Please vote for me in the contests :-D Thanks
Remember this http://www.instructables.com/id/The-Much-Larger-Board-of-Many-Ping-Pong-Balls/ ?
And this http://www.instructables.com/id/The-Board-of-Many-Ping-Pong-Balls/ ?
Well this is the third generation, and very much improved, LED display.
Dat vid doe - Finally not filmed on a potato, but still filmed by a potato though :-)
First though, a bit on why it was made.
Essentially, it was because of practical problems with the previous display, shown through use at functions .etc.
1. It was not very bright at all, don't ask me how, but I managed to get the power consumption down to 1.13 Watts whilst it was still functional; I hope this shows how poor the brightness was. In professional usage, this creates many unacceptable conditions, the most severe of these is caused when other lights hit the panel and wash it out, they don't need to be pointed at it, just beam leakage and haze can do it. This washes out the panel and changes the colour balance to the eyes.
The new panels are 200W, a significant improvement, and will output 4200lm each. These are very visible even in daylight and the colour remains good even adjacent to significant arrays of movers in haze.
2. Rigging was problematic. It could be used freestanding on the floor with its stand, but anything else was a non-starter. Want it hanging from a truss? Want it in a different aspect ratio? Want it smaller? Want a different height stand? Not possible.
The new panels have standard mounting hardware on the back. They can be mounted in any orientation, in any arrangement, using standard rigging methods, such as hung trussing, tank-traps, trees, and H-bars.
3. Control was ridiculous for professional usage. I had a folder of various combinations of code. Mixtures of my transmission code, solderlab's embedded code, my embedded code, various generation scripts, and so on. Various different non-standard protocols make the control needed for professional applications impossible.
Data is streamed over Art-Net, the standard lighting data protocol for these types of panels. sACN can also be used. This allows use of industry standard software such as madrix. This is phenomenally flexible, powerful, and professional. It makes control and advanced programming extremely easy, and can be remote-controlled from a regular lighting desk with DMX.
4. Reliability and rigging time was an issue. As the units were not enclosed, rigging involved much troubleshooting (mostly wires shorting). There was not much (or really any) proper mounting for anything.
The new panels are fully tested and enclosed. They work reliably without randomly poking wires hoping to find a short :-P
5. Ping-Pong balls and wood are flammable. Whilst the panels do not get hot, other stuff in venues will, making it a big fire hazard.
With a full metal enclosure, this is almost entirely mitigated. However, for ease of hole-drilling, the front board is still MDF, but treated to be more fire retardant than the aluminium case :-)
Whilst all these issues are no problems for installation (custom mounting) or for a few occasional discos or functions (what it was designed for), it is not practical as a full-blown professional video panel.
Whilst the previous display was great for what it is meant for, professional usage requires a bit "more", as detailed :-D
Step 1: Planning and Preparation
Obviously having a smaller pixel density is cheaper, but it is generally worse. Sometimes you can want an extra low density for effect (over 100mm), but you need high-brightness, mufti-LED modules, which are more expensive. You can also on'y go so high with pixel density before the modules themselves are too big (You can use the new WS2812 leds, but they can have reliability issues, and require more complex drivers). Make sure you allow for the wires as well, as they generally come out of the modules at an angle, not perpendicular. A higher pixel density also created problems with power distribution, so bear that in mind. More pixels also make it brighter, however, this mainly applies to casting light, not visibility in direct light. This will only apply if you are far enough away for the pixels to "merge" into a continuous image (ie you can't see the individual pixels in your computer screen, but they are there. If this you are too close for this (and you probably will be if you are building your own), then you are best just getting brighter modules, however, I have never had any problems to do with visibility. And obviously you also get more resolution...
I went with P50 (50mm pitch), enough resolution for text on one panel, but low enough to get a cool effect)
If you are making them relatively small, and a lot of them, then you can use squares. I used a rectangular 4:3 ration, which is quite good for general purpose, you can have a long banner, or a slightly rectangular shape with only two of them. You can also use them on their own in either portrait or landscape to good effect. They can be separated and vertically orientated on the back of a stage for a cool effect, either side of a drum kit, separated behind a dj, and so on. It is also an ideal ratio for tings like equalizers. If it is just for one band/dj, or an installation,then think about weird shapes. Right angled triangles either side of a drummer or dj, arrows or deltas tessellating with a gap behind a stage, arches, the list goes on. Needless to say this stops them from being general purpose :-)
Smaller panels make it more expensive to cover a given area. However, it makes handling easier. For rigging speed, there is a happy medium between small and large, it is quicker to rig a smaller panel, but you need to rig more of them. I was initially going to make twice as many panels, of half the size, but it is more expensive, and would have made rigging slower. I would not go bigger than my panels, as they may not be able to support themselves gripping to a vertical scaffolding pole, they would also be to heavy for most people to lift on their own (the size I used can be easily lifted by one man, but twice as bid would require two or more(. Any bigger and it would also not fit into a cherry picker or scissor lift.
