garage lighting project featuring Nichia 219's [wip]

I'm remodeling my garage and turning it into a shop. The lights I intend to install will be custom made arrays of Nichia "NVSL219AT-H1 4500K, 92 CRI, B10 Bin" LED's from Illumination Supply, powered by a 320w meanwell driver, hooked to a dimmer. The overall output of the luminaries will approach 20,000 lumens, and should achieve a lux of over 1,000 in this space at full brightness.

EDIT: Changed to 180 LED's and three drivers. 60 LED's per driver in 12s5p configuration, driven at 1340ma each. At 70% optical efficiency this will be approximately 40,000 lumens. Lux should be about 640. My original Lux calculation is incorrect. /End Edit

I am worried about glare. I think I will need to diffuse the led's to some extent. LED downlights in stores that I've seen are recessed cans and have a diffuser over the leds. There is no secondary optic besides the can baffle. A led with a diffuser in front of it is going to output 120 degrees and beyond, but I assume this would be ok considering other types of light sources do this and it isn't so bad. I have ordered some ledil TIR optics which are 26mm in diameter and appear to have a somewhat diffused face. I also have some ledil reflectors to try but these are not diffused. I went with wide viewing angles such as 50 degrees FWHM.

I am wondering if I should skip optics and just put a diffuser directly in front of the LEDS. I would like to mount the lights on the outside of the drywall rather than recessing them, for ease of installation and better heatsinking.

Another option is to make really mini cans with the ledil reflectors I got, and to put a diffuser film over the reflector.... I think this may be what I end up doing, barring any good suggestions.






Next will be tape and mud....

Hoop said:

...

...I am worried about glare. I think I will need to diffuse the led's to some extent..

Click to expand...

You're wise to be concerned about glare. Also shadows. I know this is not at all what you want to hear, but with that beautiful (which I assume will eventually be painted white) drywall ceiling, I'd seriously consider using the LEDs as indirect sources, non-diffused, pointed up at the ceiling.

Glare + Shadows = seriously bad lighting in a work space. I've seen it. Your hand or head tends to cast a shadow on whatever you're working on...

Ahh, shadows would be the enemy. It may help that I am going to use 70 individual Nichia's spaced throughout the ceiling. (14s5p) With this many emitters I think there should be no shadows.

Edit: There is also the option of using "led profiles" with a diffuser cover: LINK, LINK-2 (or without a diffuser, facing up at the ceiling, as per Jim's suggestion)

Hoop said:

I'm remodeling my garage and turning it into a shop. The lights I intend to install will be custom made arrays of nichia 219's, powered by a 320w meanwell driver, hooked to a dimmer. The overall output of the luminaries will approach 20,000 lumens, and should achieve a lux of over 1,000 in this space at full brightness.[/IMG]

Click to expand...

That garage looks to be 20*20' if not 20' * 25'? 20' * 25' = 46 square metres.

With 20,000 lumens, the best you can hope for is 430 lux average, and with losses from absorption, etc. count on it being less.

An evenly spaced 14s5p arrangement may be a lot of work as you will need a heat sink for every LED and then comes the issue of open connections and what you are going to do with them (and any potential fire hazards). That power supply has some pretty good oomph so a bad connection could start a fire. Just something to keep in mind. You may want to consider something easier to wire that has some inherent protection.


Semiman

Ahh, I must've recalled my math wrong. The space is about 60m², so I will only be seeing a lux of maybe 300, unless I double or triple the amount of emitters. MUHAHAH!

All wiring connections will be properly enclosed and isolated. This is an ambitious project for sure, and it might be easier to use some sharp mega zenigata's rather than a whole bunch of nichia's, but I like the idea of a more uniform spread of a whole lot of emitters.

Edit: I may go with diffused dome covers and make my own heatsinks out of 1.75"x.25" aluminum bar stock.

Edit2: Jim's suggestion of aiming at the ceiling has its advantages. For one it avoids any secondary optics and their associated cost. It means all I have to do is mount a bunch of 219's onto some aluminum bar stock and attach it to the ceiling facing up. Hopefully glare would be insignificant with this approach, and the light would be completely diffused. Cree uses a similar approach in their low bay lights. Their fixtures resemble a fluorescent shop light fixture, with the leds facing up into the housing. I assume there will be significant losses with this approach, but it will be much easier to implement so more emitters won't add much to the project complexity.

Ok, I have decided I want one of two things, or possibly a combination of both, and there is one product that will let me do either. LED light strip housings will allow me to diffuse and aim the leds down, or non diffuse and aim up at the ceiling. The "Klus" profiles are a bit expensive, and most of the diffusers are flat and the housings somewhat deep. A novel solution is empty T10 led tube housings, which have a cylindrical diffuser and are inexpensive.

