DIY Lighting Canopy, thoughts after 6 years

[BradB]

I've ran a DIY 4'x4' canopy over my reef for 6 years. I haven't seen much on this lately - DIY was big back then, but it seems to be replaced by commercial fixtures as they got cheaper.

Some thoughts on LEDs:

1) It is very difficult to assess the output or spectrum of LEDs. My PAR meter doesn't handle short wavelengths very well, and different LEDs have different spreads.
2) There is a ton of cheap high wattage LEDs at least advertised in the correct spectrum.
3) The best royal LEDs are expensive, discontinued (Luxeon K16), and give the tank a purplish tone. Right now I've shifted to cheaper generic 400nm or 470nm chips, but really don't know how well that works.
4)Bridgelux Vero led's are excellent for high power whites.
5) More expensive, branded, lower power LEDs (3 watt Crees, etc) are impractical for a setup this size.
6) Lenses haven't worked well for me. They obscure some light, are more work to mount, and one more thing that can go wrong. I prefer having nothing between my LEDs and the water.

Some thoughts on Drivers:
1) The LDDs were a huge improvement over trying to match individual drivers to LED strands. But they are a pain to connect. I have a wire nuts connecting each LDD to my main power supply from drivers located far away. I could solder this, but it is a pain to change if something fails later, and neither solder nor wire nuts hold up very well to salt spray.
2) Similarly, I use terminal blocks to connect the LDDs to wires going to each LED or circuit of LEDs. Sill a pain and they don't last, but 1-1 connections are easier.
3) I have yet to set up a working dimmer for more than 1 week. This isn't hard to do, but I always have something more important that needs done.
4) I have exactly 640 watts of capacity from my main drivers. Without replacing these (which aren't cheap) - running less means loss of efficiency and running more is impossible.

Some other thoughts:
1) I have 2 2'x4' aluminum panels for a canopy lid. They dissipate heat well, and were my only cooling for a long time. I added a CPU fan which seems to extend LED life.
2) The panels are a pain to remove, which is the only option if a LED fails. I remove them each about once a year. Losing 1 LED means losing 10-30 watts of light, and if it is a white light, is a very noticeable detraction from the look of the tank.
3) LEDs are attached using discontinued thermal adhesive. I have plans to use thermal film and screws to mount in the future, but this breaks the clean look of the top of the canopy and I have no idea how well it will work.
4) If I had to do it again, I'd use easier to remove, modular panels with heatsinks.

Compared to commercial fixtures:
1) I really have no idea if what is lower cost long term, considering LED replacements and electricity.
2) It certainly is more flexible as far as spectrum choice.
3) The initial is cost is a small fraction, roughly $400 vs $3000.
4) Mounting commercial units inside a canopy is not a trivial job.

Compared to free handing fixtures:
1) I want to avoid damaging my ceiling.
2) The light-bleed into the room takes away from the tank IMO.
3) The humidity from a large tank is not insignificant, and the canopy cuts this down significantly.

 

[Ron Reefman]

Some interesting points and a couple I'm not 100% sure I agree with, but that's more personal preference than anything. I've been using led fixtures for 7+ years now.

 

[Kampo]

Luxeon K16s are still readilly avail and cheap from china. trick with avoiding the purple look is making sure you balance it. using lime or mint leds in place of all or some of your whites, with a bit of cyan and 480nm leds will do this. a simple array based arround a k16

1xk16
3xMint/Sunplus Whites
1xBlue
1xCyan (not nessesary if you use sunplus whites, since they have good cyan coverage)
2-4 Violets

then run as many of those clusters as your tank needs.

 

[BradB]

Cheap leads labeled as Luxeon K16s are available from China I am suspicious whether they are authentic, but I have no way to measure their output.

I've found tiny leds like cyan and mint are a waste on this size tank.

 

[Kampo]

Several of us have them on here. There real and amazing

 

[theatrus]

Cheap leads labeled as Luxeon K16s are available from China I am suspicious whether they are authentic, but I have no way to measure their output.

I've found tiny leds like cyan and mint are a waste on this size tank.

Radiant flux is radiant flux, emitter size is largely irrelevant unless you are matching to secondary optics. I also wouldn't want to try assembling a whole rig using single star board LEDs (I did it once on a nano and ... the process sucks). The LED die used in the K16 shares the same manufacturing setup as the rest of the Lumileds line, so a Luxeon T is a good size substitute. I stick to C and Z footprints since it gives a ton of mounting flexibility for different layouts across everything except the violets.

