So I'm considering flinging myself over the insanity edge and building a 32 x 32 matrix (1024 RGB LEDs). I suppose there are so many different ways of achieving this so I thought, why not ask and see what comes up.
I know of two ways of possibly accomplishing this:
a) using daisychained TLC5940 drivers (or whichever is a better fit)
b) using daisychained, individually controlled ICs, such as the WS2801.
The difference is, I can use multiples of 3 TLC5940s, one per color channel, and drive 16 LEDs at the same time, or I can use one WS2801 per "pixel". If I were to use a 16-channel TLC5940, that's 1024/16 = 64, multiplied by 3 color channels that's 192 of them. With the WS2801, that's 1024 total. Both Mouser and DigiKey has the TLC5940 for $3.07 for >100. That's a cool $614 for 200 of them, whereas I can get the WS2801 for 15.5 cents a pop, so $170.50 for 1,100 of them.
Thinking further ahead, programming. How would one drive this thing? Break it up into chunks (say 8x8 sections) and update them individually? If things are daisychained one pixel at a time (using WS2801), that can potentially take a while to update the whole string. Or do I break it up into individual strings (say 32 strings of 32 LEDs each) and update them individually?
Obviously I need to figure out what's going to drive this thing, something that can handle all the pin requirements for whichever route I go. With WS2801 ICs, if I break it down into 32 rows (32 pixels each), I'd need 64 pins total for DATA and CLK (bit banging). Feed them directly from a power supply. I haven't figured it out yet for the TLCs ... bit more math there.
This is a long, LONG term project and I plan on starting small, with one of those tiny 8x8 matrix modules before I design and built a large one. I don't want to stack 16 of those 8x8's together ... what I'm thinking of making will be larger, pixels will be further apart. Meant to go in a window and seen from a distance.
So what's the "smart" way to go about this? Can we start a discussion thread and see what the collective power of overheated, tin laded, lead fumed brain cells can come up with?