I've got something that crushes all the other COBs and other lights

TogiX said:

Any proof here? I keep seeing people say to drive COBs @ 5-10w for 65-70% efficiency but when we finally get any sort of sphere data the numbers are far lower than I thought.

Click to expand...

nobody tests them that low in a sphere - every measurement costs big money and they are usually only testing at normal operating currents

i believe 60%+ efficiency is easily attainable, heres some curves of single chips in a 20" sphere (not cree but they all generally have the same shape)



so *IF* you have a cob that is 50% efficient at 50W you can see that its easy to get to 60% at 20W, but to get to 70% youd need to be well under 10W

i dont see anybody hitting 240 lm/W (~75%) with any cob at a realistic operating level (and ive tested several DB bin cobs - theres nothing magical about them -nor does PCT go anywhere close to that low of a current). most all cobs converge at low currents so if youre trying to break efficiency records by running under 10 W/cob you want to be stacking up $10-12 cobs (or single die SMDs like the quantums), not $40 crees

TogiX said:

Now I wish they'd included junction temps in the sphere tests (only ambient @ 25.1c). Not sure how these sphere tests work but something tells me they do a few quick pulses

Click to expand...

for fixtures they usually let it warm up to steady state

TogiX said:

Guess we'll have to wait until people actually start putting an integrating sphere to work around here....... It's pretty frustrating because I've been waiting on sphere data from people saying months ago that they "have one and it'll only be a few weeks until testing," meanwhile I'm still figuring whether I should buy 3 1212's and drive them @ ~16w a piece or buy one Vero 29C, drive it at 50w and be done with the whole thing.

Click to expand...

well i guess that must be me i dont know anybody else with a sphere on the board. customer orders come first before i can play/do r&d so nothing going on that front for now.

i will say that the few sphere tests i have done seem to line up with the old-style 12" spot tests so i certainly think those are valid for relative comparison

by my spot tests gen 5 1212 is as efficient at 28W as vero C is at 50W so two 1212 would be more flux than one vero C (apart from better coverage). extrapolating from a separate test (single chip gen 5 vs gen 6), gen 6 1212 @ 40W is as efficient as vero C at 50W so that seems like an easy call for 1/3 of the price. which would also put gen 6 1212 right there with cxb and 1818 gen5@ 40W (dont get too excited just yet that was just one test)

CobKits said:

for fixtures they usually let it warm up to steady state

Click to expand...

but i think not at this test, because look at the voltage numbers in the test report from chiledled. They are pretty high, so it was just a pulsed measurement i think.

licoricetree said:

Defiantly wanna see sphere results cxm-32 gen3 vs cree cxb3590 db that would be interesting.But should we compare a 36v to a 36v like vero d.

Click to expand...

chip voltage is wholly unrelated to efficiency

youd have to get the cree boys to put up a chip... which they never do. i only have cxm-32 in 4000k 80 cri and it was pulling teeth to get that sample. but yes based on what ive seen so far i dont see anything touching cxm32/clu058 gen 6 at the current time. even the flagship vero C runs neck and neck with 1825 gen 5, so theres at least 3 chips above that (1825 gen 6 and 3618 gens 5 and 6).

i have to retest cxm 32, if i recall, at a cursory look it did well, but i lost my logsheet before i could put all the data into excel and graph it

Malocan said:

View attachment 3888257
but i think not at this test, because look at the voltage numbers in the test report from chiledled. They are pretty high, so it was just a pulsed measurement i think.

Click to expand...

maybe they were running the water cooling?

CobKits said:

maybe they were running the water cooling?

Click to expand...

hm, yes true. Could be the reason maybe

those do look high

my tests on CXB:
A V
0.7-33
1.4 34.7
2.4-36.8

that 0.7 A test is being read about 1-2 minutes powered up on a 140mm heatsink with a 120 mm fan cranking ~22C air down on it. case temp cant be much over 40C in that scenario. in PCT i have to literally put in subzero Tj temps to match those voltages

CobKits said:

those do look high

my tests on CXB:
A V
0.7-33
1.4 34.7
2.4-36.8

that 0.7 A test is being read about 1-2 minutes powered up on a 140mm heatsink with a 120 mm fan cranking ~22C air down on it. case temp cant be much over 40C in that scenario. in PCT i have to literally put in subzero Tj temps to match those voltages

Click to expand...

so it must be pulsed or not?

i dunno. pretty sure all cree data is pulsed. seems weird tho. usually these labs have very accurate volt and current meters,

so keep on using those 700mA drivers or upgrading to 1050mA? i have 32 cobs here planned for a 4x8. 3500k 3590 CD bin chips

if you can afford the extra chips, sure.

theres very little difference between 1050 and 700 mA. 50% more chip cost (and holders/reflectors, and possibly heatsinks if using discrete ones) for 5% relative improvement in ppfd/W.

same with 1400 to 1050, right about 5% improvement, not a lot

now you see why 1400 mA is so popular

unless your region has you space/ plant number constrained, the same number of chips at 1400 mA and doubling your space will have a much greater return

CobKits said:

chip voltage is wholly unrelated to efficiency

youd have to get the cree boys to put up a chip... which they never do. i only have cxm-32 in 4000k 80 cri and it was pulling teeth to get that sample. but yes based on what ive seen so far i dont see anything touching cxm32/clu058 gen 6 at the current time. even the flagship vero C runs neck and neck with 1825 gen 5, so theres at least 3 chips above that (1825 gen 6 and 3618 gens 5 and 6).

i have to retest cxm 32, if i recall, at a cursory look it did well, but i lost my logsheet before i could put all the data into excel and graph it

Click to expand...

Gen 6 of citizen are already available?

If you look at cobkits.com there are some 1212 gen6's

Cree is sooo 2015.

Nenno44 said:

If you look at cobkits.com there are some 1212 gen6's

Click to expand...

How to choose gen 6?
I just see 1212 or 1818 bur not gen 6 or 5

G84 said:

How to choose gen 6?
I just see 1212 or 1818 bur not gen 6 or 5

Click to expand...

choose the 1212's I think its the 90cri i think, select you Kelvin rating and you should see gen6 in there

CobKits said:

chip voltage is wholly unrelated to efficiency

youd have to get the cree boys to put up a chip... which they never do. i only have cxm-32 in 4000k 80 cri and it was pulling teeth to get that sample. but yes based on what ive seen so far i dont see anything touching cxm32/clu058 gen 6 at the current time. even the flagship vero C runs neck and neck with 1825 gen 5, so theres at least 3 chips above that (1825 gen 6 and 3618 gens 5 and 6).

i have to retest cxm 32, if i recall, at a cursory look it did well, but i lost my logsheet before i could put all the data into excel and graph it

Click to expand...

Do you have the CXM-32 Gen3 chips? I been looking at those and it seems they are a little more efficient then the Cree cxb according to the data sheets and lm/w

no as minimum order is 300 and big chips dont sell

i got something that crushes all other cobz and other lights ,its my sisters hammer lmao

i guess he forgot to load the video hahahaha that fuckin mars panel looks pretty damn durable actually

sixstring2112 said:

i got something that crushes all other cobz and other lights ,its my sisters hammer lmao

i guess he forgot to load the video hahahaha that fuckin mars panel looks pretty damn durable actually

Click to expand...

Looks Like He Has Some Suga In His Britches And I'd Give The Win To The Marrs Panel. Hope His Next Smash Goes Better Than His Last Smash Just Saying