cheap DIY LED

[naoki]

I made a blog post about how to make Chip-on-board (COB) based LED fixtures (here is the link). The basic contents are similar to what I posted earlier in this thread. I included a little bit more assembly details (maybe too much for most people), information about driver selection, and new measurement data using PPFD. Also I came across a couple of useful youtube videos, so the links are provided at the very end of the post. I started this thread with cheap eBay COB, but the video provides the reason why these are not worth getting.

There are newer generations of COB LEDs (Bridglux Vero Version 2.0 and Cree CXB series) after I posted the messages in this thread. Some people are achieving close to 70% radiant efficiency with Cree CXB3590 (only 30% of electricity is wasted as heat, and 70% is converted to light energy). A cheaper way to grow orchids under artificial light!

 

[naoki]

For DIY COB, drilling and tapping heatsink is probably the most intimidating part. The rest is something which can be done by anyone who can strip wires and use a screw driver.

I noticed that Cutter is starting to offer pre-drilled heatsink + COB kit. It is not the cheapest, but for the people who doesn't want to drill, this could be an easy option. The shipping is a bit expensive for people outside of australia (especially if you are interested in only a few units). There is 10% discount for the first 2 items with the coupon code.

passive cooling, CXB3590:
https://www.cutter.com.au/proddetail.php?prod=cut2901
About 50W for $80 (+$35 shipping). You'll need to get your driver (about $20, including shipping) separately. I would prefer a bit beefier heatsink, but it probably works well.

The following is similar, but it comes with 4x COB (about 200W) and a driver (i.e. a complete kit). About $430 with shipping:
https://www.cutter.com.au/proddetail.php?prod=cut2900


Active cooling version:
CXB3590 ($66 + $19 shipping)
https://www.cutter.com.au/proddetail.php?prod=cut2792
CXB3070 ($56)
https://www.cutter.com.au/proddetail.php?prod=cut2791

You'll need a driver and an AC/DC adapter 5-12V (e.g. cell phone charger).

For the driver, you can use something like Meanwell LPC-60-1400 (data sheet).
Here is the current price (about $16 + shipping):
https://octopart.com/search?q=LPC-60-1400

 

[Kawarthapine]

One of the best threads I've come across.

Thanks to all who contributed.

 

[naoki]

I came across this CXB3590 kit, which was released in 1-2 month ago:

http://timbergrowlights.com/200-watt-cree-cxb3590-4-cob-kit-meanwell-hlg-power-supply-free-shipping/

This is good for people who doesn't like to assemble since no assembly is required. It will be good to cover 4'x4' (or 8'x2') area for moderate light orchids.

Similar to the Cutter kit, but for the US, it is a bit cheaper. Good components; 4x top-bin CXB3590 driven softly (50W per COB) and the best MeanWell driver. I'm not familiar with the heatsink they use. It is all pre-assembled, so you don't need to do anything. You only need to figure out how to hang them. Dimmer is included (goes down to 10%). $399 + free shipping for 200W (4x 50W). If you assemble the components by yourself, you can save $100 or so, but for people who doesn't have time to assemble or not sure how to assemble, $100 extra may be worth. They also have 5x50W version for $499. http://timbergrowlights.com/products/diy-grow-light-kits/

This is pretty much the best in terms of PAR efficacy (about 2.5 micromol/J from my calculation), and most LED grow light can't touch it. The PAR efficacy is measuring how many photons relevant to plants are produced for a given amount of electricity. In comparison, Fluece (=BML) SPYDRx, released in early 2016, is one of the very best and achieves only 2.2 micomol/J. I said "only" 2.2, but it is a very high efficiency number. You should know that cheaper LED grow lights give about 1.0 micromol/J, only slightly better than T8 or T5HO. Very good HPS is around 2.0 micromol/J. In terms of lumen (not so relevant, but people are more familiar with it), it is giving about 177lm/W.

In terms of COB front, I recently learned about a better alternative to Cree. I don't need a newer grow light for me, but if you are interested in DIY route, I would definitely look into Citizen CitiLED COB Series Version 5. There are several sizes, but something around CLU048-1212C4 seems to be the sweet spot in terms of price and efficiency ($12.55 from CDI). This is a smaller COB, so you'll need 2 of CLU048-1212C4 to replace 1x CXB3590. At the moderate driving current, the efficiency is very close to CXB3590. Maybe you can drive it at 1050mA (36.8W), which gives 169lm/W @ case temperature (Tc) of 50C, or better yet at 700mA (23.7W), which gives 179lm/W (similar to CXB3590 @ 50W).

 

[naoki]

There are quite a few interesting LED arrays released this year. But Samsung H-series is pretty excciting, and it is finally shipping. I received mine 1-2 weeks ago after waiting for a couple months. I won't have time to assemble them for a while, but I'll post about it sometime later this year. I thought that I should give a quick heads-up in case someone is considering linear (strip) array style LEDs.

