Writing to Flash Memory on a 16-bit PIC® MCU

Overview of Writing to Flash Memory

Flash memory panels on 16-bit PIC^® Microcontrollers (MCUs) are divided into pages. Each page consists of a number of rows. Each row has a set of 24-bit program words. Page and row sizes differ among the MCUs. Consult the datasheet to determine the page and row sizes for the PIC MCU you are using.

The implementation details of flash pages and rows are invisible to the application program. Rows and pages have no effect on program control or where the linker places program objects. Pages and rows are only taken into consideration when the application writes to the flash memory at run-time.

Writing to flash consists of loading the data to be written into a holding latch (or series of latches) using Table Write instructions. The flash address to be written to is loaded in the Non-Volatile Memory (NVM) address register. Once the data and address are loaded, a specific sequence of register operations will initiate the writing process. The writing process transfers the data from the holding latches into the Flash memory. During the writing process, the MCU stalls, resuming once the write operation has been completed

​​​​

All 16-bit PIC MCUs allow the application program to specify individual Flash memory words to be written. Using this word addressed write method, 24-bits of data are loaded into a write latch with a pair of Table Write instructions (TBLWTH and TBLWTL). After the latch is loaded, the applications load the destination address into the NVM address registers. After both sets of registers are loaded, the application sets the control bits, then executes the write sequence on the NVMKEY register. The write sequence causes the data in the latches to be written to the Flash address specified by the NVM address registers.

Some devices provide the ability to program an entire row at one time. To program an entire row, the application loads a series of holding latch registers with repeated Table Write instructions. The applications then load the first address of the row to be written into the NVM address registers. After both sets of registers are loaded, the application executes the write sequence. With row programming, an entire row is written to flash at one time.
Back to top

Resources used in writing to Flash Memory

Instructions

Back to top

Registers

TBLPAG Table Page Register - TBLPAG is used by the Table Read and Table Write instructions. This register specifies the upper 8-bits of the address to be accessed by the table instruction. The lower 16-bits of the address are passed to the instruction as a value in a working register. When a Table Read instruction is to be executed this register contains the page address of the word to be read. When a Table Write instruction is to be executed TBLPAG should be loaded with the base address of the holding latch(es) (#FA00 on many 16-bit PIC MCUs)

NVMADR/NVMADDRU NVM Address Registers - Holds the address of the flash memory word (or row address) which will be written to when the program sequence is executed. After loading the holding latch(es) the application sets the destination address in NVMADR/NVMADRU before initiating the write sequence. The write sequence causes the data to be copied from the latches into the flash memory location specified by NVMADR/NVMADRU.

NVMCON Non-Volatile Memory Control Register - The register through which the application selects the function to be performed when the program sequence is executed. The options in this register include Erase Page, Write Word, and Write Row. NVMCON also contains the bit to be written to start the write process (WR).

NVMKEY Non-Volatile Memory KeyRegister - To reduce the potential for Flash to be modified by a spurious write to NVMCON WR cannot be set unless a specific sequence of data values are written to NVMKEY. This sequence is commonly referred to as the program sequence. Once the sequence is executed WR becomes 'write-able' for 2 instruction cycles.

Back to top

Program (Unlock) Sequence

The program sequence for a 16-bit PIC Microcontroller (MCU) is performed by writing specific values, in successive order, to NVMKEY. Writing 0x55, immediately followed by 0xAA, will unlock WR for two instruction cycles. If WR is set within two instruction cycles of the programming sequence, the MCU will perform the flash self-write process.

Interrupts should be disabled during the write program sequence to prevent servicing an interrupt between the two write instructions.

The unlock sequence can be written in C so long as the compiler places the two write instructions in sequential order. The MPLAB^® XC16 compiler provides the function builtin-write-NVM to perform this program sequence. After executing the program sequence, builtin-write-NVM sets WR which initiates a flash write.

Back to top

Using MPLAB^® XC16 C Compiler

​​​​

Using Assembly Language

​​​​

Algorithm to Self-Program Flash Memory on a 16-bit PIC^® MCU

The 16-bit PIC MCU reference manual advises that a row or word in Flash program memory should not be programmed twice before being erased. Given this erase-before-programming mandate, the following application flow is commonly used:

Copy the Page To Be Updated Into Data Memory

• Load the base address of the Flash memory page into TBLPAG.
• Using successive TBLRDH and TBLRDL instructions, copy the contents of the Flash page to data RAM. Care should be taken when copying 24-bit wide Flash memory into the 16-bit wide data memory. One word of Flash memory requires two read instructions (TBLRDH and TBLRDL). These two read instructions will return a total of 32 bits of data.

Back to top