Current through individual Dies on MC-E / P7

[marcopolo]

If individual dies/emitties have slightly different Vf's then how do each die (in a paralell arrangement) get exactly the same current?

Take for example the P7. Rated to 2.8A: The total current available is 2.8A but if each die has a slightly different Vf then they might take:
0.7A, 0.7A, 0.6A, 0.8A for example or a whole host of variations. Is this correect?

The MC-E might be even worse if it is based on the R2 as a recent post showed that the Vf at any given current can vary quite substantially at 3.5v from 0.2A to 0.8A.

How is this addressed with a high power driver that supplies the same Vf to 4 Series LED's.?

Marco.

 

[2xTrinity]

If individual dies/emitties have slightly different Vf's then how do each die (in a paralell arrangement) get exactly the same current?

Take for example the P7. Rated to 2.8A: The total current available is 2.8A but if each die has a slightly different Vf then they might take:
0.7A, 0.7A, 0.6A, 0.8A for example or a whole host of variations. Is this correect?

The MC-E might be even worse if it is based on the R2 as a recent post showed that the Vf at any given current can vary quite substantially at 3.5v from 0.2A to 0.8A.

Severl members myself included have tested these LEDs by turning thme on and either projecting them onto walls, or looking at them through dark filters. The dice have no discernable difference in illumination from each other in either the P7 or the MC-E when driven in parallel. This suggests there is no appreciable current difference between the respective dice.

Nonetheless, the biggest causes for the effect of different spearate LEDs have different Vfs are:

1) Difference in the raw materials from one batch to another on an assembly line
2) Difference in temperature between separate LED devices in use

In the case of the quad-die emitters, all the dice will be manufactured at the same time, so effect (1) will be mitigated. My guess is that Cree or SSC may even pre-bin the separate dice for Vf before assembling the packages as well.

Also, since the 4 LEDs in both emitters are on a common substrate, all the dice will be the same temperature as each other at all times, taking care of (2). Temperature differences is the biggest idea why running separate LEDs in parallel with each other is a bad idea, but it's less of a problem for quad die LEDs thermally connected together.

How is this addressed with a high power driver that supplies the same Vf to 4 Series LED's.?

Marco.

Components in series will by definition all receive the same current through them no matter what. That is the advantage of series wiring. You can't make a driver that "applies the same Vf to each LED in series". You can however set the current of the string, and each component will drop whatever voltage it happens to drop.

In parallel, in the case of the MC-E, it may be wise to use a separate driver or ballast resistor in series with EACH die, for some sort of applicaiton that will see many thousands of hours of lot of runtime -- such as a fixed lighting applicaiton. That is not possible with the P7. Over the lifespan of any flashlight however this sholdnt' matter at all.

 

[Gryloc]

I believe that the Vf matching is pretty good, especially since you can power a SSC P7 at low currents and the die brightness of each are similar. Some day, when I get my MC-E emitters from Cutter, and after they get lumen tested, then I will run a quick experiment out of curiosity.

I will wire the MC-E in series, and power it at various currents from 10mA to 1000mA. At each current level, I will check with my voltage meter the Vf of each die after the temeratures have stabilized. Actually, where could I source a very cheap, but semi-accurate voltage meter? I would like to buy four that can measure down to tenth or thousandths of a volt. I could hook all four to the MC-E (one per die) and monitor all of the forward voltages simultaneously. I could probably capture all this on video, then review it later. Then I can graph and compare the forward voltages.

Maybe someone can beat me to it, because I am curious on how well they are matched even if I do not have any MC-E emitters. I still worry that when the dies are wired in series, then dies with significantly higher forward voltages will run hotter. Therefore, one of the dies may appear dimmer. In the long term, could there issues if one of the dies on the substrate runs hotter? Should I even worry about this? Thanks.

-Tony

 

[saabluster]

Severl members myself included have tested these LEDs by turning thme on and either projecting them onto walls, or looking at them through dark filters. The dice have no discernable difference in illumination from each other in either the P7 or the MC-E when driven in parallel. This suggests there is no appreciable current difference between the respective dice.

I have had a different experience. So far I have used 4 MC-Es. In two the dice all came on at the same time/current but one die in both was of a significantly different color(more purple). One of them was perfectly fine in both color and Vf and one had consistent color but one die was way brighter and turned on sooner than the others.

