Let's talk DIY lights


@1BigFella have you checked out the citizen 97 cri models? Those may be a lot more to you’re liking. You seemed pretty sold on the vero, but the citizen cobs are top notch.


I’m looking at them on Digikey, but they seem way more expensive per lumen then the Veros. I don’t really care about CRI: That is how humans see the light quality compared to full sunlight, so not appropriate. I want to use these for vegetating, because I plan to mostly put them outside for flowering. Do you think I should get one much shifted to the blue side? Like a 4000 K or 5000 K?


While you are essentially correct about what the cri is @1BigFella , it can change a lot as a grow light. Efficacy doesn’t really matter until you’re comparing lights of identical spectrum. Since the 97 cri peak at deeper wavelengths, and usually wider peaks, they can be equally or more effective with less luminous output.

If you plan on flowering outside then a more blue would be fine and i wouldn’t see as much advantage going to a higher cri. I wouldn’t see any reason to go much colder than 4000k though, especially since you would be potentially flowering with it for a small portion of your grow.


And digikey is great, I order from there every once in a while. But you can usually save a couple bucks and get free shipping from arrow. Their website isn’t quite as user friendly though.


I’m within easy driving distance of Mouser and it looks like they have some good deals on Cree COBs. I’ve picked up components from them before over the years. They have an interactive selection guide like Digikey’s. Nope, I take that back: They’re cheap but power-hogs. Like 65 lumens/watt at best. My HPS bulb is 133 lumens/watt. Veros are about 145 lumens/watt.


Cree? Cxb 3590 is the one you want if you go with cree, unless you plan on pushing them hard they should be more efficient than the Vero 29? You’ll just want to make sure you get the latest bin. I think it will be something like db in 4000k, but it’s been a little while since I looked. The data sheet should spell out top bin for each color temp.

If you’re looking for something that has a really high efficacy, Samsung f series or H series strips are pushing 200 lumen/watt in the colder whites. Would that be an option? Strip builds are relatively easy, and can provide pretty even coverage when done correctly. If you don’t push them too hard you can get away with mounting them to aluminum bar stick with double sided thermal transfer tape. I posted some pics of a small veg light built with bridgelux strips earlier in this thread.

Pretty cool you have that kind of access, no shipping ever!


Samsung is $413.39 for 101000 lumens in the H Series.
Vero is $171.48 for 103248 lumens.
Within reason, efficiency doesn’t matter. I have a whole bunch of 300 watt PV panels that are not hooked to anything I can use to power them.


Which Vero model, and what current do you plan on running?

Efficiency does get expensive, I’m just wondering if you calculated heatsinks cost in there as well? You’ll either need rather expensive heatsinks to run passive, or the added cost and efficiency hit of fans.

Personally, I’m a big fan of the Vero’s. But they’re not more efficient than the Cree flagship model unless driven at high currents, as you stated in the earlier post. Even the bridgelux strips are more efficient than the Vero cobs, as well as much cheaper than Samsung strip.

I’m just really confused on what you’re trying to do. In one post you seem to be looking for high efficiency, in the next low build cost. Whatever you’re trying to do is fine, I’m just not sure how to help.


I’m looking at everything. $/lumen purchase cost, efficiency mainly because I don’t want to generate lots of heat, watts to run a 18.75 square foot veg area most of the year. I think I’ll get 6 of the 400 K Veros that put out 17208 lumens at the test current, and run them at that test current. That should give me about the same lumen/square foot as I’m getting with my 400 watt HPS bulb that runs now in a smaller grow area.

I plan to run a fan anyway to exhaust the area, so it might as well pull air out through the light fixture. So it shouldn’t require a very elaborate heatsink. I can always fire it up and see if the COBs are getting hot and add more aluminum surface area inside the enclosure.

I think pointing to the LEDs has been all the help I need. Thanks!


No problem amigo!

I’m pretty familiar, that’s the 52 volt Vero 29 Gen 7 run at like 2100ma right? If I remember fv at 2.1 is around 55v, making that like a 115 watt cob?

