Luxeon Dynamic Resistance Data

[Doug S]

Thanks to Wayne J I have a few more 1W Luxeons on my workbench. I have a total of 7 Luxeon InGaN LEDs that are not permanently in use somewhere. While not a large sample, I decided nonetheless to look at the range of dynamic resistance that they had. Dynamic resistance is a simplified [linearized] model of the IV characteristic applied over a small range of I or V. The Luxeon datasheets list a typical of 1 ohm for the InGaN LEDs [green, cyan, blue, royal blue, and white] but do not specify a min/max range. You need two points to draw a line. I choose test currents of 90 and 400 mA. Voltage measurements were made immediately upon application of current while the die was still close to room temperature. Dynamic resistance, Rd, is calculated as deltaV/deltaI.
Here are the results;

Sample#**Vf90**Vf400**Rd
***1*****3.41**3.82***1.32
***2*****2.99**3.30***1.00
***3*****3.15**3.52***1.19
***4*****2.71**3.21***1.61
***5*****2.74**3.27***1.71
***6*****2.98**3.51***1.71
***7*****2.86**3.11***0.81

Samples 1-3 are recent production whites
Samples 4&5 are recent production cyans
Sample 6 is an early production RB
Sample 7 is an early production white
Other observation, bin codes were known for samples 1-5. Samples 1 and 4 would have had Vf at 350mA outside of their marked code for Vf.

It is interesting that even with this small sample size, over a 2:1 range of dynamic resistance is measured. It is likely that a larger sample would have shown a greater range.
What is the significance of dynamic resistance anyway, you may be asking? It is probably most important to those who are doing mods involving resistive current limiting or direct drive. Direct drive is a special case of resistive current limiting where the only resistance used to limit current is the sum of the battery internal resistance and the LED dynamic resistance. Here is how not accounting for the variability of dynamic resistance can get you in trouble. Say that you build a direct drive mod with a given type of battery. You know the Vf bin code for the LED you use. You run your mod for several sets of batteries and no magic smoke is released. This meets your standards for reliability. You replicate your mod using another LED with the same Vf bin code figuring that if it worked OK once it should again. If it turns out that the first LED had a relatively high dynamic resistance and the second a relatively low dynamic resistance, even though they both had identical Vf bin codes, the second will draw more, possibly much more, current and the magic smoke may show itself.

 

[McGizmo]

Doug, Doug, Doug, dang you have it all wrong...........

Yup, just kidding. Thanks for doing that test and the explaination! Very informative, to say the least. These Luxeon's really do have their own and unique characters!

Thanks again for the info.

- Don

 

[INRETECH]

The proper way to test these would be to use a piece of test equipment known as a Curve-Tracer

It is simular to a scope; it makes almost an infinite number of measurements instead of just the two you did

The X-axis is voltage, Y-axis is current

A resistor would have a diag. line, since it is a linear device; but semiconductors are not

I once saw how you could easily build your own curve tracer using a scope set on X/Y mode, a transformer and a couple of resistors

One of the most power curve tracers made by Tek was powerfull enough to curve (and cook) Hotdogs

http://www.vaxxine.com/phil/scopes/arnoud/575.jpg

 

[Slick]

Thanks for taking the time to test & post Doug.

This pretty much sums up the frustrations of any developer who's working with Luxeons..

 

[Doug S]

Originally posted by Slick:
This pretty much sums up the frustrations of any developer who's working with Luxeons

<font size="2" face="Verdana, Arial">I take a more positive view. The more you understand about the characteristics of the devices you are working with, the fewer the needless casualties on the path to where you are going.

 

[Vikas Sontakke]

Hi Doug,

Would you be able to hook up a lux-meter or something similar to measure the light output under those conditions? That would give very interesting data points.

It may also show that two identical lights based on same batch of Luxeon would give extremely large variation in the actual light output under identical batteries :-(

This would prove once and for all that any relative comparisons of two different Luxeon based lights would be strictly a draw of luck.

So the next time somebody asks a question as to which is the brightest among KL1, LS and Lambda Illuminator, we can answer

- Vikas

 

[Vikas Sontakke]

Hi Doug,

Would you be able to hook up a lux-meter or something similar to measure the light output under those conditions? That would give very interesting data points.

It may also show that two identical lights based on the same batch of Luxeon would give extremely large variation in the actual light output when driven identically (i.e. using similar batteries)

This would prove once and for all that any relative comparisons of two different Luxeon based lights would be strictly a draw of luck.

So the next time somebody asks a question as to which is the brightest among KL1, LS and Lambda Illuminator, we can tell,

"Use the outcome of a 3-sided coin to
get your answer"

- Vikas

 

[Slick]

Originally posted by Doug S:
</font><blockquote><font size="1" face="Verdana, Arial">quote:</font><hr /><font size="2" face="Verdana, Arial">Originally posted by Slick:
This pretty much sums up the frustrations of any developer who's working with Luxeons

<font size="2" face="Verdana, Arial">I take a more positive view. The more you understand about the characteristics of the devices you are working with, the fewer the needless casualties on the path to where you are going.</font><hr /></blockquote><font size="2" face="Verdana, Arial">I make that statement from the perspective of having spent considerable time optimising a design towards use of a certain "Luxeon binning", then finding it absolutely IMPOSSIBLE to specify future orders from a "like" binning...

That means back to the drawing board each time..

I've had requests to supply a continuous stream of product supply, only to have to refuse due to the fact that I cannot specifiy luxeon binning. I just tell them they'll have to wait for Lumiled's to get their act together. Of course, this also give me more time to work on personal projects.

 

[Doug S]

Originally posted by Vikas Sontakke:
Hi Doug,

Would you be able to hook up a lux-meter or something similar to measure the light output under those conditions? That would give very interesting data points.

