The not-so-nano evil cluster build

[bfliflet]

Finally completed the wiring for the 4 panels. Tried to keep the wires tidy. Left a space on the bottom to augment with amber if I get bored...

Ran into a few snags along the way. The M reflectors have pins to line things up on custom holders.

Ground down the pins with a Dremel.

LGB XTE optics were a little too snug for 405/430nm

Widened with needlenose pliers about half-way through the project. I'll widen any that fall off and re-silicone

Finally glued the optics down on all panels

Next step... Molex connectors to disconnect. Waiting on more 24pin SIP sockets for LDD board and 48v->12v DC adapter to power the CPU fan. Need to order acrylic for housing.

[milfordguy]

That looks insanely neat. You're right about the white wire, I think it really makes for a nice clean look.

 

Where did you get your molex connectors? I was wondering how to use some sort of quick disconnect in order to make the setup as modular as possible. Where will you put the disconnect? Right off the heatsink at the end of the wires you have in that last picture?

[bfliflet]

That looks insanely neat. You're right about the white wire, I think it really makes for a nice clean look.

 

Where did you get your molex connectors? I was wondering how to use some sort of quick disconnect in order to make the setup as modular as possible. Where will you put the disconnect? Right off the heatsink at the end of the wires you have in that last picture?

I am using the Molex Microfit 3.0 connectors - the 14x2 male ones I am using, for example, are #43020-1401. They're rated at 5amps. They're available through typical supply houses like Mouser, Newark and Digikey. If you go the molex route, pick up the molex crimper. Much easier to crimp, stronger and better looking overall.

 

I plan to mount the connector on the last fin of the heatsink. They'll connect to the LDD boards which will be mounted on the acrylic enclosure (~8"x~8"x~3-4"). The LDD boards will have two connectors - a 2 pin molex running to the 48v power suppy and a pair of cat5 connections. One CAT5 to the Arduino for PWM and one to daisy chain to the next fixture. I may go with something other than cat5 because it limits me to 7PWM signals (plus ground). Or I may just combine the 405/430 PWMs.

[farkwar]

"If you go the molex route, pick up the molex crimper. Much easier to crimp, stronger and better looking overall. "

 

Youre kidding right?

 

Its $350.

 

 

 

I have not found a Cat5 female block that didnt look like a botched abortion. Even with Cat5 cable, I would use the Molex Microfit 2x4(for what youre doing); with some heatshrink it would look much better than an actual RJ connection. If I bought the crimper, I would be using it on everything I could.

[bfliflet]

"If you go the molex route, pick up the molex crimper. Much easier to crimp, stronger and better looking overall. "

 

Youre kidding right?

 

Its $350.

 

 

 

I have not found a Cat5 female block that didnt look like a botched abortion. Even with Cat5 cable, I would use the Molex Microfit 2x4(for what youre doing); with some heatshrink it would look much better than an actual RJ connection. If I bought the crimper, I would be using it on everything I could.

$350? Ouch! I picked up a generic crimper, Waldom Electronics W-HT-1921 from a local Fry's Electronics for around $20, well worth the money. It's not ideal for Molex Microfit but works well. Thanks for the advice on the CAT5. Looks like I need to think more about that direction.

[bfliflet]

Sorry for the delay in updates. I was affected by the ever changing product requirements dilemma -- the more I built the more I wanted to tweak. I built two light panels to test things out and assist in designing an enclosure. In the attached photos, you can see two without the enclosure, the enclosure and the old lighting scheme in play. The photos are pretty realistic. You might also be able to see the difference between full spectrum vs old school CW/RB. I must warn people that these evil cluster builds are insanely bright. Even with a PAR meter I had a couple of corals that started to bleach. On the upside a pocillipora colored up nicely to almost a forest green. Ignore the wire mess. I am still designing the PWM signal cabling. Thanks to evil, that problem might be solved. I also need to adjust the panels so the old ceiling cables will get moved.

To do list:

- Mount the two temporary lights into enclosures

- Mount missing panel into enclosure. Wire. Hang

- Add temperature sensors to control fans

- Convert from loose Cat5 PWM cable to DB15 serial or VGA cable

- Add molex connectors for PWM cable.

- Move ceiling mounts to minimize overlap / dead spots.

- (Possible) Add amber LEDs to panels and wire in.

[Mr. Microscope]

Very nice! Thanks for the update. They look incredible! It's nice to see the comparison between the old school and the new.

[milfordguy]

Man this ended up coming out incredible. I know you still have to put the finishing touches on the build, but it really came out nice.

 

I like the acrylic enclosure. It looks better than just hanging up the plain heatsink. Now that you already have the Cat5 in place why are you thinking about switching the the DBA15 or VGA? Is it just for a better quality wire when compared to Cat5?

 

You should snap some pictures of the enclosure/back sides of the heatsink if you get a chance. I'm curios as to how you ended up laying everything out. Are those just the fans you have resting on top of the heatsink?