I don’t want to be a downer or anything, but while the efficacy is still decent, I don’t think the electrical efficiency is great that high. Like probably less than 50% par watts vs heat watts. Which brings me to the conclusion that you’re probably going to need a very large heatsink, plus it’s own fan blowing directly onto the cooling fins. Off the top of my head, I’m thinking it’s around 6 sq in per watt of heatsink if you’re actively cooling . But it’s been a while since I sized an active system, I could be off a little. That information shouldn’t be too hard to find though.

I’m just a little concerned about the amount of heat there is all.


It’s 148 lumens/watt. 17208 lumens at 1.71 amps with forward voltage of 68.1 volts. That’s more efficient than my 400 watt HPS bulb and I am handling that heat fine. I realize the LEDs have to be kept a lot cooler to work well. They use about 700 watts total. Digikey selection guide says to keep them below 85C. Fortunately they radiate heat a LOT better than my HPS bulb out the back into the aluminum heat sink. I see some stuff on the internet that suggests a perfect 100% efficient lamp would generate about 683 lumens per watt. That leaves (683-148)/683 watts of heat per watt total, or about 550 watts of heat.

I plan to use extruded aluminum fin heatsink material and use enclosure to force all the exhaust airflow through the fins. The important thing is that the heat doesn’t get dumped back into the grow chamber. It gets vented outside.


On a separate but related topic, why do you need to move the lights up and down? If you had only a single light source and an infinitely large greenhouse, then the light would spread at the square of the distance. But you don’t. You have a scrog that covers the entire floor of the tent or grow chamber, reflective walls, and multiple lamps. As you raise the lights, a particular plant gets less light from the lamp directly overhead, but then it gets more from the other lamps and reflective walls. I think all the light comes out the front of the LEDs, and gets absorbed by plant leaves. So were does the unabsorbed light go when you raise the lights?


Do Its Like 116.5 watts. I think you may be confusing tc with ambient? Case temp is much different than ambient. The data sheet is telling you what you need to keep diode temp at in order to maintain that efficacy, and the listed life expectancy of the diode. I’m just telling you from experience, that it takes quite a bit to do that at 75 watts, let alone almost 120.

I run the 36 volt Vero at 2.1, for just over 70 watts. On a $25 pin heatsink with good air circulation they are pushing 70c, sometimes more. It’s entirely different than maintaining ambient temp with an hps. It can be done, but I think you’re gonna need a pretty a serious heatsink and active cooling.

Like I said earlier, pushing the lumen/watt issue isn’t a big deal. There are lots that are as good or better than 148. It’s just my opinion that you’ll be better off targeting your desired light using more cobs at lower output. It will drive your build cost up, but your light will perform better for longer. And I believe you’ll be a lot happier in the long run.

That being said, I don’t want to be telling you what to do either. If you think it will work, give it a shot.


I was going to run them at the test current. I think the data sheet says they can actually handle twice that if you get serious about heat control. You are running a different model at it’s test current, likewise being conservative. The whole heatsink problem has so many variables that I was just going to build one, measure, and then throw more aluminum in there or get a stronger fan to get it right. Do you think 70 C is safe enough? It says 85 C on Digikeys parametric selection guide but I like to stay below the maximums.


I understand. I guess my point was 75 watts is hard to cool, 116 will be more difficult. Watts are the watts, and you will have as many if not more heat watts as my vero’s have. Which model I run compared to the one you run won’t change that, right?

The test data vs max data is there, and it’s also subjective. They can take twice the test current, if you keep them below 85c, before they will fail. They will fail very shortly below max current if you don’t keep them cool. Even at the test current, if you don’t keep them at 85c or less, you sure as hell won’t be getting 146 lumens/per watt. The charts on the data sheet should show that, and they pretty much stop at 85c for most. The last I knew anyway. If they go beyond 85c, it usually fall straight to the bottom meaning they have become nfg.

I’m just saying that going with 8, or even 6 cobs here is a much better alternative all the way around. Cooler, more efficient, and you’ll probably get some bonus light as well.


