> old models, say 16F84 not having them

The F84 does have them, and I think legacy and backward
compatibility is partly the reason

You'll notice that interrupt-on-change is also on PortB, which
I'm sure was originally intended for keypad interfacing. Manuals
still cite a keypad and IOC example

=================

Here's a little (hah!) problem I had yesterday, pull-up related.
Eventually I had the dim recollection of something like it before,
which led to the solution

I have a handful of 18F1320s as slaves to an 18F4550. A
12F675 provides a 1Hz /tick to all on a common line

I want to use one of these 1320s for analogue measurement,
and set it up to do so. But the 4550 wouldn't recognise that
1320 in the system. I spent quite some time with the scope and
an analyser, going through the 1320 s/w to try and figure out why
it wasn't acknowledging the 4550's identification request. I could
see nothing wrong anywhere, or physically, and it's s/w that runs
OK on other 1320s in the system

Then I had a look at the tick line, which is involved in the initial
synchronisation between all micros. I found low-going pulse wasn't
going below 2V, and therefore not being detected by the 1320,
which therefore didn't allow it to proceed to the ID section

So, this time init s/w was pulled apart. But what is there to look
at ? TRIS and set the AD registers, that's about it

         movlw   b'00101000'
;                  0          bbclk
;                   0         bbdata
;                    1        /MCLR
;                     0       Vee
;                      1      Vref+ (5.12V)
;                       0     IRQ out
;                        1    AN1
;                         1   AN0
         movwf   trisa

         movlw   b'00010011'
;                  0          tick LED
;                   0         sample LED
;                    0        n/u
;                     1       UART Rx and 10ms reference
;                      0      SDA
;                       0     SCL
;                        1    UART Tx (temporary)
;                         1   tick i/p
         movwf   trisb

         movlw   b'01000000'
;                  01          Vref+, Vss
;                    x
;                     000      channel 0
;                        0     go/done
;                         0    A/D module off
         movwf   adcon0

         movlw   b'11111100'
;                        0     AN1
;                         0    AN0
         movwf   adcon1

         movlw   b'10000000'
;                  1           right-justified
;                   x
;                    000       Tad
;                       000    clock
         movwf   adcon2

         bcf     tick          ;synch
wait_tk  btfss   tick          ;INT0IF=1
         bra     $-2

;a bit more code, then

         clrf    intcon
         clrf    intcon2
         bsf     intcon2,7   ;pull-ups off
         clrf    intcon3

Then the penny dropped. I had this happen once before and
the solution is .......

Move the intcon clearance to BEFORE the tris setting or the
whole thing just doesn't work. It seems to me that the logic
gets its panties in a bunch for some reason and a TRIS = 1
setting is ignored, making the desired pin wholly or partially
an output, which in this case was fighting the output of the
12F675

BTW, another 1320 is using 4-bit IOC, and b4 is also Rx,
which is a bit of a nuisance. Similarly some pin multiplexing
on the 4550 means it isn't a drop-in for earlier 18F45xx

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist