Top bin COB comparison

Status
Not open for further replies.

EfficientWatt

Well-Known Member
It's a simple equation : Total Watts consumed by led = Par watts + Heat loss watts

SO IF:
- You know how much energy goes in (electricity consumed / Watts)
- You know how much light is emited (= electricty energy converted into light energy)

Then:
- The difference between both is the inefficiency part, ie direct heat.

:peace:

thanks, I suppose that is a good answer for the real world. I was asking how you guys know 50% of the power put into a LED comes out as PAR watts. But maybe that math would be getting kind of crazy and I really don't need to know that. :-)
 

BOBBY_G

Well-Known Member
thanks, I suppose that is a good answer for the real world. I was asking how you guys know 50% of the power put into a LED comes out as PAR watts. But maybe that math would be getting kind of crazy and I really don't need to know that. :-)
thats ascertained from the "efficiency" numbers for any given LED. modern monochromatic LEDs are ~45-70% efficient (blue more efficient than red). modern cobs are 35-50% efficient when run hard and up to 65-70% efficient when run super soft (using more cobs at a higher initial cost.

see this post, its very descriptive

https://www.rollitup.org/t/cxb3590-1500w.878136/page-2#post-11778236

see this thread to see some efficiency discussion:

http://rollitup.org/t/cob-efficiency-spreadsheets.865238/
 

ketchup45685224

Well-Known Member
Thanks Bobby,
Those links give me something to work with. I'll do to experimenting of my own. I'm getting a lab power supply. I also have a cheap ebay light meter, some Vero 29s and a dozen cheap ebay 100 watt LEDs. I wonder if the efficiency goes up with these cheap LEDs when driven softly. Wouldn't be something if it went up like the name brand ones? Maybe into the 40% range? I can buy 10 of them for the price of one Vero 29. I could drive them at 500 mA or lower and its still worth it.
 

SupraSPL

Well-Known Member
Even the cheapos will increase efficiency significantly at lower power. You can get generic "100 Watt" COBs for $3-$4 ea. They claim 100lumens/W (30% efficient) in warm white, which I do not doubt they are capable of at moderate power. At low current, maybe 120lm/W (~35% efficient) could be possible.

That said, the Vero 10 is only $4.37 and driven at 20W they are 100lm/W ($0.71/PAR W) and probably much better quality control than generic COBs, which are often partially damaged.

The Vero29 driven at 142W is 111lm/W and cost $0.55/PAR W. From what I can tell the generic COBs cannot compete with the Vero series in terms off efficiency or cost.

Also there are the clearance CREE CXAs that pop up on Newark. For example the CXA2530 2700K T2, driven at 30W, 124lm/W and cost $0.60/PAR W.
 

SupraSPL

Well-Known Member
Hmm maybe if so that would definitely change things for me is cxb3070 bb was just as efficient as cxb3590 CD at same current/wattage.
Another way to phrase it, they probably both use the same level of blue dies, but the 3590 should consistently outperform the 3070 at any given drive power. The dies in both 3070 and 3590 appear to be the same size, but the 3590 has more of them (168 vs 132) and they are spaced further apart in the 3590. The 3070 has 11 strings of 12 dies in parallel and the 3590 36V has 14 strings of 12 dies in parallel

So if both the 3070 BB and 3590 CD are driven at 50W, the 3070 dies will be running at approx 127mA and the 3590 dies are running at 100mA. Or you could say each die in the 3070 is dissipating .38W and each die in the 3590 is dissipating .3W. Since the 3590 dies are running lower and cooler, they will be more efficient.
 

Stephenj37826

Well-Known Member
Right
Another way to phrase it, they probably both use the same level of blue dies, but the 3590 should consistently outperform the 3070 at any given drive power. The dies in both 3070 and 3590 appear to be the same size, but the 3590 has more of them (168 vs 132) and they are spaced further apart in the 3590. The 3070 has 11 strings of 12 dies in parallel and the 3590 36V has 14 strings of 12 dies in parallel

So if both the 3070 BB and 3590 CD are driven at 50W, the 3070 dies will be running at approx 127mA and the 3590 dies are running at 100mA. Or you could say each die in the 3070 is dissipating .38W and each die in the 3590 is dissipating .3W. Since the 3590 dies are running lower and cooler, they will be more efficient.
Right that is what I already gathered was just confused on that statement honestly lol.
 

ketchup45685224

Well-Known Member
Thanks SupraSPL,
When I build the Vero 29 prototype. I crudely compared the Vero 29 to the cheap LEDs . When both were driven at 2.1 amps the cheap Leds produced 68% as much lux on the cheap light meter. When, my Power Supply gets here, I'll setup a real testing rig.
 

Greengenes707

Well-Known Member
Bonjour
Why cob are round?
It is strings of dies in parallel so why giving them this shape?
Wouldn't it be more convenient to make square one...it will be better to fit our square or rectangular grow boxes!?
CU
I've heard SDS say that square is better(thermally or electrically...can't remember) than radial(circular). I never looked into it as I DIY panels...not cobs. But it is one of those statement that sticks in your head and always remember.
 

SupraSPL

Well-Known Member
Yes the generics are arranged in square shape, probably easier to manufacture that way. Cree staggers the strings so they can be arranged in a round pattern. The strings in the center are straight across no staggering. I assume the round shape is more ideal for making and even beam pattern and slightly more even cooling when using radial heatsinks.