So rather than guess at the data sheet numbers I decided to pull them up…

I’m pretty sure I located the cob you’re looking at. Do you see where it says the same cob at 1140ma and 25c is 180 lm/watt? Or at 1710ma 161 lm/watt?

Not that you’ll be able to keep them at 25c, but even at 50c you’d only be giving up a few. Double the amount of cobs and cut the current in half and everything will go together a lot better for you. They’ll be much easier to cool, and you’ll have a much better spread of light. The average builder is probably going to recommend 12 cobs minimum for a room your size, their effective footprint isn’t all that big. You can certainly bump the current a little to cheat that, but I don’t personally feel like going that far would be in your best interest. 4 of them at 75 watts each is probably going to create more par than a 400 watt hps, regardless of the amount of lumens produced.

The lowering and raising of light is indeed to change light intensity. You make a somewhat valid point about the reflection, but reflected light isn’t 100% efficient. So raising your light decreases the light intensity everywhere. If you have to raise a light to get the wall, you can also assume the amount of light then getting to the reflective wall will be less, and less than 100% of it will reflect onto your canopy.

This is also partially the reason you don’t see 4x4 light fixtures take up the entire top of a 4x4 tent. Light manufacturers play the reflective wall they assume you will have into their favor. Most of the quality led lights you’ll see a window where the light is raised and center par readings will go down, but edges will come up. Personally I prefer dimming over moving. But dimmable driver and dimming source cost extra, and some people aren’t as lazy lol.


Oh, yeah. I assume a little over 85 C and they go into massive heat production mode and fail rapidly. That’s the main limiting factor for all blue LEDs. Forever fail, as in no recovery. But even 70 C is pretty dang hot. That’s 158 F, for all you US non-scientists. Fry your plants good if it gets too close. Might even be smart and include a temperature sensor in the heatsink to turn down the current if it gets too hot. Typically the tiny microprocessor that will run the switcher has an input for an I2C or SPI temperature sensor and they cost about $1. It’s not like it needs full light flux when the ambient air gets too hot.

The good thing, is I can run some experiments with just one COB. So it’s just $28 if I fry it, but I don’t plan on hurting it. I can run different currents through it and see how much heat sink and fan I need, to get what temperature. My grow room is pretty much never over 85 F because I supply air conditioning if it gets hotter. If the fan and heatsink requirements are too crazy, then I can just buy more COBs and run them at lower current, Better light spread anyway.


Ok @1BigFella we’re on the same page lol. I would definitely do some testing when pushing them. Depending on your heatsink and cooling, you may be fine. And I double checked data sheet, 105c looks like the no go zone. Perhaps I seen the 85c from another manufacturer, or just talking out my ass. But 85 is indeed pretty damn hot.

When you get around to it, throw pics up or whatever of the build. There’s a few guys here interested. I’ll definitely follow along!


Nope, I think you are right about the failure temp. Digikey selection guide lists 85 C as the TEST temperature, not the maximum. I think you can run them a lot hotter than we (as growers) want to run them.

I’ve got lots of time for experiments: I don’t harvest outside until mid-October, then I have to regenerate all my different strains back into moms and at that point I need some lights for vegging inside.


Be careful with your interpretation of the test data. The test data means they ran it at said current, and captured the data at listed temp. It will run at said temp, at much lower efficiency than the other temp they list. Tc max on data sheet is listed at 105, so it’s not that much more so to speak, 20%.

The test data is not where anyone necessarily intends to run them. As an led builder, I look for where the gains are. Pounding the watts out of a couple of cobs is not where the benefit of led is. If you started from square one, where would this light rate vs your hps? It’s gonna have a little more par per watt, and run slightly cooler. For the cost, you’re not really gaining much. But when you start dropping the drive current those gains become larger and larger. The downside is that the build costs go up. So when we start talking about kick ass cob lights, they bring a heavier price tag is all. If you run 1000 watts and can get a 10% boost in electrical efficiency, that takes your draw down to 900 watts with same par. On a veg light running 18 hours a day, 100 watts can go a long way. Will eventually pay itself off.