DiY LED - Cree CXA3070
- Thread starter SupraSPL
- Start date
indianajones
Well-Known Member
got a message yesterday that the floods were ready to ship.
hopefully they are as slick as they looked online. going to take
them to a local place that does injection molding for metals
and see if i can't get them reverse engineered and manufactured
in 'murica instead of importing.
hopefully they are as slick as they looked online. going to take
them to a local place that does injection molding for metals
and see if i can't get them reverse engineered and manufactured
in 'murica instead of importing.
Attachments
iPot
Well-Known Member
Where did you get that heatsink?View attachment 3292436 Got the heat sink sanded and have a piece of 1/8" aluminum for the top. Will measure and cut hole for fan later. I have the holes for the cob holders drilled and tapped. The driver came today, it's the dimmable one http://www.meanwell.com/search/hlg-120h-C/HLG-120H-C-spec.pdf. I just need to know what to get and how to install it and whatever it is to make it dimmable. I need instructions on dimming for a dimwhit.
I tested Purple Growers 'AB's' with my cheap meter. All I can say for certain is that it was '2.5-3" brighter than Z2 from digikey'
ReeferDance
Well-Known Member
Well that just makes me want to wait till I can get some AB's.......But they seem so rare now-a-daysI tested Purple Growers 'AB's' with my cheap meter. All I can say for certain is that it was '2.5-3" brighter than Z2 from digikey'
I'll honestly take whatever i can get my hands on though when the time comes
tbh I wasn't expecting such a noticable difference but it probably is (20% or something?), there is a slight chance .. maybe.. that the AB could even be a Z4, i'll buy a real AB one day!
Yes, grab what you can, even my Z2 are fine!!!
(though I feel slightly underpowered... it would have worked better for my small grow to run 4x Z2 @ 1a than .7a, AB may have bridged the gap well enough to stick with AB/.7a).
Yes, grab what you can, even my Z2 are fine!!!
(though I feel slightly underpowered... it would have worked better for my small grow to run 4x Z2 @ 1a than .7a, AB may have bridged the gap well enough to stick with AB/.7a).
medicinehuman
Well-Known Member
I got on ebay, hard to find good ones there.
BuddyLeeCU
Member
I don't understand why all the ocd behavior with the heat sinks. Theres no need to sand or polish if youre using a good quality thermal paste. There's no need to run a massive heatsink. Its simple and cheap to actively cool with 80 or 90 mm fans. Buy a cheap 12v DC converter off amazon with screw connectors and match amperage for the number of fans. Get deans connectors to make it easy to put together or pull apart.
http://www.amazon.com/gp/aw/d/B006NTNGN0
The 2 amp version will run 6-8 fans. The tab has screw connectors.
http://www.amazon.com/gp/aw/d/B006NTNGN0
The 2 amp version will run 6-8 fans. The tab has screw connectors.
Last edited:
SupraSPL
Well-Known Member
You are correct it is not necessary to sand them, especially if you are using good paste. Neither is it necessary to use gigantic heatsinks. But, by running as low of a Tj as possible you are eliminating lumen depreciation, squeezing out some extra photons and reducing the overall heat of the system. It is true that modern whites/blue LEDs stand up to heat very well (in terms of temp droop.) But I also run reds and deep reds that have poor temp droop characteristics, so running cool makes a huge difference to them.
So if you want to run passive cooling for whatever reason, you really have to tend to your thermal path. Even with polishing, top notch thermal paste and gigantic heatsinks, my passive cooled lamps still run warmer than if I had used any old CPU cooler with a fan. BUT, I already had a ton of heatsinks on hand so I figured why not put them to use. This setup would be a lot more messy if I had to wire up a bunch of fans. You only have to do it one time, once you flatten a heatsink it is good for life. As LED efficiency continues to rise, we can install more and more watts onto these.
Another thing to consider, for those who are running a bunch of COBs at 1.4A, if they were all on separate CPU coolers imagine what a mess the setup would be. These large heatsinks make a system like that more practical and give a decent spread to the COBs. The lamp in this pic dissipates 1040W, if we went with any less heatsink/fan power the Tj will really start to rise and output would suffer. I don't see a better solution except maybe even longer heatsinks to give more spread. (These heatsinks are actively cooled with 140mm fan at 12-15V)
You may have seen reports, there are many commercial lamps that literally burn out due to insufficient cooling and poor thermal contact. Here is how bad the bottom of every single aluminum star really is. Why not take a few seconds to fix it.
