L3D Cube Takes the Work out of Building an LED Cube

Building an LED cube usually means a heck of a lot of delicate soldering work. Bending jigs, assembly jigs, and lots of patience are the name of the game. The problem multiplies if you want to build with RGB LEDs. [Shawn and Alex] are hoping to change all that with their L3D cube. Yes, L3D is a Kickstarter campaign, but it has enough good things about it that we’re comfortable featuring it here on Hackaday. What [Shawn and Alex] have done is substitute WS2812b surface mount LEDs for the 5mm  or 3mm through hole LEDs commonly used in cubes. The downside is that the cube is no longer visible on all sides. The upside is that it becomes a snap to assemble.

The L3D cube is open source hardware. The source files are available from separate software and hardware Github repositories. Not next week, not when they hit their funding goal, but now. We seriously like this, and hope all crowdfunding campaigns go this route.

The L3D cube uses an open source Spark Core as its processor and WiFi interface. Using WS2812b’s means less I/O pins, and no LED driver chips needed. This makes it perfect for a board like Spark or Arduino.  On the software side, the team has created a Processing Library which makes it easy to create animations with no coding necessary.

L3D has all the features one would expect from an LED cube – a microphone for ambient sound visualizations, and lots of built in animations. It seems [Shawn and Alex] have also created some sort of synchronization system while allows multiple cubes to work together when stacked. The team is hoping someone will come up with a 3D printed light diffuser to make these cubes truly a 360 degree experience.

The L3D cube campaign is doing well, [Shawn and Alex] are close to doubling their $38,000 goal. Click past the break to check out their Kickstarter video!

29 thoughts on “L3D Cube Takes the Work out of Building an LED Cube

  1. BOOHOO someone is making money Boohoo. It’s open source who cares if it is on kick starter. Props to hackaday for linking to an open source kickstarter, where a non-corporate entity has a chance to make some bank to pay bills and hack some more. Hopefully when I get there they will do the same.

    If you think about it practically everything on here is an AD because if you want to build something you have to buy the components. I don’t understand the constant QQing HAD is a website I will try to visit everyday until my untimely demise at the hands of a microwave transformer, a self built killer robot or flying into a mountain with a home built jet pack.

    1. I think it’s doable. The bottom plane might be changeable to allow two strips, or maybe even redo the strips so LEDs could be placed on both sides. Back side soldering will require some part pre-gluing on reflow, but I understand it’s not extremely difficult.

  2. I bet they always film the cube from the only angle that looks good.
    And I also bet that we will see Chinese clones at half the price on Alibaba soon.
    Bur nevertheless thumbs up. The software looks nice.

    1. @Erik, we experimented with a couple diffusers. We tried hot glue blobs, putting a white heatshrink tube around each LED stick, embedding mirrors in hotglue to reflect the beam from the center of the LED, putting a plastic straw around a stick, a couple little lasercut acrylic ideas, and more.

      The way we see it, there are two big parts of the diffuser problem — the first is getting a light to diffuse evenly around a circle, which is tough. The second is to not scatter light coming from other LEDs — our first PCBs were white, and that was actually a big problem — the inside of the cube just got too darn lit. We switched to black PCBs, and that made a big difference. When I tried the white heatshrink, it did an ok job of diffusing, but it was basically the same problem as having white PCBs — lots of visual noise.

      We are very interested in improved diffuser designs. If you come up with something awesome, post it up and tell us about it at info@lookingglassfactory.com

      @Timbo, JRDM, Scott, a very straightforward way to stack the sticks is to place a WS2812 on the backside of each PCB — the footprint is the same, and the data and power connections are all right there. The LEDs are the most expensive part in our BOM, and we didn’t think the increased cost would be worth the benefit (see the explanation of the LED’s reflections below), so we only put LEDs on one side. For cube purists, it would be an easy change to the design, and the eagle files are in our hardware repo (linked from the article), if you want to give it a crack.

      @cyk, we were surprised to find that the cube does look O.K. from behind the LEDs. The light from each LED reflects a bit off the black PCBs directly in front of it. It has an interesting effect of showing the same pattern as the front of the cube, but shifted on pixel forward (because we’re seeing the light off the back of the PCB one stick in front of the LED that produces the light). It’s brightest and most visible from one quadrant of a viewing sphere, but because of the reflections, I’d say that you can see a visualization from a full hemisphere. I just shot a youtube video of a 360 degree walk around the cube, so you can judge for yourself:

      happy hacking,

  3. The fact that the 16x16x16 cube burns 1.2KW at full brightness is impressive (5V x 245A).
    The fact that the 8x8x8 cube needs 30A and they supply a 2A adapter is …. well, you be the judge of that.

    Still, this cube looks really impressive.

      1. These are smart LEDs, each has it’s own PWM generator that is free running at about 450Hz. You will have to control each LED at 25*8Hz for the small one which gets too close to the 450Hz so you will get weird things.
        they are probably relying on the fact that you don’t use the thing at full brightness on all pixels + thin wiring to achieve some sort of intrinsic current limitation, so the power supply is not overloaded.

