PIC Table Write

Do you have a question? Post it now! No Registration Necessary

Translate This Thread From English to

Threaded View
I am attempting to test the table read/write of the PIC18f2431.  I
understand that tables are written 8 bytes at a time meaning that the TABLAT
register is written to 8 times.  I start by setting the table pointer as
follows:

MOVLW        0x00
MOVWF        TBLPTRU
MOVLW        0x10
MOVWF        TBLPTRH
MOVLW        0x00
MOVWF        TBLPTRL

This should set the pointer to address 0x1000.  Next I load the TABLAT
register 8 times such as:

MOVLW        0x##
MOVWF        TABLAT
TBLWT*+

*
*
*



I then set the EECON1 and EECON2 registers as follows:

BCF                EECON1,CFGS
BSF                EECON1,EEPGD
BSF                EECON1,WREN
BCF                INTCON,GIEL
BCF                INTCON,GIEH
MOVLW        0x55
MOVWF        EECON2
MOVLW        0xAA
MOVWF        EECON2
BSF                EECON1,WR
NOP
BCF                EECON1,WREN

At this point the 8 bytes that I transfered to the TABLAT should be written
to program memory at 0x1000 right?  When I attempt to read back the data, I
set the table pointer to 0x1000 just like I did before the write sequence.
I use the TBLRD*+ and then read the contents of the TABLAT register.  I get
0xFF.  When I download the program memory from the PIC into MPLAB, I can see
that the data was written starting at 0x1008.  What's strange is that the
last byte I loaded into TABLAT before the write appears at 0x1008.  The byte
at 0x1009 contains the first byte that I wrote to TABLAT and then the
sequences continues as expected.  Is this correct?  I would have expected
the first byte that I transferred to TABLAT would appear at 0x1000, the
second at 0x1001 and so on.



Re: PIC Table Write
Quoted text here. Click to load it

That is correct. The example code in the 18Fxx31 datasheet is wrong. Have a
look at the 18Fxxx2 datasheet for the correct code.

The low 3 bits of the TBLPTR registers is used to address the 8 holding
registers during a short write. On the actual (long) write, the upper 19 bits
hold the block of 8 bytes to write to. You code uses post-increment:

Quoted text here. Click to load it

so after 8 writes, TBLPTR ends up pointing at the 9th byte, in other words the
first byte of the second block of 8 bytes and that is where the holding
registers get written to.

The code is the 18Fxxx2 datasheet has two important differences. Right after
the erase is complete, a dummy read with auto-decrement is done to move the
pointer back one byte.

TBLRD-*

In the loop writing to the holding registers, pre-increment is used:

TBLWRT+*

So, after all 8 bytes are written, TBLPTR is still pointing into the correct 8
byte block so the holding registers get written into the right place.

You might look to see if there are any errata sheets for the 18Fxx31 datasheet.
Microchip might have caught the error already.


Re: PIC Table Write
Quoted text here. Click to load it

Actually, I would think that 0x1008 would have the first byte. Hmm.


Site Timeline