After you sort out this stuff, you need to buy stuff, more specifically the stuff below
Electronics and Electrical
Serial LED Pixels, make sure you buy spares as many will turn out to be defective
5v Power supplies (if you use the pixels from me, then it is 10A per 150 pixels)
Raspberry Pi and SD card, 1 per panel
Neutrix ethercon and Powercon TRUE1 duplex connectors.
Wire and solder (lots...)
Mechanical and Build Stuff
Plywood and fire retardant treatment (if you can't be bothered drilling a few hundred holes in sheet metal)
Square Tubing (is it only me who thinks that this is a stupid name, I even googled it trying to find a better one :-D )
Standard Scaffolding Clamps
I think that's it, but don't count on it, you may find while reading through that I missed something off (very likely), if so comment and I will change it :-)
Read about the nerdy techy stuff at
Had no pictures for this, so this is completely irrelevant, just looks cool, and is perhaps a hint to my next instructable :-D
Step 2: Thanks To
Gareth Tomas - Helped with the build, a lot
Howard Dean - The never ending source of industry knowledge
The OLA project - For their awesome ArtNet for RasPi code
An extra special thanks goes to Madrix, for sponsoring this project with £1000 of their awesome software (seriously, just download the demo and see what I mean)
Step 3: The Metalwork Bit
Unfortunately again, the shape of panel will change this bit, so I will just go over design principals.
As you will soon realize, I have absolutely no metalworking knowledge, I am just repeating like a parrot what was picked up from Al, who, in the biggest possible way, does know what he is talking about, but please forgive me for all the wrong stuff below (memory of a goldfish...) :-)
Whatever your shape is, you will need at least one square tube support in the center going the longest way, as it will be onto this that the clamps will be attached, that and it is needed structurally :-)
This should be welded to both the back and any sidewalls it touches.
Try to get as much of the sidewalls as possible on the same piece of metal as the back. (Bend the side walls in, with a lip to screw the front onto)
2-3 inches is a good length for the depth/sidewalls, as it needs to include excess wire, and the transformers.etc
Screw the transformers onto a big, thick bit of metal to act as a heatsink, and then screw that to the case.
This is best illustrated with a picture, the shortest sidewalls are separate and welded on, the longest (vertical) ones are the back bent into shape
Oh, and you need to drill lots of holed in some wood/sheet metal for the front, not a big job at all... he says
Step 4: Pixels
Next you need to add the common ground. This bit is very important, and when done poorly can be the cause of many signal problems. The best way to get an idea is to look at the second picture. Get some THICK wire (important because voltage drop will affect both power and signal) and strip intervals (same as LED spacing) with a Stanley knife (roll the wire with the knife to cut 2 circles, then slice a vertical line between the two so you can remove the piece. Run this down the longest side and solder the ground wires to it (you will need to cut the ground wire between the pixels).
You then need to do the same for the Vs line. The best way to do this is to do the same thing AT THE OPPOSITE SIDE (maintain constant voltage across the LEDs, preventing colour distortions). If you are feeling lazy, you could just use the supplied power injection wires, but there is a significant difference if you do. If a job is worth doing, it is worth doing right.
Step 5: Rasi and Transformer Stuff
I will release an SD image in a few days on my blog, so you can get one there if you want.
OLA has excellent documentation, so I will keep this short so you can read theirs. The main thing is that there is a fork called SpiMux (https://code.google.com/r/hypher-spimux/source/browse/README) created by "Hypher", that uses a shared buffer to allow more than 1 universe of DMX to be outputted to the pixels (SPI)
I couldn't get it to compile, so I made a few dodgy "hacks", that, with me being an appalling coder, I am sure some of you can do better, but in the meantime, either compile it yourself or use my image.
After this, and after you configure it, you need to wire it up. Wire the data wires from the pixels to the SPI lines of the Pi. Wire the pixels' and RasPi's power lines into the power supplies, and wire the mains from the power entry connector to the power supplies. I will leave this up to you, as your physical construction may vary, but I will say this, more screw terminals=better, allow stuff to be removes easily, and you can use the last "out" connector from the pixels to make a quick-connector for the data going to them.
Screw on the front panel, and you are done (meaning ready to test)
Step 6: And Yer Done :-D
Madrix warrants a separate instructable, but they have their own user guides which are very good.
I will leave you with some videos
2 Panel Test
My sister had a birthday party in the garage, and I volunteered to put up some lights
Below is me filming it, the video didn't come out well (note the big strobe at the bottom, just looks like camera flicker)
The panel is actually the most visible thing in the room, but thanks to my shoddy camera work, the beamlights wash out everything
The "Busking" wasn't that good either, due to it being done in the small hours
There are a few more videos on my youtube channel if you want to watch them.