So if you want to run passive cooling for whatever reason, you really have to tend to your thermal path. Even with polishing, top notch thermal paste and gigantic heatsinks, my passive cooled lamps still run warmer than if I had used any old CPU cooler with a fan. BUT, I already had a ton of heatsinks on hand so I figured why not put them to use. This setup would be a lot more messy if I had to wire up a bunch of fans. You only have to do it one time, once you flatten a heatsink it is good for life. As LED efficiency continues to rise, we can install more and more watts onto these.
Another thing to consider, for those who are running a bunch of COBs at 1.4A, if they were all on separate CPU coolers imagine what a mess the setup would be. These large heatsinks make a system like that more practical and give a decent spread to the COBs. The lamp in this pic dissipates 1040W, if we went with any less heatsink/fan power the Tj will really start to rise and output would suffer. I don't see a better solution except maybe even longer heatsinks to give more spread. (These heatsinks are actively cooled with 140mm fan at 12-15V)
You may have seen reports, there are many commercial lamps that literally burn out due to insufficient cooling and poor thermal contact. Here is how bad the bottom of every single aluminum star really is. Why not take a few seconds to fix it.
Last edited:
DonPetro
Well-Known Member
What grit do you use?
SupraSPL
Well-Known Member
The US and Europe grit is very different. Going by the US numbers, I start at about 120grit and finish at 1000 grit. The vast majority of the work is done with the starting grit. Once you have it flat, then you can very quickly go through a few steps to get it to 1000 grit smooth. If you want, you can continue to 2500 grit and then polish it but we could probably bump into the law of diminishing returns above 1000 grit. On the other hand polished heatsinks may reflect some scattered light into the canopy?
Last edited:
I Hate Usernames
Member
I like the idea of using a switch but these are rated at 45C, should I be looking for a higher rated switch? Running at 1400 mA the datasheet shows max of around 100C, I don't want to have that high of temps but I also don't want a switch that cuts off early.
Greengenes707
Well-Known Member
SupraSPL
Well-Known Member
The 45C would be your heatsink temp. Typically our heatsinks run no more than 35C and some stay around 30C. So if something went wrong with your fan or ventilation and your heatsink rose as high as 45C, the switch would cut the power long before any major harm was done to your COB.
I Hate Usernames
Member
Thank you both, I wasn't sure what the typical temp was. I think I have most things figured out (I hope anyways) but I do have another question. Lets say I am running 3-3070 cobs in series off of the HLG-185H-C driver and I want to add some deeper reds to the string. Is this possible running at 1400 mA or would I be required to make a new string? Seems like most of these red leds have a max current of only 700 mA.
Positivity
Well-Known Member
I don't recommend running reds at even 1a. They run real hot at higher currents. 700ma should be good though
I Hate Usernames
Member
Thank you, so a separate string for the reds it is then. I was just hoping to use up some of that left over voltage!
Greengenes707
Well-Known Member
you could split it into 2 parallel strings after the series cobs. 1400/2=700ma a string. No one I know has done it...but it can be done.
Hey Supra,
I'v been following this and other threads lately, you've persuaded me to start myself with you efficienty-charts
.I've been looking for the best heatsink setup, goal is to have as low Tj as possible and at the same time as stealth as possible (closet is in my living room 136W x 58D x 229H)
I don't care about work or cost, it's gonna be my only closet and not the first of a few
Would it be worth it to add a copper plate to the heatsinks to dissipate the heat as fast as possible from right under the cob ?
At Burminghamaluminium you can calculate thermal resistance, and there's quiet a difference...
been going forwards and back between those heatsinks :
pieces of 20cm 9800HS with 12cm fans (0.037°C/W a piece) or 2x120cm 9700HS with 20cm fans on top (0.55°C/W / a piece) both 1m/s
Last edited:
SupraSPL
Well-Known Member
Hello Kuif That is a nice sized closet I am sure you can make some magic happen in there. If cost is no object you can run them soft and significantly reduce the heat that needs to be removed, making the grow ventilation very quiet. You could panasonic bath fans that are near silent. If you can scare up some 3000K CXA3070 ABs and run them at 700mA, they are 49% efficient minimum and dissipate 24.5W ea. So that is only 12.25W of heat per COB, relatively easy to handle, especially with active cooling. Cost is about $4/PAR W.
