Heat Question

[videoman]

Which of these produce less heat ? ONE MC-E or TWO XP-G's , both same color temperature and both setups adjusted by the driver to deliver same lumen output, say 600 lumens, and both setups on the same heatsink.

 

[lolzertank]

Two XP-Gs.

 

[videoman]

Thanks for a quick reply. Is the difference close or by a substantial amount like 20 to 40 % less ?? I would like to fit 8 XP-G's in the same enclosure that had 4 MC-E's but was running very hot without any thermal shutdowns. Besides the lower cost, I was thinking also in terms of heat issues and longer battery run times by going the XP-G route.

 

[jmodic]

I think the difference is quite substantial. Quick calculation (approximations based on data and graphs from Cree datasheets):

1. 1 x MCE, M bin (430 lm @ 350 mA)
For 600 lm you would need to drive each of four dies with 520 mA. Vf will be approximately 3.3V per die, so total power consumption will be around 6.9 W.

2. 2x XPG, R5 bin (139 lm @ 350 mA each)
For 600 lm you would need to drive each of two leds with 800 mA. Vf will be approximately 3.2V per led, so total power consumption will be around 5.1 W.

So the diference in power consumption is more than 30% in favor of two XPGs over single MCE.

 

[videoman]

Thanks, that's great news. Would that also equate to a reduction of heat by the same amount ? since the wattage is that much less ?

 

[HarryN]

Hi - I am not convinced that the answer is quite so clear on 1 x MC-E vs 2 x XP-Gs. (both nice LEDs BTW)

I looked again at the white light testing thread here http://www.candlepowerforums.com/vb/showthread.php?t=89607&page=13 post 336 I think.

A rather warm MC-E was tested and produced 600 lumens at 7.7 watts, while a quite cool XP-G was tested and produced 300 lumens at 2.9 watts. While it first seems that there is a 1.8 watt difference, this could easily be the difference in tint and "reality" of what will happen under real operating conditions.

The MC-E also has 1/2 of the thermal resistance of an XP-G, so some of this might depend on how closely you pack the LEDs and how well you can move away the heat.

When you get within a 10% on two partially related data points, it is not so easy to know for sure without a 1:1 direct comparison.

 

[jtr1962]

Harry is correct. The MC-E I tested was a K bin. An M-bin would undoubtedly have done better. Just going by the middle of the bin averages (400 versus 460 lumens @ 350 mA per die for K and M bins, respectively), a cooler M bin MC-E would emit 600 lumens at roughly 475 mA. Assuming a similar Vf to the MC-E I tested, input power would be about 6.0 watts, not much over the 5.8 watts required for a pair of XP-Gs to produce the same 600 lumens. Given that the emitted spectrum of both is probably more or less the same since both are cool white, you'll have about the same amount of power coming out as light (600 lumens equals ~1.8 watts luminous flux). So the MC-E would produce roughly 4.2 watts of waste heat compared to 4.0 watts for the two XP-Gs. An MC-E which falls towards the high end of the M-bin would do even better, perhaps even marginally beating the XP-Gs. Same with one which has a low Vf. In short, the numbers are very close here if comparing like to like.

 

[IMSabbel]

Which strikes me a bit funny, though:
Somehow it does not compute that with the same emitter surface, an MCE should be more efficient than an R5 XP-G with lower Vf.
Something has to be missing there...

 

[Curt R]

Nothing missing, just two years of die technology.

Curt

 

[videoman]

To make a light that outputs say 2700 lumens (not a flashlight ) I was thinking of using 9 XP-G's. Must be neutral white (4-4.5K) when available. Must be dimmable.I am thinking about a Shark Buck 3A. Basically what I am after is the same or more lumen output as 4 (neutral) MC-E's but with an increase in run time and decrease in heat. Battery used is a 14.4V DC of 120 AH rating (brick).

 

[space]

One of the best ways to increase efficacy, is to use more LED's (more die area) at lower current/power. More LED's mean lower Vf = less heat and lower die temp, both lead to higher efficacy.
It will of course cost you.

space

 

[IMSabbel]

Nothing missing, just two years of die technology.

Curt

That would be the other way round. If the XP-G were running cooler.
But the claim was that an MCE of a good bin, with the old EZ-1000 dies, is more efficient than two XP-G R5.
That is what i think does not computer somehow.

 

[HarryN]

The thing to keep in mind is that all of the LED makers are constantly improving the die making process, even for the same "part". If they just kept stamping out the same exact die and putting it in the MC-E or other package, they would soon go out of business.

It is unlikely that the process used to make one die vs another at Cree or anywhere else is dramatically different for very long. Making LED die requires major, continuous R/D investment that needs to be spread over as many parts as possible. The only real exception is when they have large orders for LEDs with specific properties (say cell phone backlights). If they make them "better and brighter", it actually is a bad thing - because they are now "different".

 

[jmodic]

Which strikes me a bit funny, though:
Somehow it does not compute that with the same emitter surface, an MCE should be more efficient than an R5 XP-G with lower Vf.
Something has to be missing there...

Indeed, something is missing here. Or better to say, there is a tiny lapsus calami in jtr1962's calculation:

Harry is correct. The MC-E I tested was a K bin. An M-bin would undoubtedly have done better. Just going by the middle of the bin averages (400 versus 460 lumens @ 350 mA per die for K and M bins, respectively), a cooler M bin MC-E would emit 600 lumens at roughly 475 mA. Assuming a similar Vf to the MC-E I tested, input power would be about 6.0 watts, not much over the 5.8 watts required for a pair of XP-Gs to produce the same 600 lumens.

This claimed 5.8 wats required for a pair of XP-Gs is the wrong number. From jtr1962's own measurements http://www.candlepowerforums.com/vb/showpost.php?p=3115908&postcount=354 , two XP-Gs would yield 600 lumens at 800 mA, where the power consumption would actually be 5.1 W (2 x 2.56 W), which is the same number as I was claiming. That 5.8 W probably came from misreading the table - it would be 2.91 W per led if they were driven @ 900 mA!

Cheers.

 

[IMSabbel]

Indeed, something is missing here. Or better to say, there is a tiny lapsus calami in jtr1962's calculation:



This claimed 5.8 wats required for a pair of XP-Gs is the wrong number. From jtr1962's own measurements http://www.candlepowerforums.com/vb/showpost.php?p=3115908&postcount=354 , two XP-Gs would yield 600 lumens at 800 mA, where the power consumption would actually be 5.1 W (2 x 2.56 W), which is the same number as I was claiming. That 5.8 W probably came from misreading the table - it would be 2.91 W per led if they were driven @ 900 mA!

Cheers.

Thanks!

That seems much more reasonable, a 10-15% difference would make sense seeing the R5 vs R2 maximum for EZ1000 dies specification differences.

 

[jtr1962]

This claimed 5.8 wats required for a pair of XP-Gs is the wrong number. From jtr1962's own measurements http://www.candlepowerforums.com/vb/showpost.php?p=3115908&postcount=354 , two XP-Gs would yield 600 lumens at 800 mA, where the power consumption would actually be 5.1 W (2 x 2.56 W), which is the same number as I was claiming. That 5.8 W probably came from misreading the table - it would be 2.91 W per led if they were driven @ 900 mA!

Yep, I just relied on what you wrote being accurate without bothering to check if the numbers matched those in my lumen testing thread. :crazy: :o

So getting back to what I wrote, a top bin MC-E would still indeed make maybe 4.2 watts of waste heat to produce 600 lumens, but a pair of XP-Gs would only make about 3.3 watts (5.1 watts input power minus ~1.8 watts leaving as light). That's a HUGE difference in waste heat production!