    1. @bogdan Alex here, electrical engineer behind the cube. Your math on the current draw is correct — a WS2812 with its red, green and blue LEDs on full blast draws about 60mA at 5V, or 0.3W. There are 512 LEDs in the 8x8x8 cube, and 512* 0.060A = 33A. When we were initially designing the cube, we stacked bench power supplies in parallel so that we could source 30A if the cube demanded it. That worked, but the cube was _blinding_. Like, ridiculously blinding. We had to squint just to look at it, and it was wicked hard to see what was going on with your retinas smoldering.
      As it turned out, we had a bug in our first board design — we’d put in footprints for both a spark and an arduino, and the plan was to run the micro off of USB and power the LEDs off a separate plug-in adapter. By mistake, we’d left the spark’s 5V bus connected to the LED bus. After a couple days of working with the first prototype, I turned off the bench supply and was surprised the see the LEDs stay lit, just drawing power from my USB port. Even more surprising, it looked pretty good.
      We immediately liked the idea of just having one cord running to the cube, so we built that into our next design. As you might expect, if you told all the LEDs to turn on full white, the USB port would freak out and disconnect. So, we experimented with limiting the LED brightness in software by scaling back the “full-on” value for an LED — instead of making white (255,255,255), we’d make white (64, 64, 64). Power consumption dropped dramatically, and we could run most of our animations directly off of my mac’s USB port (drawing 5W or less), and they were quite visible in a well-lit room.
      The next thing was to cover the edge cases — what if someone told the cube to display (255,255,255) on all LEDs? Would the system freak out? I wrote a torture-test program that flashed the cube on at full brightness and back to black at varying intervals, from 1 Hz to 100Hz, and we set to work trying to make it break. We built a 1.7A current limit into the USB power bus, before it gets to the LEDs. After a couple revs of the current-limit, we were happy with the results, and we made the current limit a permanent part of the design.

      For people who really want **BRIGHT** (burners are some of our most enthusiastic customers), we basically made the entire baseboard a high-current bus that can handle 30A, and we put in a separate 4-pin header for 5V, GND, and data in/out for the LEDs that bypasses the USB current limiter.

      We ship with a 2A USB adapter, and all the shots in the video were powered with either a computer’s USB port or our 2A adapter.

      For the 16x16x16 cube, your math is also correct. In the video, I powered it with a 5V, 30A adapter. We haven’t built in a current limiter yet, because we’re still figuring out what the right max current would be (the 16x16x16 cube that we show in the video is our first and only prototype — we’re making our next iteration right now). We probably won’t ship with a 250A power supply, but there will always be the option for someone to wire in their own, higher-current power supply if they really want it.
      The way I see it, turning on 4096 LEDs on full brightness is about as fun as looking into a floodlight (actually, it’s more like looking into 7.5 500W halogen floodlights, given the higher lumen/watt output of LEDs vs. halogens — see http://en.wikipedia.org/wiki/Luminous_efficacy for more info). I view the full white cube as an edge case for how people will want to use the cube, and we’re optimizing the cube for showing LEDs that can be seen outdoors in full sunlight, has the power to go dazzlingly bright, but is a brightness that’s nice to look at. Check out https://www.youtube.com/watch?v=s6AatDesUE4&list=UUR1mPBF8raKBf0cMeBBdvfg&t=0m30s for some raw video of the big cube on my dining room table.
      I’m not making any promises before we finalize the design, but we’re looking in the 50A (250W) range for the 16x16x16 cube’s power supply.

      Anecdotally, we just set up the 16 cube on stage at an 80’s-vision-of-the-future party at a club here in Providence, and even with the 30A supply, we had to keep the brightness turned way down so the animations were visible (which you want), but we weren’t flood-lighting the dance floor (which you don’t want).

      It’s good that you’re critical of our design, and your intuition is also spot-on. Feel free to check out our design (eagle files for the 8-cube and 16-cube are linked from the main article), and if you have any other criticisms, we want to hear them.

      Happy hacking,

      1. @Alex Hornstein, @Shawn Frayne
        Completely agree with you guys, in my experience those LEDs are too bright to stare at(even one) at full brightness. I was sure you had some tricks there to make it look that great. I am glad that you made a foolproof design, please keep up with the good work.

  4. Hi @bogdan,

    Shawn Frayne here of the L3D cube dev team,

    Just want to clarify on the power issue:

    Specifically, we current limit the Regular L3D 8x8x8 Cube at 1.7A, in the baseboard PCB. And we actually turn the brightness of the LEDs wayyy down in software, to about 20% full brightness. Anything brighter at close-range is pretty blinding.

    This lets us power off of USB in a lot of cases – I’ve a cube running off the USB port in my computer as I’m typing this – although we still recommend using the included 2A adapter in case you run a visualization with greater than average demands, up to 1.7A.

    We have a bypass build into the baseboard, for freaks that want to melt retinas at Burning Man by letting their cubes suck down their max current.

    And to @cyk on the question of chinese clones – sure, but not at half the price we’re offering. Our BOM is around $120 on the first 20 cubes we made, and we’re aiming to push that down to $100 – $110 in 1K quantity – anyone else would be facing a similar bill of materials, since we’re automating a bunch of our production and don’t have high labor costs to shed. These WS2812B LEDs are expensive and until those get cheaper I don’t see the cubes getting much less expensive than we’re offering now (I doubt these addressable LEDs will get substantially less expensive in the next year – they are made by a single source, Worldsemi, and without competition there’s no macro reason I see for them to get cheaper)

    p.s. Alex, where’s the meme?!

    to the future!
    Shawn, Looking Glass

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s