What's new
TerraForums Venus Flytrap, Nepenthes, Drosera and more talk

Register a free account today to become a member! Once signed in, you'll be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!

DIY COB LED info

  • #101
So...It seems that Digikey sold out of the 2' strips. :censor: If Arrow doesn't come through with their stock, it looks like I'll be waiting a while.

I did finally pick out drivers. I'm not too comfortable overdriving the LEDs by 10% (and I have a feeling the drivers may overvolt trying to do so) so I'm going with some more expensive dimmable drivers where applicable.

For the old 3' shelf - LPC-35-700 which can either run 3x1' + 3x1' or (1'x2') + (1'x2')
For the low light shelf - LPF-25D-54 (470mA) with potentiometer dimming for 2x2'
For the high light shelf - LCM-60U (will probably run at 900mA) with potentiometer dimming for 4x2'
 
Last edited:
  • #102
I suppose I still don't quite understand all the different units used to measure light intensity. Lux is easy to understand, at the very least, if not the most accurate way of doing it. What I've read recently is that most plants achieve chloroplast saturation at about 70,000 lux (though I'm not sure about the light source for that number) and I'm having a difficult time figuring out how that converts.

In any case, I find it odd that the saturation point would actually be different in greenhouse-grown plants. Would it increase if they were placed outdoors? Would the plants be just as healthy under each set of conditions?

Yes, it is confusing. lux and footcandles are basically interchangeable, they differ whether it is measured in m^2 or foot ^2. 1lumen/1m^2 is 1 lux. 1lumen/1ft^2 is 1 footcandle. If we know the emission spectra, we can convert them to the plant relevant unit of photosynthetic photon flux density (PPFD), whose unit is micromol/m^2/s. For some light source, here is the conversion factor: Conversion - PPF to Lux If it is under natural sun, 70,000 * 0.0185 = 1295 micromol/m^2/s. Full sun is about 2000 micromol/m^2/s. So the PS saturation in that paper is amazingly low. Actually many things influence the PS rate (temp, CO2 and O2 concentration, RH, C3 vs C4 etc). Also different species have very different relationships between PS rate and intensity. I'm guessing your number is for some crop plants.

Since plants don't have legs, they generally have higher acclimation capacity (than typical animals). Plants modify the leaf structure (e.g. shape of parenchyma cells or number of cell layers etc) to adjust their photosynthetic capacity for the given condition. So you are right, they will have higher light saturation point if they were grown outside. There are some research showing that when plants were brought under continuous artificial light (from gradually changing natural light), their PS system needs to be acclimated (more specifically their circadian rhythm needs to be adjusted). After a couple weeks, they will adjust to the new environment.

Sounds good with the driver selection! It looks like SI-B8U11156HUS (3500K) seems to be in stock (here). Or SI-B8R11156HUS (5000K) here. I would be happy with either of these.
 
  • #104
The 70,000 lux value was given as a typical value for "most plants," though of course CPs are not typical plants. That may be for artificial lighting since elsewhere it lists 1000-1200 micromol/m^2/s specifically for sunlight.

I'm going to give Arrow some time to see if they actually get any 4000k modules in stock today, if only so that the color temperature matches my COBs (it would also be cheaper!) Otherwise I suppose I'll order some 3500k modules since I understand that red light is generally more photosynthetically efficient than blue, especially at lower light levels.

I ordered the thermal glue and picked up some electrical plugs and c-channels yesterday too. I have to say the tolerance on the channels is very tight - they're 19mm wide (interior measurement) and the LED modules are 18mm wide. I have plenty of wire available though I believe it's braided. I'm sure I can just tin the wire for use with the terminals on the LED modules.
 
Last edited:
  • #105
Let there be light! Well, almost. UPS was supposed to deliver the drivers today, but delayed the package due to bad weather - which is showing up TOMORROW. So I'm probably not going to get those until Tuesday since the whole region's going to be covered in ice until then. In any case, I got all the bars assembled today.