Figuring things out more; the array will consist of 15, 1200mm (4ft) T10 led tube housings. Each will contain 12 Nichia 219's, spaced evenly in the tube, three per foot. This will be a total of 180 leds, 60 per driver in 12s5p configuration, driven at 1340ma each. At 70% optical efficiency this will be approximately 40,000 lumens, for a theoretical lux of 640. This is quite good I think, as I've decided that if I really need more light than that over a workbench or something I will add a fixture above that bench. Total system power will be approximately 930 watts. (180x4.7 /.95)

Hoop said:

Figuring things out more; the array will consist of 15, 1200mm (4ft) T10 led tube housings. Each will contain 12 Nichia 219's, spaced evenly in the tube, three per foot. This will be a total of 180 leds, 60 per driver in 12s5p configuration, driven at 1340mah each. At 70% optical efficiency this will be approximately 40,000 lumens, for a theoretical lux of 640. This is quite good I think, as I've decided that if I really need more light than that over a workbench or something I will add a fixture above that bench. Total system power will be approximately 930 watts. (180x4.7 /.95)

Click to expand...

Went over the math quick and 40,000 lumens seems pretty reasonable.

What would cause me concern is the use of 1200MM T10 housings. These typically are running 15-20 watts, maybe a bit more. You are looking at almost 50 watt per housing. They may get hotter than what you want.

I am not sure what you are building in your workshop, but do you really need 90+ CRI? It certainly looks great, but you are taking a real efficiency and cost hit compared to what may be possible in 80CRI, 4000-4500K.

Semiman

Good points. Well, I am making my large residential garage into a small machine shop, and I am a tint snob, so, but since the leds are fully dimmable, powering them at 30% (400ma) would still give me 190 lux, and should give me over 70 lumens per watt even after the optical losses.(70% efficiency worst case) The driver itself is 95% claimed efficiency; one of the high end Meanwell drivers. I can always supplement the heatsinking capabilities of the profile by bolting on an aluminum strip so that the profile would then resemble an upside down T.

It seems that I cannot make purchases through aliexpress because of their escrow service, so the T10 housings are out... I will probably end up going with KLUS strip light housing, the GIP style, with a square diffuser.

Hoop said:

Good points. Well, I am making my large residential garage into a small machine shop, and I am a tint snob, so, but since the leds are fully dimmable, powering them at 30% (400ma) would still give me 190 lux, and should give me over 70 lumens per watt even after the optical losses.(70% efficiency worst case) The driver itself is 95% claimed efficiency; one of the high end Meanwell drivers. I can always supplement the heatsinking capabilities of the profile by bolting on an aluminum strip so that the profile would then resemble an upside down T.

It seems that I cannot make purchases through aliexpress because of their escrow service, so the T10 housings are out... I will probably end up going with KLUS strip light housing, the GIP style, with a square diffuser.

Click to expand...

I would be careful about Aliexpress ... have heard horror stories about shipping times. Obviously that does not apply to all suppliers.

It sounds like you know well what you are getting into. You should be quite happy with the results.

It's a shame the 4000K Citizen or Bridgelux high CRI modules are not easier to get.

Semiman

Mega Zenigata's are easier to get though and have a 92 cri 4000k version that puts out 5,000 lumens. I have one but have not utilized it for anything yet.

After calling the local metal shop, I will save a lot of money by using .375"x1.25" flat bar instead of the KLUS profiles, and this will also be a beefier heatsink with more surface area. I will use the KLUS square GIP diffusers and make the aluminum portion myself. Total cost on the metal bar stock will be $116. Total cost on those darn diffusers will be $285 unless I come up with a cheaper alternative.

SemiMan, could you comment on Jim's idea of aiming the leds up at the ceiling? If I aim the leds down, the ceiling will be relatively unlit. Aiming them up will result in a very bright ceiling, and well diffused light, but I do not know how much of an efficiency hit this would make. Worse than %70 on white paint? Is a very bright ceiling desirable?

Flat bar is near useless as a heat sink, especially once you put a diffuser on it that blocks one surface from diffusing and if you planned to make it flush, there goes the other side.

Go with U-channel, perhaps 2" * 1" * 1/8" ... 3/8" is overkill. The U-Channel would also give you a surface to mount a diffuser to .. or diffuser material and hence get your costs way down. Milky white diffuser loses a ton of light. Good clear diffusers are > 90% efficient. A u-channel also allows you to put an end on them so you have a place to mount a cable gland or connector to make for a cleaner design.

Anything that points upwards in a shop very quickly gets covered with dust so you will need to cover any LEDS. You can get highly reflective highly diffuse paint intended for the interior of signs. It is > 95% reflective ... till it gets covered in dust too.


Semiman

Ahh. All good points. I didn't think about dust or super reflective paint. As for the flat bar, I guess I am underestimating the heatsinking required. Since I plan on suspending the fixtures, with my current mock-up there would be a total of 69 in² of exposed aluminum per fixture, 7.6 in² per led, although only 29.25 in² would be vertical, 3.25 in² per led; the rest would be horizontal on the backside of the fixture, facing up at the ceiling, so probably less effective. U channel would work well, although extrusions such as this are a bit more expensive than bar stock or plate. I'll get a quote on some. I will be machining these either way, so the diffuser will have some dovetail or what not to grab on to. The frosty diffusers I've looked at are like %70 efficient. The only other option I have come up with is to get a clear lens and apply some diffuser film to it, but the diffuser films are either too expensive or block too much light as well. I am ok with a %70 efficient frosty diffuser if it does a great job of diffusing the light. At 400 ma I will still get 70 lumens per watt even with the diffuser losses.

