Martin,

You're transferring the information the wrong way in the version that
doesn't work.

You probably want to say

movwf   INTCON  ; transfer contents to W to INTCON

rather than

movf    INTCON,W        ; transfer contents of INTCON to W

Cheers,

Andy

-----Original Message-----
From: Martin McCormick [mailto:martin@DC.CIS.OKSTATE.EDU]
Sent: 31 January 2000 14:39
To: PICLIST@MITVMA.MIT.EDU
Subject: Correct Way to Set INTCON


        I made an interesting discovery while writing some 16C71 code
using mpasm to write the code and mpsim to test it.  I am using the
TMR0 internal interrupt along with the prescaler.  If I set
INTCON as follows:

        bsf      INTCON,GIE
        bcf     INTCON,ADIE
        bsf     INTCON,T0IE
        bcf     INTCON,INTE

the code works as expected in mpsim.  I use e to execute it and it
runs until tmr0 rolls over and then it goes to the interrupt vector as
expected.

        To save a couple of program cycles, I tried

        movlw   B'10100000'
        movf INTCON,w

        That produces an executable that runs with TMR0 rolling over
and over without producing an interrupt.

        As I understand the documentation, the lower four bits of
INTCON are flags that one reads to see what caused an interrupt in the
first place and then clears to get ready for the next interrupt.

        Why does the two-step setup process kill the interrupts?

        At this time, the whole program is a skeleton that does
nothing neat yet.  Here is the whole main loop.

main
;initialization code which only runs on power-up is here.
    bcf      INTCON,T0IF ;Turn off TMR0 int flag.
;Safety, first.
        BSF STATUS, RP0 ; Select Bank 1
        clrw    ;Set w to 0 to make all outputs.
; initialize data direction
    movwf1      TRISB ;Macro contributed by Dwayne Reid.
        movlw B'10000000'
    movwf1      OPTION_REG ;This appears to have the proper effect.
        movlw   samplespeedset
        movfw   TMR0;Take a few counts off of TMR0.  Appears to work
correctly.

        bsf      INTCON,GIE
        bcf     INTCON,ADIE
        bsf     INTCON,T0IE
        bcf     INTCON,INTE
;Turn on the two enables needed here and make sure the other two are off.

running;In subsequent loops, the forground loop goes back to here.

; remaining code goes here

        nop ;busy work

        goto running; continuous loop for main routine
                END                     ;That's all, folks.

        Is mpsim pretty faithful to the way a real PIC would have
behaved?  I am simply curious as to why the binary load of the INTCON
worked differently than the individual setting of the bits.

Martin McCormick