On Thu 15 Jan, Keith Howell <keithh@ARCAM.CO.UK> wrote:
> Subject: Re: Which Port For I2C?
>
> No, it really is okay to use bsf and bcf to twiddle SDA and SCL.
> I do. I twiddle the bits in the TRIS registers.
> AFAIK this does not disturb other bits in the register, or I/O bits.
> So I2C events do not disturb other I/O bits.

True, but it is _the_other_way_around_ that might be giving headaches,
like having an interrupt routine twiddle some bits on the same port as
you're using for I2C, right in the middle of an I2C transaction.

[snip]

> BTW, the last chap's I2C file looks flaky for a start.
> Look at the code:

Look again.

> > SendI2CStart
> >       movlw   ~(SDA_MASK|SCL_MASK)   ; make sure output
> >                                       ; lines are low
> >       andwf   I2CPORT                 ; when outputs are enabled
>
> There's nothing to guarantee the timing.
> The I2C spec says the start condition is SCL falling when SDA is low.
> AND the time between SDA falling and SCL falling must be
> at least 4 microseconds!

[snip the macros]

> So that you do a START like so:
>
>         page1                           ; meaning data page
>         I2C_SDA_LO                      ; SDA goes low during SCL high
>         doWait3microseconds             ;
>         I2C_SCL_LO                      ; Start clock train
>         page0                           ; return with data page == 0

But you never set the state of the data register!!
OK, since you are so eager to point out that my code is wrong, here is
the full routine (heavily macro'd as usual).  Note the two 20us delays;
these are to make life easier on PIC slaves, and also go guarantee the
required I2C bus timing.  This code runs on 12 and 14 bit PICs.


; +---------------------------------------------------------------------+
; |                                                                     |
; |   Send I2C start condition.                                         |
; |                                                                     |
; +---------------------------------------------------------------------+
;
; In:   -
;       SDA and SCL floating high
; Act:  Generate a slow but otherwise normal I2C start condition.
; Out:  W=0
;       SDA pulled low, SCL floating high

SendI2CStart    movlw   ~(SDA_MASK|SCL_MASK)        ; make sure output lines are
 low
                andwf   I2C_PORT                    ; when outputs are enabled

                if CLOCK_RATE>=600000
                ldw     CounterI2C,CLOCK_RATE/(3*200000)
                loop    CounterI2C,$                ; 20us delay
                endif

                ifdef __14BIT__
                bsf     status_rp
                bcf     SDA                         ; pull SDA low while SCL is
 high
                bcf     status_rp
                else
                ifdef TrisCopy
                ld      TrisCopy
                andlw   ~SDA_MASK
                str     TrisCopy
                else
                movlw   SCL1_SDA0_TRIS
                endif
                tris    I2C_PORT                    ; pull SDA low while SCL is
 high
                endif

                if CLOCK_RATE>=600000
                ldw     CounterI2C,CLOCK_RATE/(3*200000)
                loop    CounterI2C,$                ; 20us delay
                endif

                retlw   0


; +---------------------------------------------------------------------+

Note:   LDW is MOVLW <constant>:STR <reg>
        LD is MOVF <reg>,w
        STR is MOVWF <reg>
        LOOP is DECFSZ <reg>:GOTO <label>


Of course, there is nothing in my code to protect the I2C from rogue
interrupt routines doing RMW instructions on the same port.

Frank
------------------------------------------------------------------------
Frank A. Vorstenbosch                           Phone:  +44-181-636 3391
Electronics & Software Engineer                 Home:   +44-181-544 1865
Eidos Technologies Ltd., Wimbledon, London      Mobile:  +44-976-430 569