Bootloader and 3.3V VDD

Hans Wedemeyer · Joined: 15 Sep 2003 Posts: 226

I implemented a bootloader and while the loader is running my code
verifies the program code by reading it after it has been written.

The loader completes the task.
After it's done I use ICD-U40 to read the program to HEX file.

In the read file I'm finding several lines that contain only FFFF's

Here is an example.
:100280000EF0F29CF29EE824AF0FF66E020EF76EAF
:10029000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF6E
:1002A0000800F5500EBCF28C0EBEF28EF96ED6012F

I can confirm that during the programming the line starting with :10029000 would have been verified against the real code.

Yet when reading it with the ICDU-40 it reads all FFFF's

Should I be using LVP ? The chip is running at 3.3V ?

Are here other issue when using a bootloader and system is at 3.3V

Chip is PIC18LF6720

mskala · Joined: 06 Mar 2007 Posts: 100 Location: Massachusetts, USA

LVP is not related to writing Flash during program execution.

I have no problems using 3.3V on 18LF chips with bootloader.

In the read back hex file, any byte not programmed will return as FF. Check against your .lst file if there is any code supposed to be there.

Hans Wedemeyer · Joined: 15 Sep 2003 Posts: 226

mskala · Joined: 06 Mar 2007 Posts: 100 Location: Massachusetts, USA

It uses write_program_memory(). Bits from various places

SherpaDoug · Joined: 07 Sep 2003 Posts: 1640 Location: Cape Cod Mass USA

How does a bootloader work at 3.3V? According to the PIC18LF6720 datasheet page 319 table 26-4 parameter D132 you can't erase the program flash memory at less than 4.5V. How do you write a new program without erasing the old?
I have been using PICs for years, but I am just learning about bootloaders now. All the 18 series PICs I have looked at have the same 4.5V minimum for erasing program flash.
_________________
The search for better is endless. Instead simply find very good and get the job done.

Hans Wedemeyer · Joined: 15 Sep 2003 Posts: 226

Hans Wedemeyer · Joined: 15 Sep 2003 Posts: 226

Guest · Posted: Thu Sep 06, 2007 11:42 am

I´m posting this from memory, but i think that 5V are only needed to erase the whole chip at once. If you erase only a page at a time you can do it at lower voltages. This also answers a question from someone who recetly posted that he could only program the pic 32 bytes at a time: there are 2 different limits to consider when writing a bootloader: the erase block size and the write block size. Each pic has different numbers for them (damn microchip). Sometimes they are the same, sometimes the erase block is bigger than the write block, sometimes the opposite happens. You must study this information for your PIC. See AN910 for a good understanding (but not full details) about what happens when you program a PIC.

mskala · Joined: 06 Mar 2007 Posts: 100 Location: Massachusetts, USA

Hans Wedemeyer · Joined: 15 Sep 2003 Posts: 226

I tested to see if it is possible to erase program eeprom with VDD at 3.3V.

Using erase_program_memory (address) I was able to confirm it does work.

write_program_memory ( address, buf, num ) also works.

With these two items working at 3.vV the bootloader code should also work.

But it does not:

According to the CCS help file for write_program_memory()

if ( getenv("FLASH_ERASE_SIZE") > getenv("FLASH_WRITE_SIZE") )
{
then writing to the block[0] location will erase a block.
}

for the PIC18LF6720 FLASH_ERASE_SIZE is 64 and FLASH_WRITE_SIZE is 8

Test:

Already programmed locations at address 0X0100 :

I the used this statement write_program_memory( 0X0100, buf, 8 )

I then shut down and read the chip back to file using "Read To Hex File"

I expected to see locations 0X0100 thru 0X0107 contain the new values and the remainder of the block 0X0108 thru 0X013FF filled with 0XFF.

That was not the case !

So it appears the automatic erase when writing to lowest location of a block does not happen.

Unless:
I don't understand what or how a block is defined ! I assumed that
FLASH_ERASE_SIZE determined the block boundaries ! Is that a correct assumption ?

Guest · Posted: Mon Sep 10, 2007 12:36 pm

you said "it wasn´t the case" but didn´t say what happened. So... what happened? Smile

Hans Wedemeyer · Joined: 15 Sep 2003 Posts: 226

treitmey · Joined: 23 Jan 2004 Posts: 1094 Location: Appleton,WI USA

Perhaps the compiler is SMART enough to see you only write 8 bytes...
thus it reads in the 64 byte area. does the erase of the 64 byte area,((it has to do it this way)) and then write the new data. BUT it assumes the new data is the old 64 byte with your new 8 bytes.

Just a thought, can anyone check this?

Hans Wedemeyer · Joined: 15 Sep 2003 Posts: 226

Called CCS and got confirmation:

1) Yes it is possible to use write_program_memory() on a 3.3V system.

2) Only thing that cannot be changed is the configuration register which requires 5V to erase.

3) PCH V3.249 has issue with write_program_eprom() and suggested it's best to use PCH V4.055 or later.

I have no time at the moment to battle the PCWH IDE V4 debugger issues so it will be a while before I complete the boot_loader.

Thanks to everyone for the input and help.