At one point when I was feeling overwhelmed with my own circuit design and wiring schematics, it seemed like the cycle of stripping wire, tinning, soldering, screwing, and testing would never end. Now it's all done and I have my evenings back, and here is the story.
Introduction:
Somewhere around October of last year I made the decision to upgrade my stock lighting. Both of the Oceanic 36W ballasts had died, I was having wiring problems within the hood, and I was due for new bulbs. Instead of shelling out $40-50 for the same old mediocre 72 watts of PC, upgrading became attractive because I could forgo the new bulbs and absorb the cost into the total price of the project.
I looked at the Nanotuners 4x36 PC and LED kits, RapidLED kits, and 3rd party fixtures of all types. The 3rd party fixtures left an open top and without an automatic top-off system, keeping up with that rate of evaporation in a tank this small is prohibitively difficult. It was obvious that keeping the fitted stock hood in place was the way to go. That left only a few options. An LED retrofit seemed the most fun, effective, and efficient, even if it was not the cheapest solution in terms of up-front cost.
I'll skip ahead and mention here that after weeks of shopping around, I went with RapidLED for a lot of the parts. As one of the few places that sell pre-milled heatsinks for projects like this, it wasn’t a difficult choice. Their customer service is top notch, the prices are great, and I stand by them 100%.
Anyway, so began three or four months of reading threads, reviews, comparing prices, calculating, visio drawings, and then MORE reading threads. I think I probably visited every page of every LED retrofit and DIY thread on the Internet at least once. A wise man learns from the mistakes of others, I say, and so I did my best to avoid many of the pitfalls and challenges that beset fellow hobbyists taking the plunge into this type of project.
A side note: Instead of making this into a lengthy step-by-step how-to, which I had originally intended, I figured I would just talk about the different components and design considerations. Honestly if you want to find out what "forward voltage" or "constant current driver" means, how to solder, the importance of heatshrink, proper wire gauge sizing, how to connect components a DC circuit, and all the other myriad things you need to know before starting, there are far better explanations written by far more qualified people. That said, reading this should give you a fairly good idea of how I put this together especially if you have already done some research and have some electronics knowledge. Finally, IF YOU AREN'T COMFORTABLE WITH MAINS WIRING OR ELECTRICITY, DO NOT UNDERTAKE A PROJECT LIKE THIS. YOU CAN EASILY KILL YOURSELF.
Objective:
No external modifications to the hood or the tank, and a very finished and sleek look; that meant hidden drivers and other control components and bundled wiring. Aesthetic and ease of use was a priority. It was to look and feel like a finished product, something I could have bought at the store. Lastly, the light needed to be more than sufficient to keep the most demanding coral or clams, and the color temperature and overall brightness needed to be adjustable.
Drivers, Emitters, Dimming, Heatsink:
Looking up dimmable constant current LED drivers for an array of this size leaves one with a very small, specific selection of available products. At this point anyone in the hobby has heard of the Meanwell brand and their ELN-60-48D. To the same point, CREE makes several models of LEDs that are as tried and true as can be considering how new the technology is to the aquarium scene. I purchased 10 XP-G cool whites and 10 XR-E royal blues from RapidLED as part of their nano retro kit. Nothing out of the ordinary here. There are some really creative people on ReefCentral who have designed and assembled their own drivers from scratch, but that goes way further than I wanted. Maybe one day. :)
The ELN-60-48D (not the “48P” version!) allows for dimming using a 0-10V input, with 10V being the bright end. Each can run a string of LEDs with up to a 48V drop at 1300mA. In my project I sent 900mA through each string of LEDs as the max brightness because I didn’t see a tremendous increase in luminosity above that current. This also reduces heat, and therefore increases the lifetime of my components by a little bit. The drivers are dimmed with a 0-9V signal; they will not tolerate much more than 10V across the dimming leads or a diode inside can fry. Again, it’s better to err on the lower side. That was convenient because my LM317 voltage regulator circuit’s dropout voltage meant that with a ~12V supply, I could only get ~9V out the other side. It works fine if you treat 9V as the highest potential that the Meanwells will ever see on their dimming leads, and the only difference in the end is that there is a little bit less resolution in my dimming knob. The resolution and linearity of the dimming on the Meanwell drivers is questionable at best, so it wasn’t a big deal.
