Find all your DIY electronics in the MakerShed. 3D Printing, Kits, Arduino, Raspberry Pi, Books & more!
twittertable01.jpg
twittertable02.jpg

Jason Moungey of macetech dropped me an e-mail yesterday to let me know that the 9×9 RGB LED matrix table they were showing off at Maker Faire Bay Area 2009 (as shown above) has been upgraded with a Bluetooth shield that lets it receive tweets that control the display. Construction and programming details are available here, and there’s a live feed of the table here.

[Thanks, Jason!]

Sean Michael Ragan

I am descended from 5,000 generations of tool-using primates. Also, I went to college and stuff. I write for MAKE, serve as Technical Editor for MAKE magazine, and develop original DIY content for Make: Projects.


Related

Comments

  1. Andy says:

    The ShiftBrite and similar devices are great when you need nodes in an isolated pattern… But there are far better ways to drive matrices, such as “classic” matrix driver approaches (see Rainbowduino as one example, which can provide an 8×8 matrix for a total cost of under $1/pixel), or multichannel PWM controllers such as the NXP PCA9635 (16 channels, I2C interface, not too expensive) and TI TLC594x series (various 16 channel units with built-in current limiters, so need for individual resistors for each LED, just one resistor to set the peak current for all channels.)

    See, for example, the videos of YouTube user ekhamburg.

  2. Garrett says:

    That’s true, though there are still advantages to this approach. If you’re trying to light a large area in normal room lighting levels, do you want to use a matrix scanning approach that reduces your brightness level by a factor of eight? Or, if you overcurrent the LEDs during matrix scanning to recover some brightness, a code failure will immediately burn out a row of LEDs when matrix scanning freezes in one spot. ShiftBrites current limit each LED and maximum brightness is possible, but if something goes wrong the LED will not burn out from overcurrent.

    Next there’s the wiring and mounting issues. For every RGB LED, you have to find a way to mount them securely and then you have to solder and route at least four wires, usually more in a matrix setup. That’s a lot of wire cutting and stripping, soldering, and room for making wiring mistakes. Of course that’s a challenge some people are willing to face. All you need is lots of time and patience. ShiftBrites have LEDs and headers mounted to a PCB and we have cables you just plug together to make the chain.

    Finally, there’s the actual control. With most microcontroller matrix setups you’re not going to get 10-bit PWM on every color of every LED, there just isn’t enough processing power. Once you implement a matrix of any kind, the rest of your microcontroller code has to be very timing sensitive or you’ll mess up the display. And then matrix setups typically require a good number of I/0; at least 16 for an 8×8 matrix unless you use shift registers, but then you need even more CPU time. ShiftBrites have PWM in the modules so they’ll maintain a color you send to them, and only need four microcontroller output pins.

    Matrix arrangements are great if you aren’t too concerned about brightness control of each pixel, and if you’re looking directly at the LEDs so the reduced light output isn’t a problem. And of course the commercial LED display panels use matrix arrangements, though they typically have custom ASICs to provide the large amount of processing needed for decent PWM bitdepth and pulse overcurrent control with failsafes.

    The NXP and TI chips you mentioned have a lot of the same benefits. Except for the TLC5940, which doesn’t have internal grayscale clocking so your microcontroller has to be timing sensitive again. But we’ve got some products coming out using the TLC5947, which has 24 channels so you can control 8 RGB LEDs, and has an internal oscillator.

In the Maker Shed