High-Power LEDs

[Amateur_expert]

Hello guys,

Does anyone have exp with LEDs?

I am well versed in LEDs but I lack a PAR meter. I currently have two 840 lumen (5,000K or 5,700K) set up 10 inches from the mount.

They put out about the same amount of light that is in the shade of a tree to the naked eye. I'm guessing the PAR readings would be similar to under that tree.

Does anyone have suggestions? Better LEDs? These will be on display in the living room so no just red/blue LEDs.

What are you running for LED lighting?

[naoki]

With 5000K LED, if you multiply your footcandles by 0.13-0.15, you get approximate PPFD (micromol/m^2/s). My measurement of this comes from Bridelux Vero 29 5000K, cheap eBay cool white 100W COB, Cree 18W daylight (5000K) household bulbs. If you use warmer warmer LED (3000K), you multiply the fc by 0.17-0.18 (from Cree CXA3070 3000K, and Cree Softwhite 9.5W household valve).

Here is the conversion factor between PPFD and fc:
Conversion - PPF to Foot-candles

Sun is giving 0.2 micromol/m^2/s for a given foot-candle. So it is similar to (warmer) LED, but more than LEDs. Perception of light intensity by your eyes is closer to foot-candle than PPFD (might be helpful when you are comparing the shade to your light). But eyes are amazing in the flexibility, and not a good gauge to guess the absolute values.

It is difficult to tell without knowing the exact LEDs (beam angle will influence the PPFD quite a bit). But Cree 9.5W household bulb is around 800 lumen. With this and 6.5" reflector, I get 39.5 micromol/m^2/s (20fc) @ 12". But this is from 3000K bulb. So with two of them, you are probably getting good light for Phals and Paphs.

If you want to maximize the efficiency, Cree CXB3070 (which became available recently) and Bridgelux Vero series (version 2.0, I like vero 29 for the ease of construction), which should be available in a month, are one of the easiest and best performance/cost choice. They bumped up the performance of warmer LEDs (3000K).

Here is a bit more info (a bit dated) about CXA3070 (previous version before CXB, message #18):
cheap ebay DIY LED

[Amateur_expert]

Great info. Thank you.

Here is a link to the datasheet
http://www.lumileds.com/uploads/354/DS103-pdf

The LED's are LXR7-SW50

The LED's do not have a culminator or any optics, they are just set 1/2' inside a heatsink. You can see from the picture where the light throw is.

I really do need a PAR meter. I don't know what my FC would be. I do not know how to get that without a meter.

Would those other LED's be better than what I've got running?

---------- Post added at 01:41 AM ---------- Previous post was at 01:36 AM ----------

I've seen those CREE LED's before. I've also seen huge ones similar to that. I might switch over to them eventually if mine truly will not work. I see there is a 3,000K but I'd also like to get to one that is a bit whiter. Futureelectronics.com has all of them.

[naoki]

I'm not familiar with this LED, but the spec. looks pretty good! Thank you for letting me know about this.

As you know, the efficiency of LED depends on the driving current. From the data sheet, the nominal spec is 700mA @ 11.2V, so 7.84W. It emits 1100 lumens, so 140lm/W. This is a pretty good efficiency. It is better than Cree & Bridgelux COB LEDs driven at the nominal current (Vero 29, 5000K is 130lm/W @ 77.3W), but most of us drive these a bit softer to get higher efficiency.

The price of lumiled is quite a bit more expensive ($10 for 8W vs Vero 29 $30 for 50-80W). With Vero 29, 50W driver can be around $12, and the heatsink is about $10, so the total cost is around $60 for 50W (I need about 200W to cover 4'x4' area). Also it is much easier to use a big COB LED than using bunch of smaller ones. But if you don't have to cover a wide area, yours looks like a very nice one. Can you tell me what other components you are using (driver, heatsink, and the gooseneck thing)? Is passive cooling enough?

