I found a concise article describing how LED matrices work here:
http://www.acm.uiuc.edu/sigarch/tutorials/ledarray/
This is more or less the way I've always seen it done.
Quote:
|
Originally Posted by danshaffer
one way you could try doing this is with n-bit counter chips, attaching an LED to each bit of the counter, then count up on the counter until you get to the state you want. processor should be able to do it faster than you could notice, although using 4-bit counters doesn't expand your ability too much
|
It's a pretty good idea, but instead of a counter just use a shift register (as described above).
Quote:
|
Originally Posted by JBotAlan
They have cool multi-color effects and fading in and out...I'm sure it's all pulse-width because I can see them flicker.
|
As has already been discussed, the only realistic way to build a large LED array is to scan through them using rows and columns. Once you have that in place (using just ON or OFF states for each LED), you can then implement brightness by scanning faster. For example, if you normally turn on one of the LEDs for 10 milliseconds when it's supposed to be "ON", you can subdivide that 10ms further to implement brightness. ON 2.5ms OFF 7.5ms for 1/4 brightness, ON 5ms OFF 5ms for half, etc. The challenge here is not having so many LEDs that your scan time for each LED is too low. Depending on how many LEDs to want to have total, you might have to break down the overall matrix into smaller grids and have a microcontroller dedicated to each grid. If you want color, then repeat the whole process with a different color LED. Once you do that, then you need to figure out how to communicate with all of them. As you can probably tell, the basics of an LED matrix are straightforward, but the practical implementation can get complicated.