Below is a rough layout of my circuit that doesn’t include a lot of the intermediate points of contact on my eurostrip and other solder points… It’s just the essential logic. There are more details further down. I opted to use a voltage regulator instead of a simple resistor to bring the 12V supply down to the safe range for my dimming leads. It’s a little bit safer and, well, it looks cool. Instead of building my own LM317 circuit, I bought one on eBay that was extremely cheap (~$6) and has worked exactly how I expected thus far. It was nicer to have this as a self-contained unit than making it by hand and having the components loose or connected via a janky bit of breadboard; the circuit is actually pretty simple and has been described many times on DIY forums.
 |
| Testing the output. Works like a charm! |
The heatsink was something I wasn’t super particular about. I could have bought my own and drilled/tapped it, but it seemed FAR easier to purchase one of the ones that come milled, anodized, and as part of a kit. That and price was one of the major selling points of going with RapidLED.
 |
| Planning my LED layout. Visio and I had our differences but we're friends now. |
Enclosure
An extruded aluminum project box would house both drivers and all the wiring- two IEC C14 inlets for mains power, LM317 voltage regulator circuit, switches for the drivers, potentiometers for dimming, and everything in between. I did a lot of searching and it seemed like the type of enclosure I wanted came in common lengths, the largest of which was 8.66" long. This wasn't long enough for the drivers and assorted control components; a custom box was going to be necessary. One evening I stopped by and visited with the lovely people at San Jose Scientific in search of some chemistry supplies (for easier water tests). While I was there I coincidentally noticed that they carry extruded aluminum project boxes. One thing led to another, and I left that night with a custom sized ~12" x 6" x 4" box that was just right. I highly recommend you stop by if you're ever in the area, and having legit chemistry equipment (test tubes, tube rack, syringes, graduated cylinders) has made water tests and manual dosing a breeze.
This is the part where my dad’s many years experience as a machinist added a professional edge to this project. These enclosures are typically grooved on the inside and meant to house PCBs that slide in from the end. Instead of leaving components rattling around or otherwise fastened inside the enclosure , I wanted to designed and fabricate an aluminum mounting plate that would slide into the enclosure so that components could be fastened to it on either side. After a couple hours of careful measuring and planning...
...and another couple of hours of work, a bit of scrap aluminum from my dad’s shop was transformed into this:
 |
| Bask in its glory. Bask, I say! |
 |
| Grounding screws. |
The enclosure itself is made out of two identical shells which snap together. It has two end plates that screw into the shell pieces, and these end plates needed to have holes in them for mounting switches, pots, and power outlets/inlets. I don't have a handy picture of the back plate, but you can see it later in this post.
 |
| Front plate with spacer for potentiometer mounting. |
The above pieces were central to the success of the whole project and they came out perfectly. Thanks, dad.
Enclosures like this typically come anodized but mine was bare aluminum... which doesn’t handle a saltwater environment very well. Since anodizing is fairly expensive I went with three coats of RustOLeum primer and two coats of their industrial enamel. Sanding and tack cloth in between the primer coats helped smooth the finish. I highly recommend you set up some kind of dust-free environment for doing spray painting as it’ll have a huge effect on how glossy your finish looks. There were a few specs of dust in mine, but no biggie.
 |
| Painted receptacles? Yep, I went there. |
Wiring
I had to get power from the enclosure (in my stand) to the hood for both LED arrays and the two 60mm fans that cool them. The Biocube hood has a very thick and inflexible pair of cords that hang out the back of it which normally carry power to the two PC bulbs and stock fans; one of these would suffice for my retrofit (6 pins, 3 circuits). What made this convenient was it was already built-in and was terminated by a splash resistant screw-locked coupling. The wiring was all 18 gauge; more than adequate. The only tricky part was reusing the ballast's side of the cable and working its bushing into a hole in my enclosure. A little bit of applied brute force, some chiseling, and a trusty old table vice did the job, but it wasn't pretty. The ballast got the worse end of the deal.