DIY LED strip lights by Nimbulan, on Flickr

I would highly recommend a power saw of some sort for cutting the aluminum. Doing it with a hacksaw is a lot of work, takes forever, and makes a big mess.

Also interestingly, I noticed the label on the LEDs rates them at 24.2V, 1200mA (600mA for the 1' model) though it doesn't specify a light output at that power level. It seems the 1050mA drivers would be perfectly safe, though obviously less efficient.
 
Last edited:
  • #106
I also want to add that I found these LED strips to be slightly curved towards the LED side, so the ends were lifting up off the aluminum. I set up a makeshift clamping system with some rubber bands and mahjongg tiles (first thing I found that was an appropriate size) to hole the ends of the boards down, taking extra care not to put pressure on the LEDs themselves, but only the circuit board.
 
  • #107
Alright I am up and running. I wired everything up with 23 gauge solid wire out of some heavy duty cat-5 I had lying around. It's easy to bend and easy to insert into the connectors. The lights are all functioning as expected though I am calculating a color temperature of 4000k, -12 tint in pictures of the lights, which I believe indicates that they are emitting less yellow and more red light than would be normal at 3500k. For comparison, I calculated an almost perfect 4000k for my 4000k Cree COBs.

Adding everything up the total cost was about $290 for ~18,000 lumens which is certainly expensive (would be less if all the strips were hooked up to a single driver, or without buying the more expensive dimmable drivers) but it should save money in the long run with significantly lower electricity costs compared to T5's and no bulb replacements.

LED strip lights by Nimbulan, on Flickr
This definitely opened up my old plant shelf - hopefully it will also result in fewer burned flower stalks.

LED strip lights by Nimbulan, on Flickr
You can see how I wired it here, running the wires straight along the channels to the end.

LED strip lights by Nimbulan, on Flickr
And my hopefully effective lowland greenhouse. It doesn't seem to hold heat all that well though so I may need to add some insulation around it. I set up a second shelf below that one for higher light plants but I don't have any of those yet so it's not currently plugged in.
 
  • #108
Looks nice, numbulan. So did you use mahjong to hold down the ends while the adhesive is drying? Does it stay after it dried? I don't know well about this type of adhesive, but I wonder how strong it is. You are supposed to be able to remount it without breaking the module.
 
  • #109
I suppose I should have taken a picture. I placed the tiles right on the edge of the boards held down with rubber bands, being careful not to touch the LEDs with them, and left it for about 36 hours (I had to wait for the drivers anyway!) It seems to be holding since I did that. I wanted to avoid creating a hot spot due to the air gap on the ends.
 
  • #110
then what you use like material ( lamp, driver , etc ) ?

jeff
 
  • #112
Thanks Naoki, for putting your guide together. I also decided to build an LED fixture after seeing this thread. I was previously using 2x moonlight modules (like RSS uses) to try out LED technology. I was able to obtain nice coloring on a cephalotus, but they didn't seem powerful enough. I wasn't able to develop good coloring on mexican pings.

Like Nimbulan, I used the 2-foot modules. However, I decided to use heatsink tape instead of Fujik glue. The 20mm tape by Sywon was on Amazon for under $15. Since the bars are 18mm, you have 2mm of overhang, but it's easy to slice it off using an x-acto knife. I also did everything with wago connectors to keep the build solderless.