With that efficiency, you are probably getting quite a good amount of light. But it may or may not be enough for the Cattleya. You mentioned 700lm, which suggests that you are driving it at below the nominal 700mA, is it around 5W?

You probably know it, but even the cheap Apogee PAR sensor costs around $150 (you can make a measurement by connecting it to a digital multi-meter). Or USB interfaced version is about $200 (BioTek Marine BTM-3000. It is fun to play with PAR meter, but a cheap light meter (something like this) seems to be good enough if you are using white light.

[Amateur_expert]

Hey man,

The size v performance is why I originally got this LED for my wedding. I had to run a string of them in series for the colors we were looking for. It worked well, purple room with pink corners.

The set-up isn't terribly expensive if you mount them to the stars yourself like I did. It turns out that is easier than I thought it was going to be.

I know what you mean about the COB being easier than a bunch of smaller LED's. I already had a bunch of the LED's to use though. They also fit into nifty looking fixtures easier that a 1.5" wide LED (w/o heatsink).

So on to the meat of the situation. I run them at 700ma so the output should be spec minus heat dropoff. At 1,000ma, they will put out a high lumen count, but I didn't have the driver for that already - just lots of 700ma ones. The driving I'm using is a EUC-026S070DS (EUC-026S070DS | EUC Series 26 W 0.7 A 37 V Dimmable Outdoor Constant Current LED Driver Module | INVENTRONICS - Future Electronics).

It has plenty of juice to run 4 of these LED's and it's dimmable if needed though I didn't hook that up. The dimmer will work fine with a 10K ohm potentiometer (I'll have to spec that if you're interested). This puck does get warm though. I mounted a 2.5"x2.5"x1" heatsink to it using thermal epoxy and that is plenty sufficient, even being mounted on all sides except the heatsink surface.

The LED's on the other hand are a bit hotter. The lamps they are mounted it were made for a 300ma LED and could handle that heat just fine. I drilled eight 1/8" air channels into the housings. They work well, but the housings still run hot. I'd rather have active air cooling on them, but alas, no room. Plus an active cooling setup wont fit into anything but a large fixture that takes away the modern sleek look I was going for. So if these burn out in a year or two I will just solder in a new set and maybe they'll have a better housing I can work with then that will accomplish all my requirements.

As far as using a bunch of these and the costs of heats sinks individually etc, here's what I would do (and I'm still considering doing this for my seed racks) and how I'd set it up. If looks aren't your thing, I'd take as many of these LED's that my big powerpack can handle (The Xitanium can go up to 210V @ 700mA, so it can handle 18 of these white LED's and many more if they are blue or red []LEDINTA0700C210FO | Xitanium 150 W 0.7 A 210 V UL Class I Outdoor LED Driver Module | PHILIPS LIGHTING - Future Electronics).

I would take that power pack (which I would actively cool), get myself a length of square or aluminum tubing (1/8" wall x at least 2"x2" - bigger is better), mount all LED's to the tube using thermal epoxy or thermal paste and screws, then I would use my Vortek Tornado 92mm computer fan to run mass air through the tube - push or pull - though pushing may be better for the life of the fan. If you're willing to get into liquid cooling that may be more efficient. If you need a more concentrated light I would do the same thing essentially. But instead of a tube I would use a round sheet of aluminum (The size you need) then mount the tornado fan blowing towards the upper side of the disk. Adding pin heatsinks would be beneficial to remove heat faster - home made would be 4-32 x 1 or 2 inch countersunk machine head screws to act as the pins. It would take a few hours to build, but at the end, it would be substantially cheaper overall. Fixed cost being around $75 if you already have the thermal epoxy and tools to make holes for the screws (McMaster-Carr). With a round setup like that, you could always braze or even use thermal epoxy to put a full coil of the bendable 1/4 refrigerator water supply line and run water through it. That would be effective. Computer water cooling radiators aren't terribly expensive and water pumps are around $15-$20 on McMaster or any other reputable website.

I did see that apogee PAR meter, but I really want that big one. The Li-Cor LI-250A meter and the LI-190SA probe.


