Cree's 174lm/w at which amp? 700MA?

Hi,

Cree's website states that the maximum efficacy for the 3590 is 174lm/w . Can someone tell me which MA this is at?

I'm looking at the "relative luminous flux" charts and calculating my lm/w based on that data but i'm wondering how are doing it?

Thanks!

89ers said:

Hi,

Cree's website states that the maximum efficacy for the 3590 is 174lm/w . Can someone tell me which MA this is at?

I'm looking at the "relative luminous flux" charts and calculating my lm/w based on that data but i'm wondering how are doing it?

Thanks!

Click to expand...

Yep. Between 700ma(~25w) and 1000ma(~36w) will get you 170lm/w+. Depending how you cool them...1400ma can do it. Assuming the newer DB bins.

Use the PCT with a Tc= ~55*c for less than 1000ma, and Tc=~70c for 1400ma or more, which is easily doable even passively at those currents. Actively cooled can realistically looking at 40c or lower Tc's.
http://pct.cree.com/dt/index.html

Those numbers and figures correspond, and are confirmed by independent sphere testing of the older "CD" bin cxb3590 on I believe a 140mm sink.
Tc was not recorded for the sphere, but my own testing, as well as others testings, show ~58c+/- 5c for a no air movement 140mm heatsink test. So 70c in a lab setting room inside a sphere... is very accurate, and what the PCT predicts given the parameters such as the Tc's.





***That is all assuming no diver or optics losses. So if including at least the driver...1000ma or less should keep you at ~170lm/w for your system.

just for fun when i multiply my recent ppfd sphere measurements by 3.8 for CXB DB on two different days (diff temps). trend seems to fall along same lines

my 2.4A measurement was a little higher than yours as it was aggressive active cooling and measurement taken before steady state



so if that has basis it seems cobs can get to 180 lm/W at 10-15W and top 190 at 5W - but it seems we are at the practical limit there

@Greengenes707 do you think PWM would help those low current measurements push 200?

Thats actually calibrating to your CD sphere measurements. so assuming DB is +5% over CD and other current COBs are right there too we technically should be somewhere about ~190 lm/W at 10-15W and 200 lm/W at ~5W. something more like this:

CobKits said:

just for fun when i multiply my recent ppfd sphere measurements by 3.8 for CXB DB on two different days (diff temps). seems to fall along same lines

my 2.4A measurement was a little higher than yours as it was aggressive active cooling and measurement taken before steady state

View attachment 3924260

so if that has basis it seems cobs can get to 180 lm/W at 10-15W and top 190 at 5W - but it seems we are at the practical limit there

@Greengenes707 do you think PWM would help those low current measurements push 200?

Click to expand...

Those measurements I posted are from a CD...not DB, so they are not lining up. But when you see it in the PCT you can side by side them. It would be a factor of 4.2 to be inline for DB...specially considering active cooling.

I use a storm controller and dim via PWM for all my testing. And then measure current and voltage while it going.

Greengenes707 said:

Those measurements I posted are from a CD...not DB, so they are not lining up. But when you see it in the PCT you can side by side them. It would be a factor of 4.2 to be inline for DB...specially considering active cooling.

Click to expand...

i actually realized that and went back and redid that post before i saw your response. isnt 4.2 a 10% bump? thats a bit more than i would anticipate from a chip spec upgrading from (presumably the higher range of) CD to (presumably the lower range of) DB. as midpoint bin to bin is like 7%. they do tout the recent 10% bump in efficacy but then we would expect every model to jump 1 bin minimum and 2 in some cases, no?

Greengenes707 said:

I use a storm controller and dim via PWM for all my testing. And then measure current and voltage while it going.

Click to expand...

so on a standard dc voltage measurement between controller and diode is the the PWM invisible? and if so, how do you quantify overall input power?

CobKits said:

i actually realized that and went back and redid that post before i saw your response. isnt 4.2 a 10% bump? thats a bit more than i would anticipate from a chip spec upgrading from (presumably the higher range of) CD to (presumably the lower range of) DB. as midpoint bin to bin is like 7%. they do tout the recent 10% bump in efficacy but then we would expect every model to jump 1 bin minimum and 2 in some cases, no?


so on a standard dc voltage measurement between controller and diode is the the PWM invisible? and if so, how do you quantify overall input power?

Click to expand...

I get what you're trying to say, but just too much speculation going on there though. The PCT is based on steady state operation and min figures for the cx chips and is really accurate when accurate Tc's are used...like shown in the CD test. Those chips were bought from us a long time ago via fletch420...or something like that...and chilled gave him gen1 modules for the trade of info from what I can deduce. So mid generation CD's. He never went through with the side by side, or at least reported on it, and now we are looking at gen 2's. But it's the only, just chip test, and is right on with the pct.

I sort of get what you're asking about the pwm now. The PWM/controller are supported by a separate 12v supply os I have it on when I power the driver, gets a fairly accurate reading. But for what you're trying to get is more precise than my needs.

just wondering because measurement of a pulsed signal is foreign to me. i figure if cheap AC multimeters cant even read AC accurately that there is prob some special way to measure PWM. and of course if you read the input power to the controller you are adding that efficiency as a variable (which may or may not be constant/proportional to input power). not criticizing any method of course, just trying to learn/discuss.

it all goes back to the (b-lux?) mfr rec that cobs ran at ultra low currents are more stable driven via PWM as opposed to CC - stable in what way? even at 0.1A my PAR measurements are stable (extra stable in fact with near zero temperature effects)

BUT, i would think if there was a limitation, we wouldnt see the efficacy curve going nearly asymptotic at low currents, it would level out if there was some hard limitation. i do see a lot more scatter in low current measurements but ive attributed that to a combination of :
1. stray light on meter (hard to work in absolute dark and take numbers
2. rounding errors of voltage and current measurements
3. power supplies may not be as accurate at low currents

now that im using the sphere i think #1 is the predominant factor. it will be cool to build the big sphere and not have a form-factor limit, so i can do a ton of low current measurements with midpowers vs cobs and push them to their absolute limits

Dimming VIA pwm is not diming using pwm. Not sure what you are counting?...the power for the controller has nothing todo with the output if chip output is what your measuring.

It's all CC diming with MW

The way I test...is Build a complete light, then send it to a sphere. Not bench measurements of components.

PWM dimming is voltage oriented. PWM at 0-5v converts to 0-10v linear to drive the current internally by the MW circuitry with a 10v power source.

I drive cxb3590's at 700mA for veg and it is fantastic growth compared to when I used T5's. I don't have a light meter, but I do have a voltage meter connected to my rig for SAG and amps, watts, volts kwh. I believe SupraSPL pointed out based on the cree tables that the maximum power for 3590's is between 15-25w per cob delivering the maximum joules available. More brightness does not mean more joules or energy. Plants thrive on energy absorption and transformation, i.e. joules

It took me about 20mins to figure out what you guys were talking about since i'm pretty new to the scene...

I'm curious to know what the difference in yield might be for a 5%-10% increase in efficacy. @Greengenes707 , have you grown with the CXA3070's? and if so, what was the difference versus something like a 3590 CD bin?

Also, what's it cost to have testing done with a sphere?