I decided to build a 2x2-foot and a 4x2-foot and connected the fixtures with flat aluminum bars.
kn7z7gHjLDGqVpaxauU-Ls_6btHJZjgPvD-sBSKgWUHcv7JCSGq86DNA_lCQFnJFk0rULN_XiwF7IcYQo_ifit_TM54m8Yms0g6h4IXRjOzcIl1oU1lBeruV_7kq4jHfSaTqELY7S-Y-5eJKXRXNuvig7X_L34SVm7va7AX14XoQDZfZe67AgPGRD5Uy0tk5BQEGJpwAiEn_fyny2q0a-Mdcna6ADw82ahUzDV4P-xXAyjyNbL7afxfsHGPKdrh2qBJaSZqdGlFVCYZ5Po4DTau0hAhllA9Wa7al0rlHBlh6XgxcjJgKxIFWotldRvMziVwc0Y23cq1UATW8e895xLzYTsZgOj9AhrlXS5rpTQqR0tj_W5m3Lbrad-T4iWnUSafbO2XzG8ng0CZitolAyrxvoyZajnFhGmwxks2qRKor7nJT98PDfsuEiHwJZVepAg65W3OaXne8JmBnvg5lH1-DFHABz80hierJ2Sv1UdTGId8hq88fYAw7DDeIbGCVWE4WDWTI8KgiMkimr540zKQgx9F3jxlxXMsPaFpb5m2lWnSBdWRdW4uYT6Gh1XgOaCVB25go0qkSPK_0b41Katt8HFblaDXr7xW7lj13_p88Pi2U-wIL=w2286-h1286-no


The 2x2 is powered by an APC-25-500 and the 4x4 is powered by LPC-60-1050. Based on kill-a-watt mater, the 2x2 appears to draw 26.1W and 360mA. I thought the driver should be driving the lights at 500mA. Could the length of the wiring is causing resistance and causing the drop?

Anyway, I'm still testing everything out and adjusting what plants I want under them but I'm very pleased by the lights so far. I'm not sure how far away to keep the lights--right now I'm at about 6-7 inches. No numbers on the 4x2 yet. Haven't really used it much yet.

3ugHdPYBGn7xX8QlftkKr126ZpdghuD-2j80OKu4zS9AeOI4NMe3HGngmMd5_NooRf973lLHJ2X5AEv6kIWs4CDVTZ8Qxe7udgA-XRkyagjNTF08BcNjzkyCKwqL3VW28N9-DyQQBj1B2LM4k00357uoK695vxL2-7x6uv8W18-sW4kduq13rB3YrpUXzkRpnJuyNwd-HyPJtUJ2ntIpfikwhJ_kGYGKMgqrpwwCkATNmNM8VNd8PO5r5zpcijyZ3h0OvGIkTNpSVDSAh_p3JBfI3VFHYZFJ6sivNQ6lyy9aZs-IDYikWtvJv9qsJjODD82WRJ3liOGQPb1ouWQN5t2gQIQ71f5-mBt5hrPZgG-Z8Dwx3Ru1xU5esVr6cIaYa8B-8Yn3hbAoWtA5lS6sFwTwTRfcEa-RmL-xtlnmKJRhoyuKugWbTJThu6WBPK3LhvAdF9Loj1Qtw_S4_lNdpPn0L7BgY8AdS5laWJRLvLme-klZndMlFbwgpvVI1V5mZ_YqJwADAOQzhspmeEvjj8yQ8qfUzUsArp5QkoJWyBSy5-Qxe30Qz6Q0xHI7wkEo6JxnqSBijlW0YHyoHM78QYfxtosnVZP6Kw50uM_nLU6QPbB2FqwhPT2hwH7FvT1d95Ch1ZzSBZ-4cQcGCBXNSECb6mk7Tfdr2aNsZ83xww=w2286-h1286-no
 
  • #113
I assembled another set of LED strips today and ran into an odd problem. the tube of Fujik thermal grease I purchased a year ago for my first set has started to separate, so I occasionally run into pockets of very thin, clear liquid (I'm assuming the glue) as I squeeze the tube out. Considering there's probably enough in these tubes for about 40 2' strips, I may have to buy a smaller one next time to avoid this problem.
 
  • #114
I'll be adding more of my electrical engineering later on, But for now I'll add some general cautions and tips for these setups.
#1 Chill out (literally)! Heat causes the lifespan of LED lights to shorten dramatically. If you want the longest life out of your led's go all out on heat sinking and depending on your system, a fan might be called for

#2 A little knowledge is more dangerous that none. Being misinformed may lead you to make very costly and even fatal mistakes, so don't plug it in until you have double checked the ratings! NEVER touch the LED's while they are on (they're hot anyways). Even if you measured only 5 volts across the LED with your meter, it's potentially live at 120V thanks to how these things work.