I'm not sure how important it is, but I always use silver/tin solder on every joint - never just a screw on terminal. I think that improves efficiency over an array with less resistance.

They do have higher volt powerpacks, but the 210 is what I used and know works well. It does get warm though and at the wedding I had all power packs mounted on the interior walls of a tube - no heatsinks - with a 110cfm 120mm fan blowing air over everything - kept everything working flawlessly.

[naoki]

Which method did you use to mount LED to the MCPCB star? Heat gun, kitchen oven, or something like that? I don't have enough skill, so I usually use simpler DIY like COB or array. But it is a handy skill. And the scattered light source with smaller LEDs (instead of a COB LED) has an advantage to provide more homogeneous light, which can reach to the lower leaves (less shadow effect).

For our purpose, where we run the light for a long time (>12h per day), it is usually a bit more economical to run at the lower current (like 700mA, which you do, instead of 1A). Especially with this lumiled, under-driving seems to be pretty beneficial (in terms of efficiency, I didn't do the calculation, but just looking at the data sheet).

Using the aluminum screws to make the heatsink is an interesting idea, which I haven't thought of. It sounds quite tedious, though.

I've seen others used liquid cooling of LEDs. It would be nice for the enclosure for cool orchids (if the light is inside). I had thought of it, but it does become a bit more expensive. Do you know the reliability of the water pump?

If you can afford, Li-Cor LI190SA (mine is this model) is good. This class of PAR meter (newer Skye and Kipp&Zonen perform similarly well) are all > $1000. Also, if you want to keep it accurate, you need to send it for calibration every or every other year ($150 from Li-Cor).

[Amateur_expert]

Hello!

So surprisingly enough, it wasn't terribly hard to mount LEDs to stars. I took the star, put a lot of flux on the copper contact points (on the star but for the LED - not the wire contact points) and then just whipped the tip of my soldering irons over it with the silver/tin solder. It doesn't stick to the anodizing - just copper contact points. Once that's done, I let it cool and put more flux on it. Then I put the LED centered and correctly on the star. The LED is kinda stuck there by the flux - so you don't need to hold it - and you shouldn't or it will mount wrong.

Then I put the whole star on a hot plate. The flux becomes liquid before the solder, but its still a thick liquid that keeps the LED centered. Once the solder melts the LED will look like it got sucked down 1/32nd of an inch and youll see it clearly. Then I take (with tweezers) the whole star and put it into water gently so it doesn't knock off the LED but it cools quickly so it doesn't injure the LED via heat. This sound complex and hard, but I made over a hundred of these for my wedding and it took half a day. Also, remember, capillary action will put the solder where it needs to be without spilling over between the LED and the star board. It's a lot easier than it sounds. If you mess up, just reheat the star and remove the LED with tweezers.

The screws are a bit tedious, but in reality, a heatsink that big will cost a lot and a few hours to build one is tedious, but MUCH cheaper. And it'll work.

So here's the thing with liquid cooling. it can be very simple and cheap up to extreme like water to water cooling, chill units, submerged systems etc.

But at a bare minimum that would work this would be what you could do. Take 1/8" copper refrigerator tubing and attach to the stars. Then take that same tubing and bend it into a lattice with a few computer fans. Then connect it to a water pump. Then from the water pump to the other free end of the tubing. Don't forget to fill it with water!! I'd install a bubble trap so your motor doesn't cavitate and die, but you don't 100% need to.

Little Giant pumps are good. Get one with a 100% duty cycle. High head and high GPM mean more pressure and more water respectively. So I'd get as big as I could afford. The aquarium of fountain circulation pumps are good. I've used a Little Giant blue pump for over a year on a demanding system and it never failed once.

100ft of 1/8 copper tubing from homedepot is $50. The pump is $75-$100. Computer fan is $10 and a power inverter is free (Prolly have one for an old phone or something - just look at the output volts and make sure the mA are high enough to run the fan.

I want a good PAR meter. Not a junky lil one.....