Regarding the copper plate, although it would spread the heat away from the COB very quickly, it does introduce another thermal interface between the copper and aluminum and it is very expensive in terms of man hours to prepare those surfaces correctly, unless it is machined with precision like high performance CPU heatsinks. So I think the best approach is to use a heatsink with a reasonably thick base plate of pure aluminum, all the way to the fins and to make sure you have sufficient surface area to dissipate the heat into the air (30-40cm²/W for active cooling)
That said, if you run the CXA3070 at 700mA and use active cooling, there will be virtually no gains to be had by overbuilding the cooling system beyond the levels we normally recommend. The reason is because modern whites and blue stand up very well to heat (temp droop) and tend to suffer mostly from current droop instead. So it would be most effective to spend your resources running them softer rather than overbuilding the cooling system. For example, when I tested a CXA3070 at 2.1A on a basic solid aluminum 100W CPU cooler with the fan running at 12V, the decrease in light output as it warmed up was only 2%. Based on that test, I expect there to be virtually no decrease in output from ambient to thermally to warm, when running at 700mA and actively cooling. That is awesome for us!
Everything I said above does not apply to the Cree XPE and XPE2 reds, they really suffer from temp droop and not so much from current droop. However, our standard heatsink recommendations are effective at keeping them cool, especially if the stars are flattened.
So if you want to push the envelope to new territory of efficiency, you could simply run even softer than 700mA and use twice as many COBs. At 300mA the 3000K CXA3070 AB bin is 54.9% efficient minimum (Tj 50C), but Tj would be much lower than 50C in this case, so you could reach 56.5% efficiency minimum, 58% typical. That is about 190lm/W of warm white! The downside is the cost, about $9/PAR W, 4X-5X more than some DIYers are paying for 40% efficient lamps. You could push it even further by running them at 270mA. Driver cost would be negligible, plenty of cheap 270, 300 and 350mA drivers available.
That is a nice sized closet I am sure you can make some magic happen in there. If cost is no object you can run them soft and significantly reduce the heat that needs to be removed, making the grow ventilation very quiet. You could panasonic bath fans that are near silent. If you can scare up some 3000K CXA3070 ABs and run them at 700mA, they are 49% efficient minimum and dissipate 24.5W ea. So that is only 12.25W of heat per COB, relatively easy to handle, especially with active cooling. Cost is about $4/PAR W.Regarding the copper plate, although it would spread the heat away from the COB very quickly, it does introduce another thermal interface between the copper and aluminum and it is very expensive in terms of man hours to prepare those surfaces correctly, unless it is machined with precision like high performance CPU heatsinks. So I think the best approach is to use a heatsink with a reasonably thick base plate of pure aluminum, all the way to the fins and to make sure you have sufficient surface area to dissipate the heat into the air (30-40cm²/W for active cooling)
That said, if you run the CXA3070 at 700mA and use active cooling, there will be virtually no gains to be had by overbuilding the cooling system beyond the levels we normally recommend. The reason is because modern whites and blue stand up very well to heat (temp droop) and tend to suffer mostly from current droop instead. So it would be most effective to spend your resources running them softer rather than overbuilding the cooling system. For example, when I tested a CXA3070 at 2.1A on a basic solid aluminum 100W CPU cooler with the fan running at 12V, the decrease in light output as it warmed up was only 2%. Based on that test, I expect there to be virtually no decrease in output from ambient to thermally to warm, when running at 700mA and actively cooling. That is awesome for us!
Everything I said above does not apply to the Cree XPE and XPE2 reds, they really suffer from temp droop and not so much from current droop. However, our standard heatsink recommendations are effective at keeping them cool, especially if the stars are flattened.
So if you want to push the envelope to new territory of efficiency, you could simply run even softer than 700mA and use twice as many COBs. At 300mA the 3000K CXA3070 AB bin is 54.9% efficient minimum (Tj 50C), but Tj would be much lower than 50C in this case, so you could reach 56.5% efficiency minimum, 58% typical. That is about 190lm/W of warm white! The downside is the cost, about $9/PAR W, 4X-5X more than some DIYers are paying for 40% efficient lamps. You could push it even further by running them at 270mA. Driver cost would be negligible, plenty of cheap 270, 300 and 350mA drivers available.
Last edited: