LEDBrick Project - DIY pendant w/ pucks

Re-reading what you posted about the 6-up is see you have specified Cree XP/TE for the white and blue chips. I'm curious to know why you went with these and not a Rebel white and blue?

The only reason I ask is that it appears that the warm white that Steve's caries has a peak in the red at close to 650nm (Ideal for photosynthesis and good coverage of the 600 to 650 range of red required for the expression of some sps pigments) where as the warm white at Rapid LED is shifted back towards green/yellow/orange with a peak at somewhere around 600/610nm.

When using the lime chip I would think it is much more desirable to have the white chip more red shifted so you don't end up with too much light in the 560 to 600nm range. Just a thought, but maybe you are sourcing your Cree chips elsewhere and choosing something different in the white?

Yeah, I know. I'm slow to pick up on the details... :o
 
Swapped the white for a Rebel footprint, since you do bring up a good point. Going to look at the Rebel in the blues (go all in :))
 
theatrus said:
Swapped the white for a Rebel footprint, since you do bring up a good point. Going to look at the Rebel in the blues (go all in :))
Click to expand...
Excellent! I know that the Rebel whites are quite popular for DIY on some other sites because of the red shift. It looks like the rebel chip is just a little cheaper as well, though not quite as powerful.

I'm going to provide a place to drop an NTC thermistor on the board for easier thermal measurements (and long term monitoring if you want to hook it up).
Click to expand...
Even more excellenter!! :D I figured I would have to mount something myself to monitor board temperature until we get some idea what its actual thermal properties are.

I know that some companies drive their diods pretty hard and have issues with early failure. I would rather put a few extra clusters in and run them cooler for longevity.
 
awesome build, i am a huge DIYer when it comes to my tank, i even built the rimless 150 my self. i however haven't made my own light yet is i was able to get an awesome deal on my Chinese black box lights. I like this design, and would be more than happy to help you in any way you need. i have eagle and am fairly good with it. i have been looking for the MCPCB boards so i can etch them myself, but i cant find them online except possibly a company from china but im waiting to hear back from them. i was wondering if any of you have researched them and if you have had any luck finding them. Thank you
 
I personally haven't bothered with home etched PCBs in ... 10+ years. Especially with a specialty product like an MPCB. The Chinese manufacturers, and even US suppliers, are so cheap now it makes no sense for something vastly inferior (no plated holes, vias, edge routing, mask and silk, etc).

I tend to circulate with the various web-order outfits (PCBWay, Elecrow, etc, as listed on PCBShopper.com). The Apogee amplifier boards cost less than a $1 each, shipped, from China from PCBWay. Their MPCB prices are very good as well - I'll probably give it a shot. The only thing you have to deal with is DHL in the United States, which can be highly variable :)
 
I spent some time this week bringing up the microcontroller for my smart driver boards on a little test board. Had a few false starts since I wasn't in tune with Microchip's PPS system mapping input and output to the same pin for open drain (I2C) communication - whoops. For a sub $1 part, its meeting the requirements.

I'm going to do one more pass on the three emitter designs and place an order, and then work on the 1,2 and 4 up LM3414 and A6265 (buck+boost) driver designs based on the controller module. Both should add a ton of flexibility to normal LDD deployment.
 
theatrus said:
I spent some time this week bringing up the microcontroller for my smart driver boards on a little test board. Had a few false starts since I wasn't in tune with Microchip's PPS system mapping input and output to the same pin for open drain (I2C) communication - whoops. For a sub $1 part, its meeting the requirements.

I'm going to do one more pass on the three emitter designs and place an order, and then work on the 1,2 and 4 up LM3414 and A6265 (buck+boost) driver designs based on the controller module. Both should add a ton of flexibility to normal LDD deployment.
Click to expand...

Maybe I missed it, but how is the final layout with regards to color and placement and how many LED's total? And what was thought on overall cost (guess that is)?
 
For the star protos, I'm planning on:

- 16 LuxeonZ 2 channel (8/8)
- 16 LuxeonZ 4 channel (5/4/4/3)
- 6 LED mixed 4 channel (3/1/1/1)

Since the thermal densities are high with the 16 LED drivers, I'm going to test a 8 LuxeonZ 2 channel design as well (4/4)
 
As for color combinations and cost, its up in the air a bit on the Luxeon Z side. I got a stack of the intended reflectors today and some small quantities of emitters so I can build a few samples up.
 
theatrus said:
For the star protos, I'm planning on:

- 16 LuxeonZ 2 channel (8/8)
- 16 LuxeonZ 4 channel (5/4/4/3)
- 6 LED mixed 4 channel (3/1/1/1)

Since the thermal densities are high with the 16 LED drivers, I'm going to test a 8 LuxeonZ 2 channel design as well (4/4)
Click to expand...
Cool!