It is a bit costly, but the spec. is quite amazing! Here is the link to the data sheet. It achieves 187 lm/W for 4000K (>2x efficacy of T5HO or commonly available LED bulbs). It has a higher efficacy than many COBs, so it is impressive.





I got 30x Samsung H-series H282D, 4000K, SI-B8T05128HUS LED strips and a couple of Meanwell drivers (APC-16-700 and APC-25-500). I got 1' length, but it comes in 2' and 4', too.

$10.93 per strip from digikey (link), but I got $7.83 ea for the volume discount.

APC-16-700 $6 ea from Mouser, each can drive 3 strips (link)

APC-25-500 $10.62 ea from Mouser, each can drive 4 strips (link)

I'm going to attach them to Aluminum C-channels like what I did with XF-3535L earlier in this thread. While XF-3535L is flexible, and it comes with double sided tape, H-series is rigid, and you need to use thermal adhesive, screws, or magnets to attach it to the heatsinks.

Some of these will be used for planted aquarium, but others will be for orchids in smaller enclosures (e.g. Rubber Maid plastic containers, smaller aquarium tanks etc).

 

[naoki]

I just spent a bit of time to write up a blog post to help DIY LED grow light builders. The DIY COB post is one of the most frequently accessed post in my blog for some reason (even though I think other biological posts are more fun to me), so this is a follow up, and explain more fundamental principles of LEDs. In this post, I tried to explain the basics of driver selection and the design process. Here is the link to the blog post. It talks about the multiple LEDs driven by a single driver, basics of driver and LED efficiency, and super simple thermal protection. I tried to target beginners, but it might be too much information or a bit too technical. Hopefully, it is useful for some people since winter is coming (well, it's been here for a while in Alaska).

Here is my recently cleaned grow tent with 8x COB LEDs.

Here is a PPFD map of 4x CXB3590 with HLG-185C-1400B. It is showing amazingly even distribution of light over 120x120cm (4'x4') area in the grow tent. The unit is micromol/m^2/s.

Here is the relationship of dimming vs overall efficiency. So with this driver, using dimming function doesn't decrease the efficiency unless you dim it extremely (i.e., as long as you stay above 30% of the full power, it is explained more in the blog post).

 

[terryros]

Naoki, what is the dispersion angle on your bulbs? I am guessing that your lights are 100 degrees or more, rather than focused?


Sent from my iPad using Tapatalk HD

 

[JAB]

Awesome info Naoki! I think it is a bit over my head, but none the less it gives one plenty to chew on. Thank you for sharing your experience and knowledge so openly.

Cheers
Jake

 

[naoki]

Thanks, Jake!

Terry, I'm not using any optics, so I think the beam angle is 120 degrees. I think there are pros and cons of narrower beam, and I'm not completely sure which would make it better overall. For now, I'm going for the scattered light effects (of wide angle) to reduce shadowing of lower leaves.

 

[terryros]

Naoki, there is some published data that vertical light affects leaves of some plant types differently than diffuse light. Because many of my LEDs are 40 degrees or 60 degree beam angle, I have about three years experience watching this affect and I am pretty sure that with mostly horizontal leaf type orchids (e.g. Phals and Paphs) produce thicker and more compact leaves under even low light levels (e.g. 50 micromoles/m2/sec), as though they were seeing high light. I don't see the effect with more vertical leaf types like Miltoniopsis and Cattleya. I overall like it because I get compact plants that still bloom and I don't have to use high intensity light to get it. But, it is certainly strange.


Sent from my iPad using Tapatalk HD

 

[naoki]

Thank you for sharing your experience, Terry. It is very interesting. I quickly looked at google scholar, and you are right there are some literature about it. Maybe at the forest floor, there are more diffused light, so directional light may imitate stronger light, and more compact leaves (i.e. the classic sun vs shade leaves). I should read more about it.
With this kind of COB, I can't use too narrow beam because it requires too much head space. Even without any optics, I have to place it 3' above the shelf.

 

[gonewild]

Thank you for sharing your experience, Terry. It is very interesting. I quickly looked at google scholar, and you are right there are some literature about it. Maybe at the forest floor, there are more diffused light, so directional light may imitate stronger light, and more compact leaves (i.e. the classic sun vs shade leaves). I should read more about it.
With this kind of COB, I can't use too narrow beam because it requires too much head space. Even without any optics, I have to place it 3' above the shelf.

Spaced 3 feet above shelf what is the footprint size the light covers on the shelf?

 

[naoki]

Lance, inside the reflective tent, 4 of CXB3590 covers 4'x4'. So each is covering about 2'x2'. I have dimmed down to 160W (each is running about 40W) for Paphs. It might be still a bit too much (I think it is getting more than 150-200 micromol/m^2/s at the leaf level). The PPFD map in post #66 is done with full power (213.2W). So I think 210W of this LED is putting out more than 8 bulbs of T5HO (about 400W), and I get more head space.

