PIC Microcontoller Radix Math Method

ASCII Hex to Binary

; From Regulus Berdin; untested
; Input  - ASCII number in W
; Output - binary in W

         sublw   '9'
         movlw   'A' - .10
         skpnc
          movlw  '0'
         subwf   ASCII,w

Tracy Smith says:

If you KNOW that the ASCII is '0'-'9','A'-'F' then a simpler solution for the midrange pics is:

    addlw   -'A'
    skpc
     addlw  'A' - 10 + '0'
    addlw   10

If the ASCII value is in ram, then this solution will work for the 12 bit core too:

    movlw   -'A'
    btfsc   ASCII,6
     movlw  -'0'
    addwf   ASCII,f   ;(or w)

Code:

• I think the correct conversion code from Tracy Smith should read :
  
    addlw -'A'          ; expect upper case letters
    skpc
    addlw 'A' - 0x18 -'0'
    addlw 0x11
  

  +
• Samuel Tardieu shares this code: For indirect addressing case-independant :

  | bcf     0x00,5
  | movlw   0xC9
  | addwf   0x00,f
  | movlw   0x27
  | btfss   0x03,0
  | addwf   0x00,f
  | return  

  +

See:

• /techref/microchip/math/radix/b2bu-16b5d.htm PIC Microcontoller Radix Math Method Binary to BCD unpacked 16 bit to 5 digit+

Peter Heinrich Says:

;;  Converts an ASCII character code (in W) into the integer value
;;  corresponding to the hexadecimal digit it represents.  '0'-'9'
;;  become 0x0-0x9; 'A'-'F' and 'a'-'f' become 0xa-0xf.  (This routine
;;  expects W to contain only valid hexadecimal digits.)  The result
;;  is returned in place in W.
;;
char2int:
   ; Shift the character.
   addlw    0x9f
   bnn      adjust               ; if positive, character was 'a' to 'f'
   addlw    0x20                 ; otherwise, shift to next range of digits
   bnn      adjust               ; if now positive, character was 'A' to 'F'
   addlw    0x7                  ; otherwise, character must have been '0' to '9'

adjust:
   addlw    0xa                  ; shift the result to account for the alpha offset
   andlw    0xf                  ; clamp the value to one nybble
   return

Comments:

• I read this last suggestion to use this instruction and ended up spending a few days trying to figure out why it is unpredictable. This does not work (at least not on a 12F629 *and* the assembler does not even generate a warning):
  
  btfsc WREG,6 ; not all pics can do this...
  
  That assembles to 1A00 == "btfsc 0, 6" and since W is a SFR (special function register 0) and NOT a File Register which btfsc operates on, it tests the correct bit in the wrong register. "btfsc 0,6" references INDF (File Register 0, the register pointed at by FSR), not W.
  
  There is no place in the instruction encoding for this bit test instruction to specify the difference between WREG at 0 and INDF at 0. so that code "btfsc WREG, 6" really executes as "btfsc INDF, 6". Compact, but broken.
  
  best regard from Barry L. Kramer in Mohnton PA
  +

Questions:

• tamas.rudnai~NOSPAM~ at SPAMgmail.com asks:

  This one does not care if the A-F is upper or lower case:
  
  

  hex2bin ; expects '0'-'9','A'-'F','a'-'f' in W
  		btfsc	WREG,6  ; not all PICs can do this...
  		addlw	9	; that's A-F (0x4? or 0x6?)
  		andlw	0x0F    ; we need only the low nibble
  		return
  

  
  Yes, that's 3 instructions + the return :-)
  Tamas
  

  +

  .