 

[tebore]

I have had a different experience. So far I have used 4 MC-Es. In two the dice all came on at the same time/current but one die in both was of a significantly different color(more purple). One of them was perfectly fine in both color and Vf and one had consistent color but one die was way brighter and turned on sooner than the others.

I'm in the same boat but with P7s. I've had 2 where the dies were 99% matched and 1 where two dies are dimmer than the rest and another that had 4 brightness levels.

Maybe someone can beat me to it, because I am curious on how well they are matched even if I do not have any MC-E emitters. I still worry that when the dies are wired in series, then dies with significantly higher forward voltages will run hotter. Therefore, one of the dies may appear dimmer. In the long term, could there issues if one of the dies on the substrate runs hotter? Should I even worry about this? Thanks.

Why does that all sound backwards.

if the dies are wired in series you shouldn't have a problem because all the dies are seeing the same current. Think water pipe, the restriction is the smallest part. You can't flow more than that smallest part allows even with huge pipes before and after it.

If it's parallel then you have to worry about the die with the lower VF getting more current. And as it heats up it'll get brighter because the vf will lower (thermal run away).

As much as we liked to think of unmatched dies getting 0.7A, 0.7A, 0.6A, 0.8A. It's more extreme. It's more like 0.1 0.1 0.4 1.2, getting worse as it heats up.

 

[Gryloc]

tebore, my comment does sound a bit wordy (I understand). No, I still meant what I said (yes in series). However, I gave it more thought, and there isn't a problem. I was so used to parallel connected dies, and series-parallel connected dies in LEDs, that I gave it too much thought when thinking about useing an emitter that can be wired in series.

I was thinking that even though each die seen the same current (being in series), the higher Vf of one die versus the rest would mean that the one die would be dissipating more heat (being less efficient). However, I just did the math and the difference in power draw, even with what I thought was a wide range of Vfs at a current, was too small to worry about. Lets say that the MC-E is operating at 700mA (~9.5W). Even though the four dies are on the same substrate, which does not spread heat out quickly to the other dies (parallel to the die plane), even a 0.1W extra from one die is insignificant. I guess that after that single die heats up (if it would heat up), then the Vf may drop back down again. I agree that the current hogging of parallel connected emitters is a bigger concern... So, never mind then...

I believe that I was forwards there, but my feet were running too fast and my head was lagging behind. I spoke too soon before thinking about the conclusion. Think of a cartoon character with Super Acme Turbine-Powered Running Shoes and then the exxagerated momentum of the head lagging behind after the take-off. I am going to bed now...

 

[baterija]

For the P7 or MC-E wired 4P matching only has to be good enough to make sure an individual die doesn't exceed max current to that die. Don't forget that the individual die in an SSC P4 or Cree XR-E is rated for a max current of 1A. Using the same technology in the individual emitter the P7 is rated at a max of 2.8 amps (700ma average). That gives them some safety margin that apparently is enough with whatever systemic controls they use to manage Vf balance.

Now if you run the package at 4A, YMMV depending on the distribution of Vf's and the individual die ability to stand any overdriving that occurs.

 

[space]

Have anyone measured the Vf vs. Current for the individual dies in a MC-E? Temperature control is of course very important when doing this comparison.

Just asking because I'm curious about how good they match. Cree's decision to make separate connection to every die could mean they don't do any Vf matching. SSC need to have much more control about this for their P7.


space

 

[chimo]

Have anyone measured the Vf vs. Current for the individual dies in a MC-E? Temperature control is of course very important when doing this comparison.

Just asking because I'm curious about how good they match. Cree's decision to make separate connection to every die could mean they don't do any Vf matching. SSC need to have much more control about this for their P7.


space

I did a quick check on a MC-E (ran them in series and checked the individual Vfs). On the single MC-E (not a very broad sample base ) I found they were very closely matched. However, if I were going to run these in a series/parallel config, I would pair the dies IAW their Vfs to help balance the current. Ideally, they should be run in series.

 

[TexLite]

Actually, where could I source a very cheap, but semi-accurate voltage meter? I would like to buy four that can measure down to tenth or thousandths of a volt. I could hook all four to the MC-E (one per die) and monitor all of the forward voltages simultaneously.

-Tony

I don't know if it's too late for this,or if it will be on sale for Black Friday,but this would be your best bet I think.

Craftsman Digital Multi Meter $9.00

-Michael