MCE vs XRE-Q5 output at same drive level?

[gfang]

How much more efficient would an MCE be compared to an XRE Q5 at the same drive level. Or should I say at the same power level.

Seems to me that you would get more light from a parallel MCE at 1000 ma or a series MCE at 250 ma per die than you would from a single XRE Q5 at 1000ma, because the dies are more efficient at a lower power level.

Does this sound correct? Has anyone seen a direct comparison?

The reason I ask is because I would like to replace one of my Q5's with an MCE-5A. I am aware that the 5A bin is less efficient but I was hoping that the overall efficiency of an MCE vs XRE would give me the same amount of light at a nicer tint.

 

[gfang]

Ok, after a little more digging.


@ 250 ma Q5 is about 78 lumens.
@ 1000 ma Q5 is about 237 lumens.

So the MCE in a similar bin could be around 312 lumens @ 250 ma per die.

The MCE in a 5A bin is approximately 20% less efficient so....

@ 250 ma per die MCE 5A is about 250 lumens????

If this is the case I could get the tint I want and slightly better efficiency with the MCE 5A. So what are the pros/cons of MCE parallel @ 1000 ma vs MCE series @ 250 ma. I imagine that because vf is slightly less at 250 ma for one die that the overall power usage of series would be less than parralel. However for a 3 or 4 cell NIMH setup I dont know which driver config is goung to be better, boost or buck? Is one type of driver generally more efficient than the other? Currently I am using the GD1000 from the shoppe. It works great!

 

[gswitter]

How much more efficient would an MCE be compared to an XRE Q5 at the same drive level. Or should I say at the same power level.

Seems to me that you would get more light from a parallel MCE at 1000 ma or a series MCE at 250 ma per die than you would from a single XRE Q5 at 1000ma, because the dies are more efficient at a lower power level.

Does this sound correct?

Theoretically, if the XR-E and MC-E were using identical dice (unlikely). But there's enough emitter to emitter variance and other variables (such as the two applications' heat management) that the question may not be answerable without testing the two specific emitters.

More importantly, do you really think the difference in efficiency will be that significant? So much so that you'd notice it during normal day-to-day use? You're eyes will probably notice the difference in tint, but not the difference between 237 and 250 lm.

 

[gfang]

gswitter,

I should have specified that I am considering the MCE 5A over the XRE Q3 5A because I would like to have the same output and efficiency as my Q5. I suppose the output difference between the Q3 5A and Q5 WC would not really be all that much, but I really like to maximize my possibilities. Besides, I could bump up the drive on the MCE if I wanted but the single die emitter is pretty close to max already.

It is really going to come down to a side by side comparison.

 

[2xTrinity]

Ok, after a little more digging.


@ 250 ma Q5 is about 78 lumens.
@ 1000 ma Q5 is about 237 lumens.

So the MCE in a similar bin could be around 312 lumens @ 250 ma per die.

The MCE in a 5A bin is approximately 20% less efficient so....

@ 250 ma per die MCE 5A is about 250 lumens????

It won't be exactly the same, as the die used is not the only parameter -- there's also the package/optics etc. This will tend to be more lossy for the MC-E than the XR-E.

However, there's another paramter that will work in your favor -- if you're feeding the same amonut of power into a MC-E as you used to into a XR-E, the MC-E should be running considerably cooler. This is because the heat will be spread across four dice, so total thermal resistance from inside the LED to the outside world will be less. (It has to be, for the MC-E to be abel to survive consuming more than 3x the max rated power of the XR-E)

Your maximally driven Q5 will be degraded somewhat compared to the datasheed numbers due to the fact that the real-world emitter won't be producing light at a low ambient temperature. The MC-E won't be hurt as much by this effect.

If this is the case I could get the tint I want and slightly better efficiency with the MCE 5A. So what are the pros/cons of MCE parallel @ 1000 ma vs MCE series @ 250 ma. I imagine that because vf is slightly less at 250 ma for one die that the overall power usage of series would be less than parralel.

MC-E parallel 1000 is almost identical, as far as power consumption and heating, as running the MC-E in series at 250mA, or in series-parallel at 500mA. In each case, each individual die will be seeing roughly 250 (so far, in my experience with the MC-E the dice are fairly evenly matched, so current will be divided evenly between the four dice if you run them in parallel).

