)

Dave


It depends on the model.

regular Luxeons are binned at 350mA.
Luxeon III and V are binned at 700mA.
Luxeon Rebels are (currently) binned at 350mA.
Luxeon K2 Pxx2 parts are binned at 350mA.
Luxeon K2 Pxx4 parts are binned at 1000mA.

Seoul P4 parts are binned at 350mA.
Cree XR-E parts are binned at 350mA.
 

1 what

Enlightened
Joined
Jul 6, 2007
Messages
617
Location
Australia
A big thank you Evan 9162!
Like everything else in life it's so bleeding obvious when you know the answer.
The photos of bottom contact modification were great and I now see that it was obvious in your original photos only I didn't realize what I was looking at.
I've only recently discovered CPF but it impresses me how helpful many people are on this site. Thanks again.
 

45/70

Flashlight Enthusiast
Joined
Oct 9, 2005
Messages
2,800
Location
Rural Ohio
It depends on the model.
I was thinking only of Lux III applications, as that's what I'm usually replacing. So, in effect a Lux III H bin is actually a considerably lower Vf than a SSC H bin. I've seen this, of course, in the real world. I just wanted to confirm it.

Thanks evan9162. :thumbsup:

Dave
 

Bullzeyebill

Flashaholic
Joined
Feb 21, 2003
Messages
12,164
Location
CA
I think that the vf comparison that evan9162 made is the difference in H of a lumileds luxeon and a Seoul, both being run at 350mA's to led. They are very close, the Seoul ranging somewhat higher. This would pertain to a lux I, lux III, or Seoul P4 run at 350mA's to the led.

Bill
 
Top