[Flashforth-devel] Using PIC18 Internal Oscillators with Flash Forth (PIC18F26K22)

[craig b. <dab...@ya...>]

There are a few modifications to FlashForth required to use the INTOSC and INTOSC-PLL Fosc configurations, as I've discovered trying to get it running in a PIC18F26K22.  


First:  The INTOSC needs to be configured as Primary Clock, PLL enabled and selected
  config bits required:
>      config FOSC = INTIO67    ;Internal oscillator block
>      config PLLCFG = ON        ;Oscillator multiplied by 4
>      config PRICLKEN = ON    ;Primary clock enabled
  This is probably best done in the processor's .CFG file for consistancy with the package.
;
:!!!!!!!!!!!!!!!!



Second:  OSCCON and OSCTUNE bits need to be set in the startup (main: in the ff18_usb.asm code)

   I made these changes:
;

main: 
        movlw    0xf 
        iorwf    ADCON1, F, A 
        clrf    TBLPTRU, A 
#ifdef USB_CDC 
        movlw    0x14 
        movwf    UCFG, A 
#endif 
#ifdef OSCCON 
  #ifdef __18F26K22
        movlw   0x70 

        movwf   OSCCON,A

        movlw   0x40        ;ena PLL

        movwf   OSCTUNE

  #else

        movlw   0x70            ; Use full internal OSC frequency 
        movwf   OSCCON, A 

  #endif

#endif 
;
:!!!!!!!!!!!!!!!!


Third problem was matching up the baudrate with higher clockrates
   Added this sequence to p18f-main.cfg and cleaned up conflicts:
;

#define CONFIG_RESET    0x0000  ; No bootloader 
    constant clock = .64000000  ; Hz 
    constant baud = .38400 
; Handle 8-bit BRG overflow condition... 
;#define    Use_BRG16    0 
; to use higher Fosc, BRG16 needs to be activated... 
#define    Use_BRG16    1 
 #if (Use_BRG16) 
    constant spbrgval = ((clock/baud)/d'4') - 1 
 #else 
    constant spbrgval = ((clock/baud)/d'16') - 1 
 #endif 
;
:!!!!!!!!!!!!!!!!


  ...This needs to be commented out if still present:

;;; Calculate the baud rate control value 
;        constant spbrgval = ((clock/baud)/d'16') - 1 

;
:!!!!!!!!!!!!!!!!



   ... Then inserted this into the WARM: sequence in ff18_usb.asm

 #if(Use_BRG16) 
;    Setup UART1 with high-speed 16-bit BaudRate Gen 
; 18F26K22 Workaround:  ANSELx FUMTU 
    movlw    0                ;load up pattern... 
    movff    WREG,ANSELC        ;...and BANG it in directly 
; set Baudrate 
    movlw    b'00001000'        ;BRG16 
    movwf    BAUDCON1,A 
    movlw    (spbrgval%0x100) 
    movwf    SPBRG1,A 
    movlw    (spbrgval/0x100) 
    movwf    SPBRGH1,A 
; TX enable 
        movlw   b'00100100'        ;TXEN + BRGH + !SYNC 
        movwf   TXSTA1, A 
; RX enable    !!!!! 
        movlw   b'10010000'        ;SPEN + CREN 
        movwf   RCSTA1, A 
        bsf     PIE1, RCIE, A 
 
 #else 
 
        movlw   spbrgval 
        movwf   SPBRG, A 
;        movlw    0 
;        movwf    SPBRGH, A 
; TX enable 
        movlw   b'00100100'        ;TXEN + BRGH 
        movwf   TXSTA, A 
 
; RX enable    !!!!! 
; 18F26K22 Workaround: This form Failed to change ANSELC Register... 
;        bcf        ANSELH, ANS11, A ; Enable digital RB5 for RX   
;        movwf    ANSELC,A        ;DOESN'T WORK, EITHER! 
    movlw    0                ;load up pattern... 
    movff    WREG,ANSELC        ;...and BANG it in directly 
;    bsf TRISC,7,A            ;appearently futile but not needed 
        movlw   b'10010000'        ;SPEN + CREN 
        movwf   RCSTA, A 
        bsf     PIE1, RCIE, A 
 #endif 

;
:!!!!!!!!!!!!!!!!


Finally, the change that tripped me up the longest was the ms-timer config for IDLE MODE.
My TMR3 default T3CON was clocking from Pin Or Oscillator which wasn't available running INTOSC.
The sysrem clock source (Fosc) works with the least change, instruction clock is /4.
  My change (immediately following above code):


#ifdef MS_TMR3 
  #ifdef OSCCON 
    movlw    h'41'            ; System Clock (Fosc)
  #else

        movlw   h'81'           ; prescale = 1 
  #endif
        movwf   T3CON, A 
        setf    TMR3H, A 
        bsf     PIE2,TMR3IE, A 
#endif 

;
:!!!!!!!!!!!!!!!!



Other chips and other config choices may require other mods (your milage may vary (grin)), but hopefully this can save someone else some of the frustrations I experienced.

Many thanks to Mike and Attila for the help, hearty exhortations, and sound advice.


Great Project!  Long may it rave.


 craig bair -- binary behaviorist

...not a speck of serial...