DiY LED - Cree CXA3070

AbeFroman

Well-Known Member
Supra I have a question about these spreadsheets I keep seeing. What is the $/Par w I keep seeing to the far right?

From what I gather. Using the driver I got (HLG-185H-C500A). I'm better off running the 3590 over the 3570. It appears they like to be driven over

Is 500mA ok for the 3070? Or would the 3590's be better?
 
Last edited:

nogod_

Well-Known Member
You would be better off running the cxa3590 @ 500ma.

Your efficiency is going to be crazy good @ 500ma with the cxa3070 but at the expense of overall output. With each $40+ chip dissipating <20w its going to be very rough on the wallet to build something that compares to an 1000w hps.

If you build with cxa3590 you double your voltage which means double your wattage @ 500ma. If you can find a high bin 3000k cxa3590 you are powering 5 x 35w cobs over 50% efficient on one 94% efficient $65 driver and a $40 chunk of aluminum.

=booya



Supra I have a question about these spreadsheets I keep seeing. What is the $/Par w I keep seeing to the far right?

From what I gather. Using the driver I got (HLG-185H-C500A). I'm better off running the 3590 over the 3570. It appears they like to be driven over

Is 500mA ok for the 3070? Or would the 3590's be better?
 
Last edited:

SupraSPL

Well-Known Member
To be up front, the charts are just an estimate because each COB will have some variation and the chart assumes everything is at junction temp (Tj) of 50C, when in practice Tj can be slightly lower at low currents and much higher at high currents.

That said, the $/PAR W column is a way to compare the actual cost of the photons, in terms of PAR W. The calculation takes the COB efficiency into account, which is affected by things like Vf shift, temp droop and current droop. Current droop can be significant especially with blue/white LEDs.

The orange line in this graph is where output would be if there were no current droop. The difference between the orange and black line is the droop
current droop.png

The $/PAR also makes it easier to choose between different COBs or to figure out which COB is most suited to the job you have in mind. For example if your goal is to minimize up front cost and simplify a build you can run a Vero 29 hard. Or if you want to maximize efficiency you can run a CXA3590 soft. Most DIYers go with something in between like nogod suggested.

You can take it a step further if you include the cost of the other components such as driver and heatsink, because that can change the picture and may affect your decision about which drive current to run at. For example choosing between the HLG-185H-C1400 and HLG-185H-C1050. The 1400 can run 3 COBs total 156W and the 1050 can run 5 COBs total 190W, yet the driver cost is the same and the 1050 version benefits from reduced current droop and less heatsink required to do the same job. So the $/PAR W cost of running at 1050mA is not as much higher as you would expect.
 
Last edited:

NapalmD

Well-Known Member
Been awhile since any updates. Nothing too exciting just pheno hunting. Getting well over 1 gpw every run but no record breaking numbers, probably due to my messing around with the soil mix lately. CO2 is 800-900 steady.These ladies are all at 49 days
Sin City - Alien Nightmare
View attachment 3349700

G13 Labs - Blue OG
View attachment 3349701

Mango Diesel - unknown pedigree, someone very determined drove this cutting thousands of miles. Luckily it tested well and holds its own with the other keepers.
View attachment 3349702

TGA - Ace of Spades
View attachment 3349703

Divine Genetics - ChemD X Girl Scout Cookies - fem beans #2 (hermed-a-lot)
View attachment 3349704

Divine Genetics - ChemD X Girl Scout Cookies - fem beans #3 (amazing Chem aroma)
View attachment 3349705
Sick!
 
hello all.

i am a newbie here on RIU and have very much enjoyed reading through this entire thread and a few others on LED's. there are so many great ideas and so much new info to absorb!

simply amazing work y'all ... massive kudos, esp to supra, b/c of the incredible time and efforts he spends documenting and detailing his research here.

my first HID grow was in 1985. i am finally convinced it is time for an update to this newest lighting technology ...

first question: still don't understand the formula for calculating heatsink requirements for a given active or passive system design. i have followed supra's math formula earlier in this thread, but cannot come up with the same results. tsk, tsk

will someone please explain the step-by-step process to calc the correct numbers, using an arctic 11 PLUS as an example?

what is the passive as well as active cooling capacity for this unit?
 

Swomie

New Member
First post on this forum, lets go.
How many "CXA3070-0000-000N00AB30F" can i run on a Mean well "HLG-185H-C1050A"?
I think I can run 4 or 5 dont know.