But for our purposes, if CREE made square ones that were slightly cheaper or more efficient I would go for them.
 

SupraSPL

Well-Known Member
Not trying to pick on Vero but this is the only one I have on hand at the moment, a Vero 10 4000K 80 CRi Version 1.2 (older version). 20W maximum COB but I tested it up to 32W. I also added the CXB3590 3000K CB (re tested). Both were mounted on the exact same heatsink hung in the exact same place, same methodolgy as before.

COB comparison V5.png

Vero10 4000K V1.2.png
 
Last edited:

salmonetin

Well-Known Member
I've heard SDS say that square is better(thermally or electrically...can't remember) than radial(circular). I never looked into it as I DIY panels...not cobs. But it is one of those statement that sticks in your head and always remember.
...just my pov...in cobs...

...in parallel ...the large of wire conections or strings ..counts... then square give us same distance...deal better from my pov...

:peace:

Saludos
 

bassman999

Well-Known Member
Another way to phrase it, they probably both use the same level of blue dies, but the 3590 should consistently outperform the 3070 at any given drive power. The dies in both 3070 and 3590 appear to be the same size, but the 3590 has more of them (168 vs 132) and they are spaced further apart in the 3590. The 3070 has 11 strings of 12 dies in parallel and the 3590 36V has 14 strings of 12 dies in parallel

So if both the 3070 BB and 3590 CD are driven at 50W, the 3070 dies will be running at approx 127mA and the 3590 dies are running at 100mA. Or you could say each die in the 3070 is dissipating .38W and each die in the 3590 is dissipating .3W. Since the 3590 dies are running lower and cooler, they will be more efficient.
Thanks for that breakdown. I looked at the graphs on their respective Vf as they go up in current. I was wondering if something like this was the reason.
 

bassman999

Well-Known Member
Thanks SupraSPL,
When I build the Vero 29 prototype. I crudely compared the Vero 29 to the cheap LEDs . When both were driven at 2.1 amps the cheap Leds produced 68% as much lux on the cheap light meter. When, my Power Supply gets here, I'll setup a real testing rig.
I have watched videos with the cheapos.
As Supra pointed out they are often damaged, and they light up the individual leds at different currents, and often lots of them dont work at all. The same test done with quality ones they illuminate the individual leds as the same time and all work.

That zoomed in image of the strings is crazy! What a convoluted design!!
 
Last edited:

BuddyColas

Well-Known Member
Added the CXB3590 5000K 80CRi CD (retest) from Kingbrite
View attachment 3571646
Supra,

You’re killing me.

Your attention to detail and ability to not be effect by trolls and egos are legendary. And you soooooo need a Vero 29 v2 for your comparisons. Then you can have 100% increase in your Vero collection. Shweeet!

So being relatively new to RIU and the threads. Please let me know how I can discreetly get a brand spanky new Vero 29 v2 sent to you and I will do it.

Thanks. I’ll keep a look out in my “alerts.”

Thanks.
 

ketchup45685224

Well-Known Member
I have watched videos with the cheapos.
As Supra pointed out they are often damaged, and they light up the individual leds at different currents, and often lots of them dont work at all. The same test done with quality ones they illuminate the individual leds as the same time and all work.

That zoomed in image of the strings is crazy! What a convoluted design!!
I've seen some of those videos as well. I have some of those cheap leds and they all light up when the current is high enough. Are they as good as the name brand ones. NO. But, I'm going to see how cheap, I can build a light that replaces a 1000 HPS with LEDS. I thinking of something with the back side of the heatsink is in a tube that is like an air cooled HID figure, pulling air from outside and then venting it back outside. I've seen something similier on RIU using a fence post. I'd build mine out of sheet metal as I have all the tools and a background in HVAC. I'm not sure how much heat will stay in the tube and how much will ended up in the room anyways. I figure each 1000 HPS uses about 1500 watt with the light plus the A/C to cool it. If this works and I can get rid of the A/C it'd would be worth it.
 

littlejacob

Well-Known Member
Bonjour
Just save some time and money and go straight to cxb 3590 3500ºk 36V CD!
600w of it would provide you the same
gpw amd will cost 1000 $ in parts!
So I guess cheaper than a lots of cheap chips run very low on a lot of drivers and hs...and to get to a descent efficiency you will need a lots at 5/10% of max!
I will do some math before if I was you...to be sure it will worth it!?
CU
 

ketchup45685224

Well-Known Member
Bonjour
Just save some time and money and go straight to cxb 3590 3500ºk 36V CD!
600w of it would provide you the same
gpw amd will cost 1000 $ in parts!
So I guess cheaper than a lots of cheap chips run very low on a lot of drivers and hs...and to get to a descent efficiency you will need a lots at 5/10% of max!
I will do some math before if I was you...to be sure it will worth it!?
CU
I'm planning at building my own driver. TI has some really good app notes for their UCC28180 PFC controller. I think by changing the feed back loom. It'll make a great high voltage constant current driver. I done some adding of parts and PCB. I think, I can build 800 watt one for about $75 to 100. I'm not welded to the idea of using the cheap LEDs. But, I do not think it's unlikely I'll go with Crees. I not a hobby grower and sending over a grand to replace one 1000 watter seems crazy. I have some Vero 29's aready and from what I've seen they're a great value. I will be doing some testing and comparing between the cheap leds and the Vero 29s before going ahead with anything. I should be able to build a test rig get the testing done in the next couple of weeks.
 
Status
Not open for further replies.
Top