 

[stefanm]

I run 3 k16's from China, now they're even cheaper, so cheap that it's almost free!

If you don't like the too blue look they're easy to balance out with Vero 10's 4000k, luxeon mints look like a good idea, wouldn't mind swapping my whites out for some of those.

 

[Lingwendil]

I really like the cheap K16 boards that are now ubiquitous on the ebay and aliexpress tyoe sites. They are surplus for a Toshiba DLP projector of some sort, and are the real deal. Affordable and nice quality.


Okay, prepare yourself for a lengthy reply to the OP and the points made- keep in mind I build a LOT of fixtures, and have installed a lot of them over the years. I find that certain things stand out, so I'll address them directly in separate replies. I've fixed enough homebuilds and "other peoples' work to find a few things worth paying attention to, as to not make unnecessary compromises or mistakes.

3) The best royal LEDs are expensive, discontinued (Luxeon K16), and give the tank a purplish tone. Right now I've shifted to cheaper generic 400nm or 470nm chips, but really don't know how well that works.

To cut down on the purple you need to offset the red- cyan and cool blue help, as does lime or mint. Using a higher CRI, higher kelvin white helps too.

I find longevity and efficiency on the generic "multichip" Epistar types to be rather poor. I tend to replace them after a short time due to staining of the internal surface where the individual dies are mounted, and loss of output. Yoo really need to under-drive them to get life out of them.

These are also a very, very nice and powerful Royal Blue option for larger builds, current production, and Citizen makes some very nice LEDs-

https://www.rapidled.com/citizen-royal-blue-cob-clu048-1212cf-b455/

Some other affordable options of reasonable quality are the Radion "clone" boards that have become more common over the last few years, as well as the really nice stuff Theatrus is working on for his "LEDbricks" projects.

4)Bridgelux Vero led's are excellent for high power whites.

Agreed. They are very nice for building larger clusters, or even as a base for freshwater builds... Affordable too. There are more expensive options with better color rendering these days (I have a pair of 4000k Citizen CLU-028 CoBs that are 97CRI that cost about ~$7 each, The Luxeon "fresh fish" line is another good one) but the Vero series have a certain mojo that's hard to beat, and the pico-harness connector is handy.

5) More expensive, branded, lower power LEDs (3 watt Crees, etc) are impractical for a setup this size.

Somewhat agree. It gets pretty complicated rather quickly on larger setups. That's one of the reasons I'm a fan of higher CRI base whites, as they require less supporting color to get the right "look" down easily.

6) Lenses haven't worked well for me. They obscure some light, are more work to mount, and one more thing that can go wrong. I prefer having nothing between my LEDs and the water.

I'm personally not all that big of a fan of lenses either. I do like to use a splash shield however. In my opinion a splash shield or some sort of waterproof/conformal coating is mandatory, otherwise you will encounter easily preventable issues down the road.

Some thoughts on Drivers:
1) The LDDs were a huge improvement over trying to match individual drivers to LED strands. But they are a pain to connect. I have a wire nuts connecting each LDD to my main power supply from drivers located far away. I could solder this, but it is a pain to change if something fails later, and neither solder nor wire nuts hold up very well to salt spray.
2) Similarly, I use terminal blocks to connect the LDDs to wires going to each LED or circuit of LEDs. Sill a pain and they don't last, but 1-1 connections are easier.

I like to use PCBs for them, and I like to run terminal/barrier strips near the LED arrays for easy wiring and maintenance. I find them easier to use than "brick" drivers in all honesty. Sockets on the PCB make swapping drivers take less than a second.

3) I have yet to set up a working dimmer for more than 1 week. This isn't hard to do, but I always have something more important that needs done.

Arduino is a harsh mistress, but I finally found a very nice 16-channel design over on Ultimate-Reef, it works very nicely. Bluefish controllers or the TC420 are other easy alternatives. Adapters are available to use common stuff like the Apex to control them as well.

Some other thoughts:
1) I have 2 2'x4' aluminum panels for a canopy lid. They dissipate heat well, and were my only cooling for a long time. I added a CPU fan which seems to extend LED life.
2) The panels are a pain to remove, which is the only option if a LED fails. I remove them each about once a year. Losing 1 LED means losing 10-30 watts of light, and if it is a white light, is a very noticeable detraction from the look of the tank.
3) LEDs are attached using discontinued thermal adhesive. I have plans to use thermal film and screws to mount in the future, but this breaks the clean look of the top of the canopy and I have no idea how well it will work.
4) If I had to do it again, I'd use easier to remove, modular panels with heatsinks.

