ivp wrote:
> I've got a 5 x 5 LED matrix, which is driven by an 18F43K20 on
> a one-LED-per-pin basis

I assume the PIC is neither RAM, nor code, nor cycles limited in this
application?  If so, optimize for maintainability.

If the customer is already asking for a few different brightnesses, I'd be
nervous hard coding just those few.  Sooner or later the specs will change.
I think (knee jerk warning) I'd make the underlying mechanism with a byte
per LED.  This byte holds a brightness value.  You certainly don't need to
use all 256 brightnesses, and that would make the blink rate too slow
anyway.  Maybe brightnesses 0-8 would be good enough.  LEDs don't have to b=
e
separately turned off, just set their brightness values to 0.

The interrupt routine handles the conversion from brightness to the on/off
decision every PWM slice interrupt.  Let's say you want to flash the LEDs a=
t
500Hz and use 9 brightness levels (0-8).  That would mean the interrupt rat=
e
is 4KHz.  At 10MHz instuction rate that's every 2500 instructions, which
should leave a decent amount of the processor to the foreground code.  The
interrupt routine maintains a 0-7 counter that is incremented every
interrupt.  Any LED with a brightness value exceeding the counter is lit
that interrupt.

I'd probably deal with the mapping of LED number to port pin in a
semi-automated way.  You really want this mapping to be in only one place i=
n
the source code.  I'd give each of the LED pins regular names in the /OUTBI=
T
directive.  Then I'd use preprocessor math and looping to write the specifi=
c
instructions for each of the 25 LEDs as one unrolled loop.  That code will
be ugly, but you won't be maintaining it directly.  You only maintain the
single copy that gets unrolled automatically by the preprocessor loop.  Eac=
h
loop iteration fetches the brightness value for that LED, compares it to th=
e
current interrupt count, then sets or clears the associated LAT bit
accordingly.

The preprocessor (unlike the MPASM macro processor) can do real string
manipulations to build new symbols.  It would loop over the LED number
(0-24) and build the symbol name for the LAT bit accordingly.  The
brightness values would likely be in a regular array, so you load FSR0 once
then use POSTINC0 to fetch each successive brightness value.  Don't forget
to save/restore FSR0 in the interrupt routine if you use this method.

If you're tight on program memory this is not the best way.  Otherwise, it
is nicely maintainable, the odd mappings in only one place, and about as
fast as possible.

The foreground code then just stuffs brightness values into the brightness
locations for the LEDs as desired.  It doesn't and shouldn't be dealing wit=
h
the arbitrary pin mappings.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.
--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.