- Vikas

<font size="2" face="Verdana, Arial">That wasn't the point of my testing. The fact that output is variable among samples is well known.

Originally posted by Vikas Sontakke:

It may also show that two identical lights based on same batch of Luxeon would give extremely large variation in the actual light output under identical batteries :-(

- Vikas

<font size="2" face="Verdana, Arial">A casual reading of the datasheets would lead to the same conclusion. Practical experience of CPFers bear this out. Checkout some of the CMG Reactor threads.

 

[Kirk]

Doug S,
Any idea why so much variation? Is there this much range in regular old Nichia LEDs? What makes Luxeons so darn weird? Are these hand built? Are the chips way different in size?
Kirk

 

[Doug S]

Originally posted by Kirk:
Doug S,
Any idea why so much variation? Is there this much range in regular old Nichia LEDs? What makes Luxeons so darn weird? Are these hand built? Are the chips way different in size?
Kirk

<font size="2" face="Verdana, Arial">Not hand built. While this may seem like a large variation, compared to some semiconductor parameters it is not large at all, e.g., transistor betas or op amp input bias currents [which vary thru orders of magnitude]. The Nichias that I have checked seem to be a bit less variable, typically are in the range of 20 to 30 ohms. Their dynamic resistance is much higher due to the small die size.

 

[Kirk]

Doug S,
Thanks for the info. Do you see them becoming more uniform--enough for users to do regular production runs without running into problems--in the near future or is this an item that is inherently non-uniform and has to be tested every single time? Man, if they are inherently non-uniform that will put a major damper on the big manufacturers using them. I assume the "Mag-Lites" of the world want to be able to buy a component, stick it in their product, and not have to worry about it. Even Nichia is still having problems with uniformity. I assume they don't start out to make "yellow" or "green" white LEDs, but they end up with a large? percentage that don't meet specs. And they've made how many bazillion white LEDs?
Kirk

 

[Doug S]

Originally posted by Kirk:
Doug S,
Thanks for the info. Do you see them becoming more uniform--

<font size="2" face="Verdana, Arial">I have no insight/knowledge to answer this question.

Originally posted by Kirk:

enough for users to do regular production runs without running into problems--in the near future or is this an item that is inherently non-uniform and has to be tested every single time? Man, if they are inherently non-uniform that will put a major damper on the big manufacturers using them. I assume the "Mag-Lites" of the world want to be able to buy a component, stick it in their product, and not have to worry about it.

<font size="2" face="Verdana, Arial">I do not agree with the premise of your above questions. LED manufacturers do not intend for end users to treat LEDs as constant voltage devices. They all advocate controlling current. There are various viable ways of accomodating the variability of the dynanic resistance parameter without resorting to individual sorting. It is true that many commercial portable LED lighting products ignore what I and the LED manufacturers would consider good engineering practices but that doesn't mean that it is not possible for them to do it right.

Originally posted by Kirk:

Even Nichia is still having problems with uniformity. I assume they don't start out to make "yellow" or "green" white LEDs, but they end up with a large? percentage that don't meet specs. And they've made how many bazillion white LEDs?
Kirk

<font size="2" face="Verdana, Arial">I have no knowledge of what percentage of their production fails to meet the specs of one of their four color bins. My personnal experience with their white product has been very good.

 

[MicroE]

Originally posted by INRETECH:
it makes almost an infinite number of measurements

<font size="2" face="Verdana, Arial">Inretech---Is "almost an infinite number" Infinity minus one or Infinity minus two?
Maybe Infinity divided by two?
It's been a lot of years since I operated a curve tracer. But, I do seem to recall that it had an "almost infinite number" of knobs....Marc

 

[Charles]

Hello, Doug

Ofcourse this is off subject, but my light peformed well for seven hours run time under
adverse conditions, but the cave is spectacular
and no end in site. 3500ft. H. 350ft. V. so far.

 

[Kirk]

Doug S,
I stand corrected. I learn more stuff here than I thought I ever would. Thank you and the others for bringing the technical side of lighting to "light".
Kirk

 

[Doug S]

Originally posted by Kirk:
Doug S,
I stand corrected. I learn more stuff here than I thought I ever would. Thank you and the others for bringing the technical side of lighting to "light".
Kirk

<font size="2" face="Verdana, Arial">You are welcome. I'm happy you find it of interest. There is a very wide variation here in level of interest [or disinterest] in the technical side of stuff. While I often imagine all the eyes glazing over when I start to type, I don't believe that one has to be a rocket scientist to appreciate the technical side[though optics, solid state physics, or thermodynamic scientist might be helpful

]

 

[MicroE]

Doug---I read your (and others) technical posts. My eyes do not "glaze over". Keep typing.---Marc

 

[Vikas Sontakke]

For a given bin of Luxeons, would they produce the same light for same current?

Is yield on LED (including Luxeon) similar to other semicondcutor products such as CPU's? Is it standard bell curve? Don't the semi manufacturer do the sorting after actually testing them at different speed and grade them?

- Vikas

 

[Doug S]

Originally posted by Vikas Sontakke:
For a given bin of Luxeons, would they produce the same light for same current?
- Vikas

<font size="2" face="Verdana, Arial">Under the same conditions as the binning test, output would be the "same" subject to the upper and lower limits of the bin. At other conditions, this may not be true. I have noted that the relative change in output as current is varied is different from one sample to another.

Originally posted by Vikas Sontakke:

Is yield on LED (including Luxeon) similar to other semicondcutor products such as CPU's? Is it standard bell curve? Don't the semi manufacturer do the sorting after actually testing them at different speed and grade them?

- Vikas

<font size="2" face="Verdana, Arial">I don't know.