 |
| A little hacksaw. |
 |
| A little chisel. |
The idea was by removing the "female" end of the cord from the dead ballast, I could use it with my enclosure for a seamless connection. In the end it wasn't long enough to reach the hood on its own. I cut up the spare cord hanging out of my hood (remember there were two, and I only needed one) and cut up the other ballast, stripped the cable ends, and with some soldering and heat shrink I fashioned those two bits into an extension that could plug into my enclosure's tail on one end and the hood's tail on the other. That sounds confusing so here are some pictures.
For wiring between emitters I chose white and blue 22AWG UL 1007, stranded. Tinned wire doesn’t suffer so much from corrosion. Stranded lends flexibility, making it easier to route between emitters while removing pressure from the contact pads on each star. Some people swear by solid wire, though, and either will work. After some cable-management kung fu, the hood cleaned up quite nicely.
Within the enclosure I used 18 or 20 AWG stranded plain copper for most of the connections. The 120VAC and grounds are carried by 16AWG stranded. Every single end of every piece of my wiring was stripped no further than necessary, tinned, inspected for fraying and quality, and connections were covered with one or two layers of polyolefin heatshrink wherever appropriate. It's good to get several sizes of heatshrink, from 3/32" to 3/4". A 2:1 shrink ratio is pretty standard and I find a heatgun (or a hair dryer) works better than a lighter for shrinking. As you can see in some of the pictures, I used large tortilla bag type clips to hold wiring still while I soldered. There are better solutions, but these worked great. A huge thanks to my friend (who shall not be publicly named on the Internet) for lending me some advice and his 50Watt soldering iron and station. Don’t leave home without it… by the time I was through with this project I felt like I could solder like a pro, but that could have been all the lead and flux smoke I inhaled.
Many of the parts for this project were purchased online but the true hero was a great little shop called HSC electronics in Santa Clara. It’s the type of shop you don’t find that often anymore… walking around in there, I felt like a kid in Willy Wonka’s chocolate factory. They have all sorts of gizmos, electronics components, wiring, anything that you can possibly imagine, and all at bargain prices. If you build something like this, a shop like HSC is the first place you should go for parts. I wish I had started there instead of finishing there; it would have saved me a lot of cash. They also came through in a snap when I needed parts at the last minute.
Assembly
Early on I had decided against soldering every bit to every other bit that required contact. I used a couple of 12 position eurostrip terminals to facilitate easy replacement of any single component on the board without much de-soldering required, and used jumpers to connect terminals that shared connections.
It was a slow process putting everything together but over a few nights I managed to get it all assembled. I verified and reverified all connections; you can easily use a DMM’s ohmmeter to test for continuity. What you don’t see in these pictures are my documentation for the pinout, wire map, and other notes; it is crucial that components are connected as intended or you can fry your parts, or worse yet, yourself. I sat down and worked out what was connected where on these Eurostrips using excel well before I even received them and even then it was at times confusing to lay out the wiring on the fly. Again my dad’s major contribution to the project, the mounting plate, was a tremendous advantage as you can see.
 |
| Illuminated switches for extra bling. |
 |
| These eurostrips and the jumpers that are made for them were extremely handy. I did have to flatten the ends. |
 |
| The aforementioned fridge-magnet-tortilla-chip-bag-clips. |
 |
| Coming together... |
 |
| Final testing. Note grounds! Always ground!!! |
Throughout all of this I only hit one major snag. As I was testing the 12V DC circuit, I had a really hard time figuring out why it wouldn’t power up properly. This was something I had not run into during research.
