Low Vf Luxeons?

[markus_i]

Hi all,

(hope this is the best forum for my question...) since most of the modders/builders are tending towards direct drive and correspondingly white, high flux, high Vf Luxeons: what happens to the rest? Does ARC (e.g.) have a big bin of rejected emitters?
Basically, I'd like to know if there's a reliable way (apart form having to buy a full reel) to find low Vf Luxeons (white and yellow, 'G' at most, 'H' only if I can find 'E's to compensate) with almost no requiremets regarding colour and moderate requirements towards flux (of course, Q's would be nice, but I can live with anything above N for the whites and maybe Q for the yellows).

Bye
Markus

 

[JollyRoger]

Markus, I believe many modders actually *like* low Vf luxeons! I certainly do! They provide for a longer runtime when used, for example, in step-up circuits like the BadBoy....

 

[JSWrightOC]

The *ultimate* LED would have a very low Vf, yet a very high luminous efficacy. That would mean the emitter is more efficient at converting electron flow into pure photonic radiation. /ubbthreads/images/graemlins/crazy.gif

 

[hotbeam]

JSWrightOC... you mean a X3T? /ubbthreads/images/graemlins/grin.gif /ubbthreads/images/graemlins/grin.gif Sorry, couldn't help myself.

 

[Chop]

markus,
Unless I'm mistaken, luxeons of like bin come in reels. Q3J's for example. If you happened to get a reel and it was Q3J, they would all be Q3J's. They would of course be subject to the luxeon lottery.

Going with high Vf is useful when direct driving so that you can match the output of the batts to the Vf of the luxeon. When using a circuit of some sort, I always go with a low Vf for runtime, if possible.

 

[shiftd]

Ditto to what chop's saying.
High Vf lux great for DD, while the Low Vf is preferred when you are using some step up (boost) or step down (buck); they really help in getting the runtime up-ed more.

In general, one would prefer to get Vf as low as possible, Flux as high as possible (obviously) and color as white as they can get (i don't think i need to say more /ubbthreads/images/graemlins/tongue.gif)

 

[CM]

I trust the Vf bin rating as much as the color rating. I have four R2H's that I put into my BB biased at exactly (theoretically) 350mA. One LED has a Vf of 3.06V. The other three LED's have a Vf of 3.5V. An H is supposed to be 3.03V MIN to 3.27V MAX. That means out of my sample of 4, 75% is out of spec. Granted that's a small sample but statistically, I should expect more to be closer to the mean. The fact that 3 units are OUT OF SPEC means that their binning is highly suspicious. Incidentally, I tested Q3J's (12 samples) and they ALL measured about 3.25-3.35V. The mean was about 3.28V which is what I would expect. I trust my measurement method since I also measured a resistive load to ensure the BB was regulating to 350mA and it was within 5%. Would be interesting if everyone with an R2H can measure the Vf at the rated 350mA to see if they truly have an "H" forward voltage. BAH! Lumileds.

CM

 

[Chop]

CM is, of course, much more knowledgeable than I am in such matters; however, I would just like to say that I distrust R2H's as a whole. None of the R2H's that I've gotten my hands on (6 of them, from what I believe to be different sources) lived up to the expectations that I had built after reading all of the comments about them on CPF. A couple were white, a couple green and a couple yellow. None of them were as bright as any of the Q3J's that I compared them to. The only R flux that I've had luck with are R2K's, which have been the brightest 1 watters I've seen to date.

CM, have you tried your test on anything other than the R2H's and Q3J's. Your results with the Q3J's are not surprising since these things have been great and consistent, for me anyway. The R2H's have just left a bad taste in my mouth, so to speak.

 

[JSWrightOC]

On the topic of forward voltage drop (Vf), why is there so much variation across a specific type of junction? The white Luxeon emitters use a blue-emitting PN junction, which while having a higher Vf then a red-emitting junction, still varies by quite a bit. Can anyone shed some light (no pun intended /ubbthreads/images/graemlins/tongue.gif) on this?

 

[Doug S]

[ QUOTE ]
JSWrightOC said:
On the topic of forward voltage drop (Vf), why is there so much variation across a specific type of junction? The white Luxeon emitters use a blue-emitting PN junction, which while having a higher Vf then a red-emitting junction, still varies by quite a bit. Can anyone shed some light (no pun intended /ubbthreads/images/graemlins/tongue.gif) on this?

[/ QUOTE ]

This is a good question for which I too would like to know the answer. I have made some measurements that suggest that the differences are due to more than variations in the bulk conductivity of the semiconductor material involved. This seems to suggest that variations in the properties of the junction itself are involved.

 

[AilSnail]

cm, my two r2h's were around 3.7v with a bb400. three q4h ld's were around 3.15.

