Luxeon dynamic resistance

[evan9162]

After doing some measuring, and crunching some numbers tonight, I understand why it's called "dynamic" resistance /ubbthreads/images/graemlins/tongue.gif

I measured the Vf of 3 white Q2Js (Vf = 3.27-3.51V) across multiple current levels, and two 5W emitters at the same current levels (one white, one RB). The white has a Vf bin of T (6.31-6.87V), and I think the RB was binned U (6.87-7.35V).

Here is a graph of Vf vs Current:

This doesn't reveal anything spectacular, however, let's take a look at calculated dynamic resistance based on these values.

Now, to calculate dynamic resistance, I used delta(Vf)/delta(If) = R. I believe this is the correct procedure for calculating resistance (and I applied this method to measurements taken for current vs. voltage across a resistor, and got correct results)

Now, Lumileds specifys dynamic resistance for both 1W and 5W Luxeons as 1 ohm. This certianly is an average at rated current (350mA for 1W, 700mA for 5W).

This graph shows us a few things:

-The dynamic resistance varies depending on current. It seems to be quite high at low current levels, and very low at high current levels. This may explain why the Luxeon III has a dynamic resistance specified at 0.8 ohms, whereas the 1W and 5W all specify 1 ohm.

-The dynamic resistance of 5W luxeons tends to be higher than 1Ws (I realize I do have a relatively small sample size here, but it does appear as an emerging trend from this data)

-A lower Vf and lower dynamic resistance seem to go hand-in-hand. I only have one Vf bin of 1W, and only one sample of each 5W, and their Vfs are close to each other, and close to the border values for the two Vf bins, so I don't know how much I can realistically extrapolate this to other Vf bins. Anyone care to plot some data on some H, K, L binned 1Ws, or some T, U, V, W binned 5Ws?

-A lower dynamic resistance will mean that the Vf will increase less with increasing current. If we had 2 otherwise equal emitters with equal Vf, but one with higher dynamic resistance, at higher currents the emitter with lower dynamic resistance will both use less power, AND will run at a lower temperature (since Vf will be lower at high currents, dissipated power will be lower, thus junction temperature will be lower).

-As a result of the above, the emitter with the lower dynamic resistance will also produce more light at a given current level. Thus, an R-binned emitter could possibly be outperformed by a Q-binned emitter at higher current levels.

-The dynamic resistance may (and probably does) vary depending on heat. The tests I did were all about 1-2 seconds, so temperature shouldn't be a factor. Longer testing may result in different values for the dynamic resistance.

-I just did a lot of typing /ubbthreads/images/graemlins/tongue.gif

Thoughts?

 

[DrJ]

.

 

[evan9162]

That effect is well known. The Vf of LEDs drops as they heat up. The downside to this is that you can fall into a thermal runaway scenario, where the Vf keeps dropping, allowing more current, heating up the junction, dropping the Vf, allowing more current, etc...until the LED fails from overcurrent (bond wires), or excessive heat (semiconductor failure).

I'm not sure if this is a result of the dynamic resistance changing, or if the heat would cause the Vf to drop equally across all current levels.


[edit]

My ISP's hosting server has been bouncing a lot tonight. You may get red Xs for the graphs, but I know that the URLs are correct.

 

[Doug S]

[ QUOTE ]
evan9162 said:
-The dynamic resistance of 5W luxeons tends to be higher than 1Ws (I realize I do have a relatively small sample size here, but it does appear as an emerging trend from this data)

Thoughts?


[/ QUOTE ]
Nice work.
I think if the 5W and 1W are compared on the basis of equal current per die, e.g., 350mA for 1W compared to 700mA for 5W, the dynamic resistances are similar, subject to sample to sample variation.
Awhile back, I looked at the variation among samples using the dynamic resistance defined by the slope connecting the 90 and 400mA points. The was considerable variation. Results were posted here:
dynamic resistance

 

[evan9162]

[ QUOTE ]
Doug S said:
[ QUOTE ]
evan9162 said:
-The dynamic resistance of 5W luxeons tends to be higher than 1Ws (I realize I do have a relatively small sample size here, but it does appear as an emerging trend from this data)

Thoughts?


[/ QUOTE ]
Nice work.
I think if the 5W and 1W are compared on the basis of equal current per die, e.g., 350mA for 1W compared to 700mA for 5W, the dynamic resistances are similar, subject to sample to sample variation.
Awhile back, I looked at the variation among samples using the dynamic resistance defined by the slope connecting the 90 and 400mA points. The was considerable variation. Results were posted here:
dynamic resistance



[/ QUOTE ]

/me slaps head

Of course, I don't know why I didn't think that through. I normalized along emitter current, I probably should normalize along individual die current.

It looks like I mirrored some work you did a year ago. I wonder if I'm just a year behind, or if the need to investigate random aspects of lights hits someone once a year. Looks like it was my turn /ubbthreads/images/graemlins/tongue.gif

Here's an updated graph, normalized to current per die:

Yep, the 5W fits right into the same curve that the 1W's do.