Hans de Vries wrote:
>
> I've created (yet another) 4x4 matrix keyboard decode-routine.
> I've placed 4 NOP's in it, because in a similar routine, they also were
> there (bad reason) , but I'm not sure if they are needed.
> They are there to let the output-pins when the pattern on portb is
> changed. But are they nescessary?


There are several factors that affect this decision, however just
to be safe I would leave them in but change them to goto's:

>         MOVWF   PORTB                   ;Put mask on port
>         NOP                             ;Let pins settle
>         NOP                             ;But is this nescessary??
>         NOP
>         NOP
>         NOP
>         NOP
>         BTFSS   PORTB,0                 ;IS first bit clear?
>         RETURN                          ;Yes, so that row was pressed


>         MOVWF   PORTB                   ;Put mask on port
          goto    $+1                     ;Let pins settle
          goto    $+1                     ;But is this nescessary??
          goto    $+1
>         BTFSS   PORTB,0                 ;IS first bit clear?
>         RETURN                          ;Yes, so that row was pressed

So you get the same delay for half the instructions. Another trick
to get the 6-cycle delay is to do this:

>         MOVWF   PORTB                   ;Put mask on port
          CALL    a_return                ;Let pins settle
          goto    $+1                     ;But is this nescessary??

>         BTFSS   PORTB,0                 ;IS first bit clear?
a_return: RETURN                          ;Yes, so that row was pressed


If you have other delays then they can be grouped together:

delay_6:  NOP
delay_5:  NOP
delay_4:  RETURN

And then if you need a delay of 6 cycles, just add a 'CALL delay_6'
>         MOVWF   PORTB                   ;Put mask on port
          CALL    delay_6                 ;Let pins settle
                                          ;But is this nescessary??

>         BTFSS   PORTB,0                 ;IS first bit clear?
>         RETURN                          ;Yes, so that row was pressed




Payson posted something similar to this a good while back. Perhaps
you care to elaborate John?

Scott