next topic »

[on:]

I'm putting together a project proposal and need to finalize my plans for budget before I can actually do any testing or building. I've chained RGB LEDs before, but have never used this much power over this size area.

Briefly, I'm putting together an outdoor interactive light display that will be played by an electric piano in various effects. The piano will connect to my laptop, then to an uno that will shift out to 88 Megabrites:

http://docs.macetech.com/doku.php/megabrite

Each Megabrite will be hung our housed on the ground inside a paper lantern. At the scale in the layout below, each gray circle is about 24" in diameter.
Here is my layout plan:



I'm mostly worried about voltage drop over these distances and would like feedback on setup options I'm thinking of.

Some Stats:
The diameter of the circle is 80'
The longest distance between Megabrites will be roughly 10' (all connections will be 10' to modularize construction)
Because of the zigzag, I'm guessing about 45' from power supply to the end of each direct power line
Each Megabrite can pull 300mA - 88 will pull 26.4A
22 Megabrites (each quadrant) would pull 6.6A
They recommend a chain of 5 or less MegaBrites to avoid voltage drop
They can handle between 5.5V and 9V

If all 88 are connected to the same power source, and I need 30A, I will probably need 10AWG or 8AWG to handle it and reduce drop as much as possible. I'm not sure I want to spring for 200ft of 8/2 wire if it makes more sense to use 4 power supplies at 8A each and go with cheaper 12/2 wire. Is that a legitimate concern or am I over complicating it? Would the trade off make sense? Any recommended power supplies for this application?

Also should since they can handle 9V should I go for that so I have the freedom to drop all the way down to 5.5V at the furthest load?

Thanks!

[Reply #1 on:]

If the entire quadrant load (6.6A) was connected to the END of the 40' power runs you might need some heavy wires but I suspect that since the load is distributed and the data cables help carry some current you can probably get by with 12 or 14 gage wire for the four power arms.

[Reply #2 on:]

Is that a legitimate concern or am I over complicating it?

Wire up a worst-case run and see how it works.

[Reply #3 on:]

Thanks for the replies. I'd love to wire up the worst case and see how it goes, but I don't have the money to do it, and this is a budget proposal that may or may not get approved. I'm just trying to be as accurate as possible up front.

[Reply #4 on:]

Power distribution is not the problem. The problem is TTL data over 10 foot cables. We never recommend more than 3 feet for MegaBrites, and suggest keeping it short as possible.

This is a job for OctoBars, with a separate arduino and control box for each quadrant. Or OctoBrite DEFILIPPI. Both are star diagram controllers where LED power runs over the long cables, not data.

[Reply #5 on:]

Thanks for the reply macegr,

I hadn't thought of that limitation. Is it just too much noise developing over that distance in TTL that's the problem?

Also, could I not just use 1 arduino for the whole setup with octobars? 1 data pin for each set of 3 octobars (4 data pins total) and all sets sharing the same latch, clock, and enable pin?

Thanks!

[Reply #6 on:]

Also, could I not just use 1 arduino for the whole setup with octobars? 1 data pin for each set of 3 octobars (4 data pins total) and all sets sharing the same latch, clock, and enable pin?

But that suggests you will again have the TTL data, latch, and clock going farther than 10 feet.

[Reply #7 on:]

But that suggests you will again have the TTL data, latch, and clock going farther than 10 feet.

In this case, the 12 octobars would be in a chained circle around the center power supply and arduino, probably not more than 5 feet in diameter and definitely less than 3 feet between octobars. All the distance will be covered by the power going to each satellite led with roughly 40' being the furthest distance traveled.

It'd be a big change because I'd end up using 2500+ feet of flat 6 wire cable (28AWG?) and RJ25 connectors... That could get fun quickly... I'm working on a new layout plan and budget based on those features to compare the two.

[Reply #8 on:]

^ suggesting the satellite leds are all http://docs.macetech.com/doku.php/satellite_module_s-001

not MegaBrites

[Reply #9 on:]

I mentioned a star diagram since that would probably be the most efficient. For each quadrant, you'd locate however many LED controllers you need in a central location, then run cables straight out to each light module. The Arduino needs to be there due to the data cable length issue, but as you planned already, those active USB extenders work great. The downside is having four Arduinos and keeping them straight in software, but it'll actually improve your bandwidth/response time, and also be pretty minimal in cost compared to the rest of the system.

The Satellite modules are going to be a bit more expensive than the MegaBrites. You'll get excellent brightness from them, they'll light up paper lanterns very nicely, you might want to test one of each to be sure. If your venue is very dark and you can survive with even LESS light than a MegaBrite, the Octobrite DEFILIPPI may work. You can get the smaller/cheaper Satellite modules, and connect everything with IDC 2x3 connectors and ribbon cable instead of phone cables. A system with the DEFILIPPI is probably an 1/8 the cost of an OctoBar system. However, it will be 3/5 as bright as MegaBrites, and 1/5 as bright as the OctoBar + Satellites.

[Reply #10 on:]

Thanks for the great advice. I'm exploring the two different routes. I'm not sure we want to go for lower light than the megabrite. And the idea of the octobar + satellites leds makes me drool a little more

One thing I'm considering is that the venue would be in a garden with some trees, so there would not be foot traffic within the 80' circle. This also means that the pc, which will play a minor but important role in UI, needs to stay on the perimeter of the 80' circle. Strangely enough, it's looking like it might be cheaper to go with 1000 extra ft of telephone wire then it would be to get 150+ feet of USB extenders + 3 extra arduinos. Not by much. But some.

I'll be posting a modified layout soon.

[Reply #11 on:]

I really need to finish that data extender module design...supposed to solve exactly this type of problem. Never enough time....

[Reply #12 on:]

New layout!

Any major objections to feasibility? This brings us right up to our grant submission budget ceiling.