In my opinion/experience, good heatsinking is not an option, but a requirement for these builds. At the very least run a very good fan with a contingency plan for if it fails. Large fanless heatsinks with good ventilation are my preference, and they are usually trouble free when specced correctly. Not every setup can handle multiple large heatsinks, however. I also don't musch care for thermal adhesive- Use a good quality thermal compound, with a good solid retention system, most often screws. Drilling and tapping is a pain, but honestly the best way if you really look at it. On a related note- I really wish heatsinkUSA still offered the T-slot profile heatsinks. They were perfect for this.

As far as removing panels, well, yeah- it's not the most fun thing to do. Having quick disconnects of some sort for power, dimming, and a quick release of some sort for the hanging helps a lot. I became a big fan of through pins with retainers on either end, using aluminum 8020 framing to make it easy to adjust or modify.

Compared to commercial fixtures:
1) I really have no idea if what is lower cost long term, considering LED replacements and electricity.
2) It certainly is more flexible as far as spectrum choice.
3) The initial is cost is a small fraction, roughly $400 vs $3000.
4) Mounting commercial units inside a canopy is not a trivial job.

LED replacement cost is very low on the fixtures I have built/installed over the years. I can count on one hand the number of LED failures I have had in the couple dozen installations I've done, and a couple of those were from fixtures falling in the water, and the subsequent corrosion damage after being dried and reinstalled by the owner. The others were from a fan failure roasting older violet LEDs, and one from a overvoltage condition from a failing "maxwellen" chinese driver. I design conservatively, wether large or small build- so take that for what you will. The largest setup I did was for a private frag/fishroom, and we had maybe fifteen fixtures in that installation alone, and It still saved the couple 60% less cost than if they had gone commercial/consumer fixtures. four years later there hasn't been a single problem with their setup, other than when drywall anchors pulled out of the ceiling (not my installation work, I swear!!!) and dropped a 4' MakersLED into the their LPS bin, and the subsequent damage to a handful of LEDs.

Compared to free handing fixtures:
1) I want to avoid damaging my ceiling.
2) The light-bleed into the room takes away from the tank IMO.
3) The humidity from a large tank is not insignificant, and the canopy cuts this down significantly.

Light spill is pretty distracting. I like canopies, and it makes it easy to oversize fixtures for cooling with affordable large heatsinks. Having a cover glass that is in the canopy closer to the tank, with the fixture suspended above and self contained is a handy way to not have to worry about waterproofing, so long as you have ventilation in the upper portion so you don't roast the LEDs.

 

[BradB]

A glass cover is an interesting idea. Have you done this? The problem is strength to weight, if you don't suspend from the ceiling. Keeping it clean is also a challenge. But this solves all corrosion, humidity and risk of metal contaminating your tank.

Some day I will have a house I know I will be in a long time and do a built in wall setup. For now I don't want to use the ceiling.

 

[Lingwendil]

You don't need a cover over the entire tank opening, doing it at each array works well, and you can use comparatively thin glass or acrylic to make the weight negligible. 1/16" acrylic about 3-4" from the arrays works well if supported properly. Stainless steel hardware on some aluminum standoffs to support the acrylic, on some thicker 12-20" heatsink profiles from heatsinkUSA is my go-to solution, with a 120mm or two PC fan setup on top. have the acrylic overhang the edges of the heatsink 2-3" in each direction to prevent splash from coming in sideways. Cheap and cheerful, easy to work on if maintenance is required, and some holes drilled in the right places makes hanging the heatsink very easy.

Look into solutions using 8020 aluminum framing for suspension systems, they use it for many industrial applications that require some pretty hefty weight to be supported, it's even used in the construction of aquarium stands. It's very easy to work with and quite versatile.

 

[BradB]

Covering just the array takes me back to building something like a commercial fixture. Not a bad idea, but not a new idea, and integrating it into a larger canopy to block light and humidity, allow easy access and disassembly, and isn't too heavy for the tank itself to support is a challenge.

How are you connecting your ldds?