 

Overall I think it came out great, you can definitely tell a difference in the color when compared to the CW/RB setup.

[bfliflet]

I'll upload a better interior shot later tonight or tomorrow. Switching to DBA or VGA is mainly because I am a control freak. I am using all 8 channels of the CAT5 cable (7PWM + gnd) and want to retain that functionality although it's not really needed. I am likely to add an amber LED channel to see if I can get some pink millies to emit. I am also planning to add temp sensor (yet another signal) and a way to shut off the CPU fans (and another). Having them run off the meanwell 48dc/12dc on the main power supply means they're always on. I could put the power supplies on a timer I suppose. Honestly, sometimes I think I create new projects simply to pass the time while frags grow -- worse than watching paint dry...

[milfordguy]

I'll upload a better interior shot later tonight or tomorrow. Switching to DBA or VGA is mainly because I am a control freak. I am using all 8 channels of the CAT5 cable (7PWM + gnd) and want to retain that functionality although it's not really needed. I am likely to add an amber LED channel to see if I can get some pink millies to emit. I am also planning to add temp sensor (yet another signal) and a way to shut off the CPU fans (and another). Having them run off the meanwell 48dc/12dc on the main power supply means they're always on. I could put the power supplies on a timer I suppose. Honestly, sometimes I think I create new projects simply to pass the time while frags grow -- worse than watching paint dry...

Is your controller going to be a jarduino, or a separate arduino build? I'm assuming one of your 7 pwm signals is to control the fan from the temp sensor?

 

I went with just regular non adjustable fans because of the cost. What is the 1 ground wire you are using from the cat5 going to?

 

Sorry for all the questions, but your build is so similar to mine I want to know how you did it. I have about a million thoughts of how I want to finish it, but I'm just unsure.

[bfliflet]

Part of the DIY experience for me is learning. It's not that I don't trust Meep or the Jarduino author but when something goes radically wrong, I can't call Meep up at 3am to fix it. I am a SW engineer by trade so writing the Arduino code is the easiest part. I do have 7 light channels NW/405/430/RB/CB/CY/RD (and amber is next). The 405 and 430 are close enough that they really could be on the same signal. The others I am finding are mostly subjective and I'd rather build in the functionality than wish I had.

 

I could have it wrong but the 8th signal for me, GND, links up the GND between the Arduino and the LDD such that the LDDs recognize the PWM voltage coming from the Arduino. It didn't seem to work until I did that.

 

My fan is a 3wire fan as well but I want some aspect of control. I figure the experts on here will chime in on how to control this type of fan via an Arduino or I'll google it.

[milfordguy]

Part of the DIY experience for me is learning. It's not that I don't trust Meep or the Jarduino author but when something goes radically wrong, I can't call Meep up at 3am to fix it. I am a SW engineer by trade so writing the Arduino code is the easiest part. I do have 7 light channels NW/405/430/RB/CB/CY/RD (and amber is next). The 405 and 430 are close enough that they really could be on the same signal. The others I am finding are mostly subjective and I'd rather build in the functionality than wish I had.

 

I could have it wrong but the 8th signal for me, GND, links up the GND between the Arduino and the LDD such that the LDDs recognize the PWM voltage coming from the Arduino. It didn't seem to work until I did that.

 

My fan is a 3wire fan as well but I want some aspect of control. I figure the experts on here will chime in on how to control this type of fan via an Arduino or I'll google it.

 

I know that you have to share a common ground between the Arduino and the LDDs, so that makes sense that is your 8th wire. Just to clarify you have the common ground from the Arduino attached to the LDD board ground? I was thinking about just attaching the Arduino ground directly to the PS. That probably does the same exact thing if I had to guess.

 

Are you using an Arduino mega? That's what I have and am using it with the LCD shield.

 

So your 3 wire fans don't have a pwm input? Neither do mine, but I'm sure we could figure out a way to set them on a timer through the Arduino. It seems like that would be fairly easy to do... Of course that's easy for me to say when I have almost no experience with any sort of coding, and you're the one that is a SE by trade haha!

 

 

Also I tried to test my array without any pwm input thinking they would just run at 100%... I didn't have any success. Do you know if you need a pwm signal even if you just want to test it out without dimming? I didn't have a controller or anything hooked up. Just the LDD board to the PS... I meant to post this on my thread but completely spaced it until I saw you talking about a common ground.

[jedimasterben]

If your board uses resistors on the PWM wire, they will not send power without a signal. Most boards have jumpers that disable the resistor, so pull that if so.

[milfordguy]

If your board uses resistors on the PWM wire, they will not send power without a signal. Most boards have jumpers that disable the resistor, so pull that if so.

I got my boards from 02Surplus and they do have a built in resistor if I remember correctly. That must be my problem. Thanks Ben!

 

Edit: I'm an idiot. Now that you mention that I remember they are designed like that incase you lose pwm signal you won't bleach your corals...