#3 Use Heat shrink to cover exposed wires. It's an easy procedure that will save you the unnecessary experience of an unpleasant tingling sensation and the resulting reflexes.

This isn't directed towards any individual in this thread, they're just some tips I think will help anyone who is considering building their own. More to come.
 
Last edited:
  • #115
I assembled another set of LED strips today and ran into an odd problem. the tube of Fujik thermal grease I purchased a year ago for my first set has started to separate, so I occasionally run into pockets of very thin, clear liquid (I'm assuming the glue) as I squeeze the tube out. Considering there's probably enough in these tubes for about 40 2' strips, I may have to buy a smaller one next time to avoid this problem.

Good to know. Was it still usable?

Which LEDs did you use? I haven't done a comparison, but I believe Bridgelux came out with Gen2 of EB series strips late last year.

As a related note, at the bottom of this page, Bridgelux mentions the comparison of photon efficiency (micromol/J). I'm not completely sure if it is limited to PAR range (400-700nm). But LED emission is mostly within PAR, so the value should be pretty close. The value reported there is using an unrealistic condition (case temperature, Tc, of 25C and pulsed), so the real efficiency is lower than that. Lower CRI is slightly better in terms of efficiency. Higher K is very slightly better. This qualitatively matches with my calculation from a couple years ago. Another cool thing is that their newer simulation tool (excel macro) contains PPF and photon efficiency (for Tc=25C and 85C). Simulator can be downloaded from here: https://www.bridgelux.com/product-simulator
 
  • #116
The glue is still usable, I just used a tissue to wipe up anything that was too clear/thin. I'm just hoping it doesn't degrade any further.

I'm using the rest of my H-series strips (I think we talked about it on Facebook briefly, I got these really cheap.) I'll have to compare products next time I find myself in need of lights to see what the best option is.
 
  • #117
Finally got a chance to pick this project back up...

Looks like I'll need to double check my math and set up. The wiring isn't complicated, so I'm not sure where I goofed. Power supply should be sufficient to supply voltage/amperage for each of my 4 COBs to light up in series, but only two in a series will work at a time. This is even with the dimmer jumpered, so it's not just a matter of excessive resistance there.

Must be something generating more resistance than anticipated, or I goofed on some math. Hopefully I'll get a chance to figure it out this weekend and can post some pictures. Funny to see how much things have moved on since I bought my parts, but I'm still looking forward to getting a setup going.

Great to see that this thread is still around and semi-active!
 
  • #118
Could you provide some details about what driver and COBs you bought?
 
  • #119
Had a few minutes to review my purchasing choices from like 2 years ago. haha Looks like I managed to mismatch my driver and COBs by accidentally purchasing the 72V version of the COB. I'd meant to get the 36V version, which would have worked.

So, I've got an HLG-185H-C1400B CC driver (143V, 1400mA out) and 4 Cree CXB3590s (72V version).

So, looks like I'm left with a couple options:
1) So what the math says. I know I can run two COBs, but I need to see if it'd be dumb or super inefficient. I'll have to check if I can get away running them in two parallel series of two COBs, at the cost of ~half my amperage.
2) Buy something. Either new COBs or a new driver are both pretty expensive. A single driver for all four COBs (like the HLG-480H-C1400) starts to get expensive and starts to get into pretty high voltages. I work with 5-10kV stuff routinely, but I've got a few more safeties built in there...
3) Other clever suggestions?
 
  • #120
Running two sets in parallel will be equivalent to running 4x 36V models in series, so it won't be any different than if you had bought the intended COBs. It looks like you'll get around 70% of the rated output at that current level, and of course will run more efficiently than at higher current.
 
Back
Top