I wanna cover 5x5 area with good efficiency, and thinking of a using 20 cobs @1050ma
 

Imptspd

Well-Known Member
First post on this forum, lets go.
How many "CXA3070-0000-000N00AB30F" can i run on a Mean well "HLG-185H-C1050A"?
I think I can run 4 or 5 dont know.


I wanna cover 5x5 area with good efficiency, and thinking of a using 20 cobs @1050ma
You can run up to 5 on that driver.
Edit. If you run 20 Cobbs with 4 or 5 of those drivers, you will be using right around 800 watts from the wall, this would give right around 32watts per square foot for a 5x5 area.
 
Last edited:

SupraSPL

Well-Known Member
Hello @HeadStash1111 . Not sure if you have seen the heatsink thread although the info changes so fast some of it is already out of date anyway. Here is a summary of my current understanding. As a general rule, 25cm²/dissipation watt for active cooling and 75cm²/dissipation watt for passive cooling. There are some significant variables, such as how much airflow the heatsink is receiving from the circulation fan, how tall the fins are, how wide the space is between fins, how thick is the base plate, and how efficient you are running the COB.

The good news is, it doesnt matter that much. Sticking to those general rules, best case, your COBs will experience .25% temp droop. At worst, maybe 5%.

The Alpine 11 is good for up to 50W dissipation with the fan running at 5V (.36W), up to at least 100W dissipation with fan running at 7.5V (1W). I havent tried going to higher dissipation but that would be an interesting experiment.

Actual data: CXA3070 3K AB running at 2.47A (98.9W) on Alpine 11 with stock paste: fan running at 5V=4.4% temp droop. fan running at 7.5V=3.21% temp droop. Increasing fan speed further uses a lot more electricity but only a very slight improvement in temp droop because of the law of diminishing returns, it is a net loss of efficiency to increase fan speed in this case.

Another way to think about it, heat exchange is more efficient with a larger temp gradient, so as the heatsink temp approaches ambient temp and temp droop is approaching zero, it takes a lot more energy to continue pushing toward zero.
 
Last edited:

SupraSPL

Well-Known Member
To calculate surface area of a heatsink, you first calculate the "perimeter" of the profile. The formula = (profile width X 2) + (number of fins * fin height * 2) + (base height * 2). One you have that you just multiply by the fin length and you have surface area.

There is one little hang up though, the surface area of the profile face. It is very significant for a piece of heatsink that is cut short and has a wide profile (like @churchhaze setup) To make is easy I just calculate the surface area of the base plate area (profile width * base plate height * 2)

HeatsinkUSA provides the perimeter and surface area for its heatsinks which may save you the trouble.

In regard to using CPU coolers for passive cooling, they have surprising low performance because the space between fins is optimized for active cooling. It requires a slight pressure to get the air moving but will not convect naturally, so its effective surface area is quite small. For example a $10 chunk of the 4.9" profile (4.9"X6") would perform much better and will cool about 20W passively because it has a thick base plate, thick fins, wide fin spacing, short fins and provide 75cm²/W @ 20W dissipation.
 
Last edited:
thank you for the additional info supra. just saw your latest post plus, i will check out the heatsink thread. just finished this one, the powering thread and the one on reflectors.

my LED system design basic strategy is unfolding as a one COB+alpine 11 plus per SqFt setup, each on a yo-yo, using cxa3070 3K AB's @ 1050mA ish for bloom area and vero10 5K's @ 350 mA or less for veg area.

how many watts of cooling power does an alpine 11 offer WITHOUT the fan running at all, ie. passive mode? how soft would i have to drive a single vero10 to simply passively cool it with the alpine11's heatsink and save running the fans themselves as a backup, if needed?
 

SupraSPL

Well-Known Member
Great plan, a cutting edge build :)

To make sure I clarify for DIYers, the Vero has the impressive ability to bring very low cost to grow lamps, but it is not necessarily the value point. For example in order to match that cost ($1.18/PAR W) and efficiency (40%) with CXA3590s, you would have to run a top bin 5000K CXA3590 at 112W! So that goes to show the abilities of the Vero at reducing up front cost to very low levels.

That said, in the build example above, HS would be spending $2.36/PAR for the flower lights (CXA3070 AB at 1050mA) and $1.18/PAR W for vegging light (Vero10 5K @ 350mA). But the vegging light runs 18-20 hours/day, so by spending an equivalent amount/ PAR W on the vegging lamp, you could run Vero18s at 350mA and get 47.5% (17.4% more light/W). So it may be that the value point is at a higher cost/PAR W.