 

[CM]

AilSnail,

You really need to measure with 350mA because that's where Vf is measured. The BB400's are actually driving the LED at 416.7mA so a higher Vf will be expected. Also Lumiled's specs bias at 350mA MAX (like we all listen to that /ubbthreads/images/graemlins/grin.gif) Maybe if we can trust their binning, we wouldn't have to resort to overdriving the s**t out of these things to get what we expect. OK, OK, we would still overdrive the ****ens out of them regardless of the trustworthiness of their binning. I have 10 more Q3J's, this round from hotbeam. Will repeat the test and report back here time permitting /ubbthreads/images/graemlins/grin.gif

CM

 

[Chop]

Alright CM, /ubbthreads/images/graemlins/popcorn.gif /ubbthreads/images/graemlins/sleepy.gif /ubbthreads/images/graemlins/popcorn.gif /ubbthreads/images/graemlins/sleepy.gif

 

[Doug S]

[ QUOTE ]
CM said:
AilSnail,

You really need to measure with 350mA because that's where Vf is measured. CM

[/ QUOTE ]

To fully match the test conditions defined by Lumileds for binning, it is also necessary for the measurement to be performed at a *junction* temperature of 25C. This is a bit tricky to do with limited test equipment. A low duty cycle current pulse and a scope are probably the easiest method.

 

[koala]

It sounds rather unrealistic for Lumileds to be doing all the measurement at 25c, without external heatsinking, my R2H star will be heating up higher than 34c.

Placing it on a huge piece of heatsink cools down nicely but I don't think the junction temperature is as cool as 25c.

These Luxeons are not very cheap compared to 5mms could it be easily made by those people out there? I have seen some markings on the stars different from standard one to be a little less bold.

Vince.

 

[CM]

DougS

That's a very important point in which I failed to mention the grimy details. Here's how I did my test for those willing to endure the details. I actually mounted an array of 10 LED's onto a large slab of aluminum with white thermal grease at the interface. The slab provides great thermal mass to dissipate the heat. I found that without this mass to stabilize Tj, I can watch Vf change drastically over several seconds as the junction heated up. That's when I realized I needed to keep the junction temp stabilized. I don't think it was exactly at 25C, but my house is at 74F (yeah my cooling bill is high /ubbthreads/images/graemlins/grin.gif) which is probably close enough. The slab remained cool to the touch in the duration of the test. Vf was very stable, changing only 10-15mV max during each 3-5 second test. Can you tell I'm kinda anal? Guess if I'm really anal, I would compute theta JA with my slab and then use a peltier with a PID controller to set the slab temperature for a junction temp of 25C /ubbthreads/images/graemlins/grin.gif Hmmm, now that's an interesting thought.

nukiez, Lumileds test is not really supposed to be representative of *actual* use condition. Test conditions at a junction temp of 25C is actually pretty common with semiconductor manufacturers. It just establishes a reference test condition that can easily be replicated.

CM

 

[Doug S]

[ QUOTE ]
CM said:
DougS

Can you tell I'm kinda anal? CM

[/ QUOTE ]

CM, anal can be good. At least as it relates to lab technique. /ubbthreads/images/graemlins/grin.gif

 

[Doug Owen]

[ QUOTE ]
JSWrightOC said:
On the topic of forward voltage drop (Vf), why is there so much variation across a specific type of junction? The white Luxeon emitters use a blue-emitting PN junction, which while having a higher Vf then a red-emitting junction, still varies by quite a bit. Can anyone shed some light (no pun intended /ubbthreads/images/graemlins/tongue.gif) on this?

[/ QUOTE ]

There are a lot of 'drops' that go into Vf. Some are easy to understand (like the bulk resistances that contribute to Vf changes with current), others fixed (like the work function to get the photon out). The 2 mV per degree C drop is also fairly well understood by most.

The major variable, device to device, has to do with doping, Magnesium specifically, as I understand it:

How Luxeons are made

Doug Owen

 

[JSWrightOC]

/ubbthreads/images/graemlins/thumbsup.gif

(the link) Very good reading material. I knew some of what you said, but it's nice to read it again to refresh the memory, and the rest to learn more new stuff.

As for the "drop" I was referring to, I had silently assumed a few things (I know, that's wrong), such as junction temperature and forward current. Though none of us really trust LumiLED's voltage binning structure, we can safely assume that they had good intentions in creating it, so variations to a certain degree were to be expected.

If I understand correctly, we have ohmic loss in the bonding wire, the bonding connection, and the metallic distribution layer on top...while we have semiconductor "losses" (non-linear) in the quantum well, correct? I guess the most variation is in that top metallic layer, but your statement about the doping would suggest there is also quite a bit in the well structure.

Are these kinds of variations present in your average 5mm LED?

 

[tvodrd]

In total, I have had my hands on 16 R2H Stars. The first 4 were fairly low Vf based on brightness on a 123, and I didn't know what I had, and used them. Building a CR2 DD, I went through the rest with a fresh 123 and my ammeter. They have ranged from 80mA to abt 160 mA. I subsequently was given one that pulls 240mA /ubbthreads/images/graemlins/smile.gif. Test condition was power it long enough to read the meter, so I don't think Tj got into the act much. No where as anal as CM, but it seems to work. (BTW McGismo told me I was anal. /ubbthreads/images/graemlins/smile.gif )

Larry