On Sat, 18 Apr 1998 22:19:07 +0200 marcel <marbel@XS4ALL.NL> writes (many
times):

>> I still have great difficulty on the folowing
>> it seems carry does not affect high byte
>> i am also not shure on testing for zero
>> i realy need help on this.

Subtraction on the PIC tends to confuse me.  I generally just use two
routines: "add" and "negate" (two's complement).  The "add" routine uses
the recently-discussed 'add with carry' sequence:

; Add 16-bit numbers A and B together, result in B (B=A+B).
        movfw   al
        addwf   bl,f
        movfw   ah
        skpnc
        incfsz  ah,w
        addwf   bh,f

The two's complement routine is straightforward as well:

; Negate (2's complement) 16-bit number in B (B = - B)
        comf    bl
        comf    bh
        incf    bl,f
        skpnz
        incf    bh,f
[Note that this routine can be extended to more bytes by adding another
complement instruction to the top and another incf / skpnz at the bottom.
 But that isn't important right now.]

Using two's complement representation, an ordinary add will combine
positive and/or negative numbers correctly.  Positive numbers are just
binary numbers with the most significant (sign) bit zero.  Negative
numbers have the sign bit of one.  This makes it very easy to tell if a
result is positive or negative:

        btfsc   nh,7            ;Test high bit of n (sign bit)
        goto    num_neg         ;It's negative
        goto    num_pos         ;It's positive


The other bits in a negative number are not exactly the value in binary.
The "negate" routine will of course turn a negative number into a
positive one, also clearing the sign bit.

So to implement the "slew rate limiter" logic, do something like this
(not optimized at all):

A = oldval
B = newval
call negate  (B = - newval)
call add  (B = oldval - newval) -- Difference in reading
C = B (save difference for later)
If B positive, call negate (B = -|B|)
A = max change (0x80 in this case)
call add (B = max change - |difference|)
 [If B negative now, difference is too large.  Setup to alter oldval
  by either the actual difference (C) or the maximum depending on if
  the difference is too large]

[If B negative]
A = oldval
B = max change
If C positive, call negate (make B = - max change)
call add
oldval = B

[If B positive- difference OK.  Note if you still have a copy of newval,
it could just be copied.]
A = oldval
B = C (oldval - newval)
call negate  (B = newval - oldval)
call add (B = oldval + newval - oldval = newval)
oldval = B

This should work, or at least get close.

_____________________________________________________________________
You don't need to buy Internet access to use free Internet e-mail.
Get completely free e-mail from Juno at http://www.juno.com
Or call Juno at (800) 654-JUNO [654-5866]