But if you have bit of extra vertical height and if you can wait to source them, you could really hit a grand slam if you ran a CXA3070 5K BB bin at 800mA, you would get 55.25% efficiency for $2.56/PAR W, (36.7% more light /W)
 
Last edited:

nvhak49

Well-Known Member
So a buddy and I been wanting to replace his 1000w HID and I been telling him about the DIY cobs and he's pretty interested in it Cuz he wants to cut the electric bill down and have less heat. How many 3070 or 3590 cobs what I need to compare to a 1000w HPS, what what cobs would be better and how strong should they be ran from the driver to equal it?
 

AquariusPanta

Well-Known Member
So a buddy and I been wanting to replace his 1000w HID and I been telling him about the DIY cobs and he's pretty interested in it Cuz he wants to cut the electric bill down and have less heat. How many 3070 or 3590 cobs what I need to compare to a 1000w HPS, what what cobs would be better and how strong should they be ran from the driver to equal it?
So let me get this straight, you want one of us to do all the work that comes with planning and organizing numbers regarding lighting so that your friend can replace his 1000W HID in attempts to cut down his electric bill?

Somethings missing here...:confused:
 

nvhak49

Well-Known Member
So let me get this straight, you want one of us to do all the work that comes with planning and organizing numbers regarding lighting so that your friend can replace his 1000W HID in attempts to cut down his electric bill?

Somethings missing here...:confused:
I was just wondering if anyone has already done this befor, I'm still very new to making LEDs and still ordering parts for my first DIY light. I'm not trying to make anyone do anything for me jwing if it's already been done before that's all?
 

alef

Active Member
Dear mates,

I finally did it. 3 vero 29 lights and one dimmable driver. I know that the connection is not optimal and it is better to run them continuosly but I was not able to find the cheap way to do so...

The parameters of the driver are:

U = 42V
I = 5.6A

So each LED gets almoust 2A which is an optimal load. The total strength of light is approximately 27000. I have to run the full cycle to teat it and I am planning to compile the luxometr to test the parameters.
 

Attachments

SupraSPL

Well-Known Member
@nvhak49 Yes it can be done. DIY COB is very customizeable so you can build for minimal up front cost/medium efficiency, balanced cost/efficiency, high efficiency/high cost, or very high efficiency/very high cost.

The higher the efficiency the more electrical savings but probably even more importantly, huge heat reduction. The heat reduction can be due to COBs higher conversion efficiency than HPS, removing the infrared peak that occurs at 775nm with HID lamps, but mostly because you can get the same job done with so much less dissipation Wattage. This is thanks to the higher conversion efficiency, more efficient spectrum, reduced reflector losses and potentially a more even spread.

Here are some common designs for flowering
Vero29 with HLN-80H-42 (2.1A, 79W, 37.5% efficiency)
CXA3070 with 1.45A drivers (54W ea, 43.3% efficiency)
Those examples are very straight forward, one driver/COB

You can use drivers that run multiple COBs like the HLG-185H-C1400. They are dimmable, high quality, high efficiency and highly recommended to further reduce electricity and heat.

If you have questions about wiring check out the "how to power them sticky". There are threads dedicated to heatsink info and lenses/reflector info. Post up your build and questions. Good luck!
 
Last edited:

Empherio

Active Member
These DIY builds are inspirational! I've got several pc psu's lying about- so, how could that be modded into a COB psu? It's got lots of 12V, 5V, molex, etc, 350W approx total output. Is it not feasible? Forgive if this is not the place to ask, but I recognize @SupraSPL as the DIY LED Guru. The info I've gotten on the site (and this thread) is excellent and cutting edge.
 

SupraSPL

Well-Known Member
I am with you man DIY is a lot of fun for us LED/weed nerds and the DIYers here bring a lot to the table :leaf: The designs that we currently recommend represent contributions from a bunch of different members. KNNA got us started with hand mount OSRAM golden dragons and we were building lamps that were 26% efficient. Now (7 years later) we can build lamps that are 65% efficient and cost less cost/PAR W.

Most COBs run in the 36-40V range and would take about 2A each, so the 12V would need to be stepped up by a DC-DC boost converter and the COBs would be run in parallel. That would work if the COBs had very close Vf curves or if you used balancing circuitry. So in practice it might actually end up costing more and would result in a much lower efficiency than constant current AC-DC drivers. In the range we are working with, we can get up to 94% efficiency and that is really hard (impossible?) to top. Some AC DC converters can get up to 96% efficiency but they are gigantic $250 600W units and run constant voltage.
 
Last edited:
Top