However for a 3 or 4 cell NIMH setup I dont know which driver config is goung to be better, boost or buck? Is one type of driver generally more efficient than the other? Currently I am using the GD1000 from the shoppe. It works great!

Generally, drivers are more efficient when output voltage is close to input voltage. For 4 NiMH, I'd use a buck converter and wire the MC-E in parallel. Since you're only gonig to be driving at 1A, there are tons of buck drivers that can easily handle that. That has the added benefit that the converter should fall out of regulation at roughly the exact time as the batteries are just spent and need to be recharged.

 

[gfang]

Thanks 2xTrinity.

 

[eprom]

Hi gfang,

My observation is these MC-E 5A's has very low Vf. When I am driving the Paralel wired MC-E 5A at 2.4A (~600ma per die) the Vf is only ~3.1V and even lower when driving with lower currents.

So,

When you are driving the Q5 and MC-E 5A at same current, your power consuption will be lower with MC-E 5A

example (Q5 results based on jtr1962's tests):

Q5 @1000ma = 3,5V*1A = 3,5W = 237 lumen
MC-E 5A (Paralel) @1000ma = 2,9 * 1A = 2,9W = ~240 lumen (according to MC-E datasheet 250ma per die gives the ~0,75 of the luminous flux of 350ma per die)

So..

Advantages of MC-E 5A :

* less power more lumen
* Neutral white output
* Underdriving the MC-E will give more life to LED
* Very low Vf gives a chance to use Buck topology which is cheaper than Boost topology

Disadvantages :

* Hard to focus 4 die's
* Not for throw race

Thanks,
EpRoM

 

[yellow]

I think (but have no data, of course) that this will be a difference of theory to reality

in theory that difference might be there,
in reality I doubt it

... there will be no difference at all
(except for the quad-beam being wider with the same focusing device)

has anyone done this yet?


[edit]just noticed eproms post states data, that there will be no difference [/edit]

 

[liveforphysics]

Being a multi-die device, I hadn't thought I would have good performance running parallel because I thought there would be Vf differences.

Turns out, of the 16 MCEs I've used so far, they all roughly the exact same Vf on each dice. I don't know how that is possible, but perhaps Cree has develop some new technology to ensure similar Vf's to enable the multi-die chips to be driven in parallel.

 

[2xTrinity]

Being a multi-die device, I hadn't thought I would have good performance running parallel because I thought there would be Vf differences.

Turns out, of the 16 MCEs I've used so far, they all roughly the exact same Vf on each dice. I don't know how that is possible, but perhaps Cree has develop some new technology to ensure similar Vf's to enable the multi-die chips to be driven in parallel.

I think Vf differences are generally due to minor differences in doping (that vary from one "batch" of semiconductors to others), or differences in temperature between the different devices (the LED Vf is strongly sensitive to temperature). With the MC-Es the dice are likely all cut from the same wafer, and they are thermally coupled to each other inside the same package.

There's no guarantee for a multi-MC-E build however that one package will be the same Vf as the other.

That's why for example if I'm going to wire 2 MC-Es, in 4-series, 2-parallel, rather than wire each MC-E emitter in series, and then put both MC-E's in parallel to each other, I'll wire two strings each composed of 2 dice from MC-E 1, and 2 dice from MC-E 2. This way, if one MC-E gets hotter than the other, BOTH of my parallel strings are affected in the same way. Sum of the Vf's across the strings will likely stay the same.

 

[StevelKnievel]

I have also observed that the MC-E dies are very well matched. Maybe it has something to do with the abundance of bond wires?

 

[4thpage]

(according to MC-E datasheet 250ma per die gives the ~0,75 of the luminous flux of 350ma per die)
EpRoM

So wait. Are the specs on the data sheet for the MC-E (and for P7, etc.) specs for each individual die of the 4 dice? I ask, because the data sheet gives the max forward current as 700 mA, and I see all you knowledgible folks talking about driving the emitter a 1000mA parallel?

Does this mean you could drive the emitter up to 2.8A?

Thanks for answering a newbie's question!

 

[znomit]

So wait. Are the specs on the data sheet for the MC-E (and for P7, etc.) specs for each individual die of the 4 dice? I ask, because the data sheet gives the max forward current as 700 mA, and I see all you knowledgible folks talking about driving the emitter a 1000mA parallel?

Does this mean you could drive the emitter up to 2.8A?

Thanks for answering a newbie's question!

Yes 700mA per die is the spec(so 2.8A if in parallel).