Hoop said:

Ahh. All good points. I didn't think about dust or super reflective paint. As for the flat bar, I guess I am underestimating the heatsinking required. Since I plan on suspending the fixtures, with my current mock-up there would be a total of 69 in² of exposed aluminum per fixture, 7.6 in² per led, although only 29.25 in² would be vertical, 3.25 in² per led; the rest would be horizontal on the backside of the fixture, facing up at the ceiling, so probably less effective. U channel would work well, although extrusions such as this are a bit more expensive than bar stock or plate. I'll get a quote on some. I will be machining these either way, so the diffuser will have some dovetail or what not to grab on to. The frosty diffusers I've looked at are like %70 efficient. The only other option I have come up with is to get a clear lens and apply some diffuser film to it, but the diffuser films are either too expensive or block too much light as well. I am ok with a %70 efficient frosty diffuser if it does a great job of diffusing the light. At 400 ma I will still get 70 lumens per watt even with the diffuser losses.

Click to expand...

Looking at onlinemetals, 2" * 1" * 1/8" is about $9 for each 4' length assuming you buy 8' lengths. That seemed to be inline with your bar costs.

Just another option for you would be Fluorescent lights. They make a newer T5 bulb which produced about 5000 lumen per 4 foot bulb. its going to be a LOT cheaper. I too considered using LED's for my garage but it would have been much too costly. you can get 20,000 lumen for about $75 and they are about 110 lumen/watt

SemiMan said:

Looking at onlinemetals, 2" * 1" * 1/8" is about $9 for each 4' length assuming you buy 8' lengths. That seemed to be inline with your bar costs.

Click to expand...

Nice. This means it will be even cheaper at the local metal distributor. The profile of the diffuser doesn't lend itself to the U bar, but I will simply machine both the diffuser and the U bar. I will put a dovetail on both or something.

mds82 said:

Just another option for you would be Fluorescent lights. They make a newer T5 bulb which produced about 5000 lumen per 4 foot bulb. its going to be a LOT cheaper. I too considered using LED's for my garage but it would have been much too costly. you can get 20,000 lumen for about $75 and they are about 110 lumen/watt

Click to expand...

If I were sane, I probably would have gone with fluorescence, especially considering there are some 90+ CRI fluorescent bulbs these days. For some reason I am compelled to LED's, probably because the shop where I am employed uses fluorescent lighting, cheap ones, and I want a different lighting experience. Oh, I like the fully dimmable aspect of the LED's as well. At this point I have already purchased the led drivers. Total project cost will be somewhere around $2k. A plus side is that I may come up with a commercially viable product idea from this.

Edit: Ok I have purchased the drivers and the diffusers at this time. Total cost so far is 390.53 + 241.03 = $631.56

Soon I'll get to say: "do these led's make my mini-mill look big?"

Alrighty! It's about that time. I now have all of the main ingredients for this project: 180 nichia 219's, three Meanwell 320 watt drivers, thirty 19 inch lengths of 1"x3"x.125" aluminum rectangular tubing, and 15 of the 39.4" Klus differ profiles which I will cut in half. I decided to go with 3" tall aluminum rectangular tube rather than 2" to get even more surface area for heatsinking.









The fit of the diffuser to the aluminum extrusion is not ideal and there is not much aluminum to machine away for attachment purposes. I think that I will be able to rely on the end caps to hold the diffuser in place, and I can machine small channels in which the bottom portion of the diffuser will sit. If there is light leakage from any bow in the diffuser I will make a few hold down brackets which will be hidden inside the diffuser and spaced evenly along its length.



The diffuser profiles vary in their shape somewhat so the hold down mechanism needs to be either adjustable, or floating with spring tension. The above mock up uses dowel pins as guides for a small aluminum plate, and a thru screw for adjustment. The spring would just serve to keep the assembly taut when the screw is backed off.

EDIT: The easiest solution would be to simply drill through the diffuser in a couple spots and have a button head cap screw going thru into the aluminum extrusion. This would of course be visible but I think the ease of construction is worth it. I will go with stainless or nickel screws rather than black so they are less apparent.



Now what kind of guy would I be if I didn't include a picture of the thru screw?

It's funny that I always think of a complicated solution first and then the more simple ones after a bit more thought. I mocked up a hold down clamp which is shaped such that it would put significant down force on the star, but then I decided to just use screws in an alternating fashion from star to star.

Have you done any thermal analysis or measured a mock-up yet? Just curious how hot your tubing will get with whatever thermal load you're planning...