led matrix

[dougyy]

I'm at the brainstorming stage of a new project, contemplating how to drive my output. The output is an led matrix (array?) 16 x 10, 30mA per led, so probably driven thru 2803's. My problem is the persistence of the led's, and how often to update the display. Basically the program would be "do a bunch of stuff" "update the matrix" "repeat". The picaxe will be a 40Xx and could use a slave picaxe for output.

The way I see the options:
1) Parallel interface using 26 output pins (16 for rows, 10 for columns). PRO, low chip count, lengthy but easy program. CON, ties up a lot of ports on a big picaxe.
2) Modified parallel using 6 outputs to a 4-line to 16-line decoder (rows) and another 4 outputs to a Johnson decade counter (columns). PRO moderate chip count, moderate program. CON long wait between updates.
3) Serial to a MAX6957, using an output off it to increment a Johnson counter for the rows. PRO moderate chip count, familiarity with serial output. CON really long wait between updates.
4) Using either option 2 or 3 while writing the data row by row into flip-flops, updating the matrix through hardware. PRO no worry about the display update. CON high chip count, providing a clock signal to the output.

Are there any easier options I'm missing? Anyone have thoughts on working with a matrix?

 

[techElder]

Don't forget the shift register method ...

 

[westaust55]

Are they single colour or RGB LEDs?

I did a project using an 8x8 matrix of RGB LEDs with multiplexing in the past using two MCP23017 i2c IO expander chips (each chips has 2 8-bit ports for a total of 16 IO each chip).
See here for ideas: http://www.picaxeforum.co.uk/showthread.php?10836/page3

 

[dougyy]

Thanks Texas,
I'd looked at shift registers, specifically the TPIC6595, but the only advantage they seemed to offer over the MAX was integrated drive transistors, while an additional disadvantage would be the need to cascade them. Is there another advantage I'm missing?

Thanks westy,
The thread you referenced did not come up in my search. The led's are mono color, a simple matrix of common anode for the rows, cathodes for the columns. The MCP23017 looks like a nice chip, seems like similar performance to the MAX at far lower cost. Are there other advantages?

Both of you have recommended serial, is my worry about led persistence needless?

 

[hippy]

The easiest and best solution for LED multiplexing IMO is a purpose designed LED matrix driver chip which can handle refreshing at the required rates while allowing the controlling PICAXE to update the display at a more leisurely rate.

Any software driven LED multiplexing scheme will put more workload on the PICAXE and require more complicated coding and complexity in achieving the desired refresh rates. DIY multiplexing, unless done directly from the PICAXE output pins, will normally involve more complicated hardware.

It can be quite hard predicting how a software LED multiplexing scheme will perform in practice ( including when data is being updated and how that's achieved with the rest of the system ) without building prototypes.

 

[dougyy]

The easiest and best solution for LED multiplexing IMO is a purpose designed LED matrix driver chip

Thanks hippy,
Do you have a particular chip you could recommend?