Cree CXB3590 is a bit old now, so it isn't the best selection at this moment. Bridgelux Vero 29 Generation 7 is much cheaper and more efficient now. But for non-DIY people, CXB3590 can be had in pre-assembled form (post #64 of this thread) at a reasonable price. Timber seems to have a smaller/cheaper one now; 2x CXB3590 (100W) for $235, or 4x Citizen CLU048-1212 (200W) for $300.

 

[naoki]

Happy Thanksgiving!

Another DIY LED update. I made a plastic container based grow space. I used Samsung H series LED linear modules this time. This module was announced in Summer 2016, but it started to ship in October. This Samsung basically beats Philips XF-3535L (I posted about this previously) in all aspects. It has amazing efficacy (187lm/W), even higher than most COB LEDs. I haven't calculated the PAR efficiency, but I'm pretty sure there is no commercially available grow light with this high efficiency. It is a bit more expensive per light output than COB, but it is easier to assemble, and better for a small area. Also, it provides very even light distribution. It costs about $60 to cover 2-3 sq. ft.

I built this for pygmy Drosera, so it is targeting high-light plants. But if you increase the distance or spacing, it will work with orchids, too. I listed an alternative configuration with lower output. It takes about 1 hour to assemble.

Here is the link to my Orchid Borealis blog post.

Drosera dichrosepala ssp. enodes from Scott River. Catalogue #10a of Drosera Gemmae.

 

[naoki]

I've been very happy with Samsung H-series (see post #74), but there is an improved version announced at the end of Aug. 2017. It is called Q-series. Very similar to H-series, but about 10% (actually more like 8%) increase in efficiency. Now it exceeds 200 lm/W at the nominal current! In comparison, typical house-hold LED bulbs or LED shop light you can get from Walmart etc. are barely 100 lm/W. The price is similar to H-series. It is more limiting in the driver selection than H-series, though. I've added possible drivers for the new Q-series in my previous blog post (scroll to the very end).

As a different note, which I have forgotten about, I was curious about the longevity of those cheap eBay 100W COB LEDs (the ones in the very beginning of this thread). So even though it is cheaper to replace it with a high quality COB, I kept one running. It lasted 3 years with 13h/day. It was driven at around 45W, I think. It lasted shorter than Cree, but it lasted longer than I expected. Just as a data point.

 

[Bjorn]

Naoki, have you seen the Integrated COB that has been offered for some time now? the driver is integrated onto the chip and everything becomes quite neat and you get rid of those bulky drivers. Connects directly to mains in other words.
http://www.ebay.com/itm/220V-20W-30...hash=item3aca6310d9:m:myaZytjrJAeBhaXtewCae1A
I have made a unit with 4x100W (rated but uncertain about its real performance) that is running at my Place now for a couple of weeks. Light seems to be ok (and stronger than expected-perhaps it really is 100W?) But have not made much measurement on them yet.

 

[naoki]

Bjorn, I didn't know about it until last week, but I saw it mentioned in Orchid Board. Because of the super cheap price, I got interested in, too. But after looking at this video, it doesn't look too promising. Clive knows well about electronics, and he has been showing informative videos about LEDs:
https://www.youtube.com/watch?time_continue=207&v=KKd2L9Exw0M

Yes, their rated wattage appears to be close to the actual wattage according to the video.

The problem of those so-called driver-less COB is that they appear to be targeted for flood light type application where you don't keep them on for 13h/day. So they are likely to be driving each LEDs at a high current to put emphasis on a higher total output for a given cost, and sacrificing the efficiency. As Clive mentioned, we don't know the spec. of the diodes, so this is just a guess. But the web site mentions 80 lm/W, which is the efficacy level of household LED bulbs from 2-3 years ago. So even if they are given for free, it is cheaper to buy a newer, more efficient modules for most people (With your cheap electricity, you are probably an exception!).

I think the larger makers like Philips have produced similar "driverless" LED modules. It is certainly convenient. The current offering seems to have generally lower efficiency than the separate driver + LED combo, where we can choose the driving current. But once they produce under-driven modules with super high efficiency, it will make DIY even easier!

 

[Bjorn]

Had a look at the video. Interesting. Seems that I got some that are slightly different, mine has an attached capacitor and this remioves flickering as far as I can judge. Also the size-I got the elongated design, perhaps they are better-who knows, time will show.

 

[naoki]

The long rectangular one (100W) seems to have more diodes than the square 50W ( I counted 81/82 per row, so about 249 vs 79). If they are using the same diodes, it is driven slightly softer.

100W seems to have 10 IC current controller (50W in the video had 5). So they are probably similar design.

I'm not sure flickering is an issue, though. Fluorescent light flickers, right?

 

[Bjorn]

No, do not think that plants bother. Nevertheless, I use it to grow salad and Herbs. Much faster than outdoors these days (even in greenhouses). I'll find out re. longivety.