Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Content-description: Mail message body

I can probably figure this by experiment but since it is likely to
hit others at some stage a definitive answer from someone who really
knows may be useful to others.

I have implemented PICtoLCD using 4 byte. I needed to scroll the
16*2 line screen (new char goes to line2<15>,line2<0> goes to
line1<15>, all other chars <x> goes to <x-1> line1<1> goes to
line1<0>,lose line1,0>). I did this by reading all the chars and
rewriting them to and from LCD (I know there are probably other
better ways but I wanted flexibility).

I built this on a piece of gash stripboard, wires going everywhere
and it works AOK every time. I then rebuilt the whole project tidy on
new stripboard using the strips wherever possible and now get
spurious errors (space displayed as $ or &, P as T). Looks like if
low bit is 0000 it gets read as 0010 or 0110 occassionally (in fact
more often than not). It's gotta be the read code getting it wrong
because all the non-scrolling write code works OK and most of the
other alphabet chars are OK  in scroll mode).  Fairly sure it's not a
soldering error because built it twice now and also if I position the
board on a different insulator higher above the surface of my metal
desk the pattern of $ and & changes (get all & rather than all $). It
doesn't appear to be software because works AOK everytime on my gash
board and have tried 3 different manufacturers LCD and get same
work/don't work problem.

So I guess it's a capacitive effect of the stripboard (the 16C84
is at 4Mhz) ??  Whats the best way to fix? (touching my fingers
across the LCD pins seems to work!!!). Do the unused LCD data pins
need tying low? Do I need caps from active LCD data pins to ground?
DO I have a s'ware timing error that is borderline and works sometime
(tried all sorts of delays etc and same resurlt)

 I need this to work for a demo so will circumvemt by maintaining a
circular buffer in EEPROM  and writing LCD direct from there but am
curious to understand what might be going on here.

Will attach the LCD 4 bit read code, someone might find it useful (I
couldn't find any on the net) or someone might be able to point out a
timing error.

Thanks,
              Brian
Brian Jones
Java Technology Centre
IBM Hursley




Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Content-description: Text from file 'read.asm.txt'

Read_Data
        call     Wait_Busy              ;wait for LCD
        bsf     STATUS, RP0     ; Select Register page 1
        movlw     0xF0          ; Set port to input
        iorwf   LCD_DATA_TRIS,W ; Only set upper half of port
        movwf     LCD_DATA_TRIS
      bcf     STATUS, RP0     ; Select Register page 0
        bsf     LCD_CNTL, RS    ; Set LCD for Data mode
        bsf     LCD_CNTL, R_W   ; Setup to read
        bsf     LCD_CNTL, E     ;toggle E for LCD
        bcf     LCD_CNTL, E
        movf    LCD_DATA, W     ; Read upper nibble data
        andlw   0xF0            ; Mask out lower nibble
        movwf   Temp
        bsf     LCD_CNTL, E     ;toggle E for LCD
        bcf     LCD_CNTL, E
        swapf   LCD_DATA, W     ; Read lower nibble data
        andlw   0x0F            ; Mask out upper nibble
        iorwf   Temp,F          ; Combine nibbles
        bsf     STATUS, RP0     ; Select Register page 1
        movlw   0x0F
        andwf   LCD_DATA_TRIS,W
        movwf   LCD_DATA_TRIS   ; Set Port for output
        bcf     STATUS, RP0     ; Select Register page 0
        movf    Temp,W          ;return with read data byte in W
        return