The supply that came from RapidLED could supply 500mA; much more than enough current to power these components during operation (their nominal current). I figured something was wrong with the circuit itself… I painstakingly tested each connection, searched for shorts or unintended ground, retightened all terminals, and yet still had the issue. The voltage on my supply dropped from 12V to ~5V as soon as it was powered up which indicated that the circuit was drawing too much current. At first this was very perplexing; the circuit powered up fine and stayed at 12V if I removed the LM317 circuit or the fans... but with both present, the problem persisted. I consulted my EE friends who confirmed the excessive current draw. The revelation came when I tried reconnecting the LM317 circuit once the fans were already running… it worked! I had the same result when reconnecting the fans once the LM317 capacitor was fully charged.
The issue was inrush current on the DC motors which drive the fans and the capacitor in the LM317 circuit. At first, these components allow a large amount of current to pass through them while they spin up or charge, respectively. The motors begin to generate back EMF in the wire once spinning and eventually reach equilibrium, which happens over the course of a few seconds. The capacitor doesn’t sip very much current at all once charged. It simply turned out that the combined demand from these components within the first few seconds after power was applied was more than my little DC supply could handle- the fans wouldn’t spin up so they kept passing current and the capacitor wouldn't charge. Yikes.
A last-minute trip to HSC rewarded me. I bought a 2A 12V/5V supply but the Mini-DIN connector on the end needed an adaptor before it would fit into the DC power jack on the back of my enclosure. I found the requisite parts at HSC, brought them home with my new supply, and made my own Mini-DIN to DC power adapter in no time.
Even though the inrush current on various components can be 6-10 times greater than the nominal current, a 2A supply was enough by my logic (since a 500mA supply could start up one or the other, but not both at once). I breathed a sigh of relief when everything worked on my first attempt with the new supply. All that was left was to recalibrate my dimming circuit and Meanwells to the new input voltage- the larger supply gave me ~11.5V under load instead of the ~12.3V on the smaller original, which left me with ~8.5V max across the dimming leads.
 |
| Calibrating the meanwells. |
I gingerly slid the mounting plate into the enclosure shell, snapped the two sides together, screwed on the end plates, and did a full dual circuit test for the last time before putting everything together inside the stand and over the tank. Done at last!
Conclusions and Afterthoughts
What fun. I learned so many new things... and relearned so many things that I shouldn’t have forgotten to begin with. An education in Physics really came in handy here, although most of the relevant applied knowledge was stuff that I never really paid attention to in class. Having a engineering/project-planning oriented approach really helped, as did copious amounts of research. When all was said and done, I had made countless small modifications to the design over time before I even picked up a soldering iron. The finished product came out better than I imagined it could.
At first the LEDs were dialed in to their brightest setting and I felt like I had a 200W metal halide bulb over my tank; it was that bright. I am currently running them at a little below 50% brightness to slowly acclimate the corals to the increased light and over a couple weeks will eventually get them accustomed to full power.
A lot of people complain about color banding, which is the effect of having multiple point sources of different colored light creating shadows of color around rocks. This is noticeable in places, but it doesn’t bother me; I figure it is one of the attributes of LED lighting and can’t really be avoided with using arrays of emitters like this, so I might as well cope.
After seeing the four empty spots on my heatsink, I wish I had gotten 4 extra LEDs to fill it up. My drivers could easily have handled two more LEDs each, but it would not have made a huge difference in the end; I have more than enough light.
DPST switches would have been better so that I could break both the neutral and the live connections on my building wiring. My SPST switches only break the live connection; some older buildings aren't wired correctly and this could cause some serious safety issues. Fortunately this building is fine, but if I ever moved to a new place I would want to address this with new switches, and would use DPST in the future.
The royal blues are on without the cool whites for a period of about 30 minutes late at night… it looks unreal. A camera does not do justice to the colors that come out of these LEDs and the fluorescence of the corals.
Without friends and family and the huge community of fellow reefers online, this wouldn’t have happened. My thanks go out to my dad and friends, all of you trailblazers who have fried emitters, exploded capacitors, burned diodes, started fires, and paved the way for LEDs to be the future of this hobby.