[bfliflet]

You should snap some pictures of the enclosure/back sides of the heatsink if you get a chance. I'm curios as to how you ended up laying everything out. Are those just the fans you have resting on top of the heatsink?

Here are a few more photos of the panel I was working on last night though the wiring is not complete. The unconnected white wires are the LED strings that need to be connected to the LDD board. I did modify the 24pin socket so the 48v/12v adapter could use the same board. I still need to put the rubber feet on the bolts for the fans and enclosure that rest on the heatsink to avoid galvanic corrosion. They'll also reduce vibration noise though nothing is audible.

You can see the molex 2 pin connector for the 48v source and the molex 12 pin connector for the PWM signals.

[milfordguy]

Okay thanks. Do me a favor and take down your pictures immediately. They're too clean and they're going to make me look bad! haha

[Paleoreef103]

This is an amazing build. What percentages are you running the lights on in the above picture? The full spectrum side is perfect!

Also, why are you thinking about an Amber LED? Is it not warm enough?

[jedimasterben]

Amber LEDs are some of the most inefficient out there, white LEDs put out far more light in that spectra than dedicated amber LEDs would.

[bfliflet]

This is an amazing build. What percentages are you running the lights on in the above picture? The full spectrum side is perfect!

 

Also, why are you thinking about an Amber LED? Is it not warm enough?

Thanks. My wife thinks it's a little too purple. It's probably because she sees it next to the bland CW/RB.

 

Here are the total (4 panel) max wattages / percentages currently in use.

Vero13 - 84% / 147W

Luxeon M - 100% / 192W

405 - 100% / 39W

430 - 100% / 114W

OCW Cool Blue - 100% / 40W

OCW Cyan (visibly green) - 100% / 40W

OCW Red - 61% / 24W

Global (everything scaled back) - 63%. NOTE: I've recently scaled this back to 50% because of some coral burning. At 63% i was registering a max PAR at 920 at surface (comparable to a 250W metal halide at a LFS). As a point of reference, some midlevel SPS were receiving about 250-300 PAR at during peak hours. KillaWatt reported about 420W for everything (fans/LEDs/power supplies/etc).

 

As to exploring amber, the NW is definitely warm enough. Think of it another way, one reason people load up on RB/HV is to get inverts to emit greens and other colors well beyond natural amounts. Amber is a spectrum (as is deep red among other colors) that affects certain corals similarly but if you don't spike it, you won't get noticeable spectral emissions. The same thing would happen if you only relied on the NW's inherent royal blue. You'll get green to emit but it's not pronounced. It's washed out by the intensity of the broad spectrum in NW. Hence, a benefit of spiking RB/HV. Honestly, I am the skeptic but my local reef club buddies keep swearing by it. You can also see amber spikes in metal halide bulbs that added it specifically to get that color to emit.

Amber LEDs are some of the most inefficient out there, white LEDs put out far more light in that spectra than dedicated amber LEDs would.

Good to know. If I add it to the system, it's only about 18W total but still provides a noticeable bump on the "hypothetical" spectral graph.

[bfliflet]

Correction: OCW Cyan is 84% of 40W.

[NirvanaandTool]

Wow stellar work there. Love the wiring work and the acrylic cases. Color difference between the old school is really awesome to see too. Nice job!

[jedimasterben]

Good to know. If I add it to the system, it's only about 18W total but still provides a noticeable bump on the "hypothetical" spectral graph.

No, actually it will not, again goes back into the abysmal efficiency of dedicated amber diodes.

[bfliflet]

No, actually it will not, again goes back into the abysmal efficiency of dedicated amber diodes.

The specific LED I am thinking about is LXM2-PL01 - Rebel Luxeon PC-Amber. It has luminous efficiencies similar to other Rebels. Maybe you're thinking of the LXML-PL01 which definitely are dogs? This is still an optional aspect of the build but I am reserving space for it (or some other narrow spectrum). I am heeding your warning. Definitely not in a position to debate.

[milfordguy]

Here are a few more photos of the panel I was working on last night though the wiring is not complete. The unconnected white wires are the LED strings that need to be connected to the LDD board. I did modify the 24pin socket so the 48v/12v adapter could use the same board. I still need to put the rubber feet on the bolts for the fans and enclosure that rest on the heatsink to avoid galvanic corrosion. They'll also reduce vibration noise though nothing is audible.

 

When you say put the rubber feet on the bolts are your fans just resting on the heatsink? Or are the bolts going through to the front of the heatsink?

 

Also, did you use right angle headers to attach your pwm signals to your Arduino?

 

I attached the RTC by soldering to the bottom of the board, but it came out somewhat suspect.

[bfliflet]

The fan isn't mounted to the heat sink. The bolts are resting in the grooves the heatsink. The bolt holes of the fan make it a decently snug fit and it doesn't slide around. Rubber feet will be added to prevent galvanic corruption and possible vibration issues.