Out of curiosity, how do you determine the thermal properties of a board/star?
 
The best way is to directly measure the LED package thermal spot (if available), the star plate, and the mounted heatsink. This was you can grab real life thermal resistances.

As for drivers, I have an extra set of the 8 channel LM3414 design I'm using for my current puck boards, since my ready made digital control drivers aren't ready yet.
 
I was thinking more from a design perspective before you've actually built your boards. The diode makers give you all sorts of thermal data, but how do you approach that from the board side?

As a designer I would think that you would want some idea of what your physical limits in terms of board size and diode density are.

I'm guessing there are points at which more heat sink will not make up for diode density.

I also presume that one of the reasons we do not see puck designs in the cheaper black box lights is a price restriction. Big heatsinks and heat dissipation issues with pucks would add quite a bit of cost to the unit.

I'm always curious to know that the designers constraints are and how they figure them out. I've long wondered why commercial manufacturers have not moved to a smaller puck design to better distribute light output.
 
Absolutely a good question. I honestly didn't invest a huge amount of up front design into the thermal properties, beyond some simple thermal resistance calculations where the PCB portion is a large slightly informed guess. With today's price of prototypes it doesn't cost that much to fail.

My power density is on par with a desktop CPU (up to 60W in a slightly larger surface area). Active cooling is a must. It's also in a similar problem to larger multichip arrays. However both of those designs have more flexibility in their thermal paths and will fare better than a cheap MPCB (and at least Intel has done huge amounts of validation ;))

A lot of LED lights are done as high density arrays. Kessil is one of the few (only?) that does it in a reef setting. My theory is the love of lenses to reduce emitter counts for the same PAR (and add hot spots) drives the design.

This could all be a disaster, but at least there will be some data :)
 
Well, as long as you didn't take a complete wild a$$ed guess at it. :)

For the 6 ups, if I understand correctly, at full power, they will have a max of about 14.5w and 8.4w of heat to dissipate. I've got some old copper industrial computer heatsinks (3"x3"x1/4) that may do the trick for me. I don't know if I can run them passive though, particularly since the fins are designed for active cooling.
 
Good news: prototype emitter boards should show up tomorrow.

I've also been working on the smart drivers, here is a two channel LM3414 based board. You can daisy chain multiple units together using I2C (only three wires), which lets you both set the output current (in this case, 350mA to 1A,, at ~ 30 steps) per board, and PWM duty cycle (0-100%) per driver channel. Its a few lines of Arduino or Raspberry Pi code to set - I'll make a few libraries :)

ezled-proto.png


I'm planning on making a variant based on a buck-boost (50V out from 12-36V in) and the high current (4A), all with the digital control bus to reduce wiring and give lots of programmability.

I'm torn on the mounting holes - I've spaced the long dimension to be 3x T-slot increments (0.655 in), but can go smaller if there is no preference. Thoughts? Since there is no conventional mounting box (maybe we can all agree on a good one from polycase.com ), there is a lot of variation.
 
Looking good!

pcbs.jpg


Don't yet have all the emitters needed :)

I also experimented using a piece of high-transmission ACRYLITE® Satinice (Satin Ice) Colorless 0D010 DF as a diffuser if you wanted to run the LEDs close to the water:

diffuser1.jpg

diffuser2.jpg


The disc and clip-on legs are laser cut.
 
You must log in or register to reply here.

Similar threads

Easygoing1
PFO Aquatics SOLARIS Rebuild And Questions
Replies
8
Views
402
oreo57
oreo57
C
Facing LED Woes: Controller Troubles and Compatibility Questions 🤔💡
Replies
3
Views
322
combs.eric.813
C
B
JBJ Arctica 1/10 hp Chiller Modification for External Control
Replies
9
Views
763
griss
griss
C
DIY Stand for Flatback Hexagon Tank
Replies
2
Views
472
Drummer98
Drummer98
Rpkneumann
BeanAnimal conversion of an 2005 Oceanic 150 GA Tank - possible?
Replies
9
Views
556
Rpkneumann
Rpkneumann
Back
Top