LED Disco Wall! [2m x 2m, RGB]

• 256 RGB LEDs.
• 256 ping pong balls.
• 100m of .65mm wire.
• Eight 1m x 1m hardboard panels.
• 32m of 2by1 wood.
• 15-20A Power-supply. [PSU from a PC].
• Matt black paint.

Arguments For and Against Double-resolution:

For:

• Less man hours - all LEDs come pre-wired and don't need soldering together, drastically reducing the time it takes to make a board.
• Picture quality - hi-res allows crazy picture quality or 4 lines of text.

Against:

• Number of LEDs - double resolution means quadrupling the LEDs. 256 per board, 1024 in total. This brings the cost to between £360 and £600 for LEDs alone!
• Power drain - 1024 LEDs require 60A or more to run full brightness, therefore needing 4 separate 15A power-supplies and 4 plug-sockets.
• Voltage drop - the LEDs start to change colour by the end of a 64-string. By the end of a 256-string, the last ones will probably just be red and require a lot of extra power-jump wires.
• Data stream - software allows only X amount of LEDs and would require overhaul.
• Overwhelming audience - 1024 LEDs will be too bright for a 2m x 2m square, coupled with a hi-res animation may just be too much for a disco-wall. The idea is to provide anamorphic disco lights, not a giant video screen; we have a projector for that.

Problems and Solutions:

THE BUILD:

The Boards:

I took 1 metre by 1 metre hardboard flooring panels and painted one face of each matt black. The back was then framed with a strip of 2by1 boxwood, though a second strip will be necessary later on to give enough depth for the LEDs and wires. The painted face of the panels were then measured up for the LEDs. The four 1m-square boards make up a larger square, so the most accurate way to measure and plot the dots was to have all four boards arranged together. Due to the arrangement, each set of 64 LEDs cannot be central to the board as this leaves an equal border around the outside of each board, which in turn makes the gap between the centre LEDs twice what it should be. The square of LEDs should be central across all 4 boards together, making the border around the inner two edges of each board a third of the border around the outside, shown in the diagram below.

I did go ahead and double the depth of the wooden frame and added more hardboard panels to the back of each board 

Making the wires and prepping the LEDs:

This was by far the longest and most arduous task of the project, over half of the total time spent. The LEDs come in a long string like Xmas lights, but can only be spaced about 6cm apart. This would double the resolution and use up all our LEDs on one metre-square board, so the 3 wires in between each one needed lengthening.

First, the 256 LEDs had to be cut from the 50-long strings, including the final 6 annoyingly, but at least we have 44 spares. We then stripped the ends of the 3 wires to each side of the LEDs while they were still joined, then separated the six wires, twisted their ends and tinned them, ** in total.

756 wire extensions were also needed. We cut these from a 100m spool of .65mm wire, which cost £10. Over 1500 wire-ends were then stripped, twisted and tinned individually, which was an incredible amount of extra work. Pre-tinned jumper wires were cheap, but just too skinny to do the job. Suitable ones were prohibitively expensive. I would love to say this was worth all the hassle, but I can't really say that for extending a wire that, if it was the right length to begin with, would have removed 75% of the project work. 

Assembling the Boards:

• After pushing the LEDs into the boards, the only thing left was to solder them all together, 1,512 solder points in total. Thanks to the wire-ends being tinned already, this didn't actually take very long at all. The board worked, but we noticed the LEDs were fading toward the end of the string and it became far worse with two boards, so daisy-chaining power along the boards is out and we have to give each board its own power, but at least the PC PSU we are using can do this. With 8 rows of 8, the start and end points of the string are in the top and bottom right of each board respectively, so to spread the power we just ran a length of twin-core down the right side of the board connecting the now redundant power-out wires back to the power-in socket. This removed the fading problem.
• The focus was now on insulating the solder-points. Heat-shrink would be ideal obviously, but we dismissed it as too much added cost and effort so, rather than resort to tape we tried hot-glue and found it worked extremely well. It was quite messy and not the fastest technique, but cost about £3, looks a tad professional and provides a very solid insulation for the wires that shouldn't fail over time.
• The data I/O sockets are also in the right-hand top and bottom corners and, unlike the power wires, the boards connect in series so two need relocating to avoid having long data wires dangling about. We connected the boards in a clockwise order from the top left, so data from the first board exits in it's bottom right corner. The next board is the top right, but data enters at it's top right corner, so this socket needed relocating down to the bottom left corner, adjacent to the output of the first board. The third board in the series is the bottom right one, so data is coming in at the top right corner and nicely meets the output at the bottom right of the second board, but the data output needed relocating to the top left of the third board to meet the input on the fourth board, again at the top right. The last board needs no data output as it takes a serial signal and just goes to earth, but the socket still remains in case we expand the board later. The data I/O arrangement is explained much better in the diagram below.

• Finally, the LEDs need diffusing somehow as they are too small and bright to give a clear image on their own. The ideal thing is ping pong balls and we went for white semi-translucent ones, rather than the normal opaque brittle ones. This meant the balls were made of a softer plastic than usual and wouldn't have holes drilled in them, they just bent so instead required some ingenuity and a hot bolt to melt the holes, which I detail in this post.

The Arduino and Control Software:

The LEDs are all set up, but require a serial signal to tell them what to do so you will need something to interface them with the computer. We used an Arduino, which is perfect. It's a kind of programmable microchip and can be easily flashed with the right code to run a huge board of LEDs. It plugs in to the computer by USB and then hard wires to a socket that goes to the data-input of the first board so we will mount it inside there eventually and the whole thing can just plug in by USB. All you need to know about Arduino is here - .

I don't know much about the software and control, my department is the physical workage, but I can say the software we are using is called Glediator. It's some sort of freeware Java runtime and works on both PC and Mac, but (amazingly) we recommend PC for this as the Mac version is quite latent and the PC one isn't for some reason. You can download the Glediator runtime here -.