 

[Lingwendil]

CAT5/6 cable, one twisted pair per driver, for 700mA-1A strings. Use plenum rated stuff as it stands up to heat better (it's rated for use in HVAC spaces) and you can conveniently run four LDDs per cable, with handy color coding. I use PCBs for the LDDs with socket strips to allow for easy driver swaps. Usually four LDDs per PCB, with a large Meanwell DC power supply, from 24-48 volts, depending on application. It's way better than running individual pairs to the LDDs.

 

[BradB]

Clever, but isn't it too thin? Does it get hot or measure voltage drop or current use? Are your ldds remote or near your leds?

 

[theatrus]

Clever, but isn't it too thin? Does it get hot or measure voltage drop or current use? Are your ldds remote or near your leds?

Depends on the cable. You can go a reasonable distance on 24awg CAT5 at 1A. The 30AWG stuff isn’t useful here.

 

[Lingwendil]

I only ever need a few feet, so voltage drop/heat isn't an issue with 24/26AWG. The PCB get's mounted nearby in a box, or on top of the heatsink assembly, depending on the build. Common solution is to have several heatsinks bolted to a rack made of aluminum angle or U-channel, with the junction boxes also mounted to the rails. The outputs of the boards usually feature some sort of solderless connector (poke-in or screw terminals) with screw terminals being my preferred method. No wire-nuts, no expensive plugs or wire harnesses.

 

[BradB]

Can you go into more details about " I use PCBs for the LDDs with socket strips to allow for easy driver swaps"? Is this PCB a commercial, off the shelf product? Or do you have a way to print your own? Where do I get the "socket strips"?

 

[Lingwendil]

Yeah, they are commercially available- plenty of variations, Coralux, Steve'sLEDs, RapidLED, etc.

https://www.stevesleds.com/LDD-Driver-Array--8X-Drivers_p_306.html

https://www.rapidled.com/ldd-h-4-driver-board/

http://coralux.net/?wpsc-product=ldd-5-driver-board

For what it's worth the RapidLED and Steve's versions are what I have the most experience with, and I'm working on designs for my own versions as well.

 

[garbled]

I actually have pretty good luck with the epistar/epileds style COB's on ebay. I have one set of 5 20w chips that has been running continuously for over 3-4 years with passive heatsinks. All of my COB failures to date have been due to improper cooling. Mostly non-passive heatsinks that had fans that died. I feel you can even overdrive them if you have sufficient cooling. I'm not discounting your experience, just noting mine.

As far as lenses go, I definitely prefer them. I had alot of pain figuring out how to mount them though, and had many failures until I figured it out. The method I currently use, which seems to be working great, is in my 100w LED article. I 3d-print a mounting ring, and then bolt the ring to the sink. This avoids the use of the thermal paste (which eventually fails), and if you can find a sink that has holes on it somewhere, you just need to print the ring with the holes to match, and then you don't have to drill/tap. Basically the ring puts even pressure on the lens, which puts even pressure on the COB, and then my mating problems all go away.

I prefer to use quick disconnect connectors for everything, just so if I do get a failure, I just unsnap the connector, solder in a new LED, and snap/screw it all back together. I could probably swap an LED in about 15 minutes, not counting shipping time... I should buy some spares. :)

 

[User1]

Yeah, they are commercially available- plenty of variations, Coralux, Steve'sLEDs, RapidLED, etc.

https://www.stevesleds.com/LDD-Driver-Array--8X-Drivers_p_306.html

https://www.rapidled.com/ldd-h-4-driver-board/

http://coralux.net/?wpsc-product=ldd-5-driver-board

For what it's worth the RapidLED and Steve's versions are what I have the most experience with, and I'm working on designs for my own versions as well.

I have used Steve's 8 LDD boards and while it works well the wire holders are utter crap. Granted maybe you shouldn't have to remove wires but for those of us who do due to error or testing the push release things are just garbage. I am very glad I was able to switch to @theatrus drivers which are 10x more friendly not to mention less wires.

Not bashing but just saying those boards are a bit frustrating to work with and way too many wires for me.

 

[BradB]

So you have large meanwells, run ethernet from them to Steve's Led style (his were the first I looked at, if not the first available) and then hookup wire to your Leds?

The reason I never tried Steve's is I would need to locate them inside the canopy. They don't look that robust and are a little expensive to keep replacing if they are constantly exposed to salt water. I could locate them remotely, but then have a mess of wires to the canopy.