MCS Bootloader for ATMEGA168P

MadAl · Joined: 26 Jul 2010 Posts: 48

Hi,

I have been using the MCS Bootloader for the ATMEGA328p successfully. Now I tried the ATMEGA168p (as 328 is not available due to semiconductor shortage). I adapted the bootloader accordingly and managed to flash it into the ATMEGA168P. Now comes the weird part. I can bootload my firmware into the ATMEGA only once. A second or subsequent time it won't do it anymore.
If I flash the bootloader again, this process repeats (I can use the bootloader once but after that not anymore). Looks like the bootloader is corrupted by my firmware program. But it is very small (2% for testing). What could be wrong where my same routine/settings work for 328P but not for 168P ?
Fuse settings are F8 D7 EE

thanks

(BASCOM-AVR version : 2.0.8.4 )

Paulvk · Joined: 28 Jul 2006 Posts: 1257 Location: SYDNEY

Have you set the boot loader address with the $loader instruction to the fuse setting
eg 1024 words = 0xC00 512 = 0xE00 256 =0xF00 128 = OxF80

Note the Mega328 is 1024 words = 0x3C00

Regards Paul

MadAl · Joined: 26 Jul 2010 Posts: 48

I've set $loader = $1c00 as already presented in the Bootloader.bas file for this controller. You reckon this value is wrong?
In the fuse settings I also set %1c00

Paulvk · Joined: 28 Jul 2006 Posts: 1257 Location: SYDNEY

The data sheet says 1024 words = 0xC00 same as the Mega88 so it seems so.
Regards Paul

MadAl · Joined: 26 Jul 2010 Posts: 48

OK found the problems. There was an issue with the pull up resistor on the reset pin. And the program I tried compiled at 87% which apparently overwrites part of the bootloader space.

MadAl · Joined: 26 Jul 2010 Posts: 48

Value 0C00 now allows multiple bootloading firmware. However a small led flasher firmware doesn't run. When I change back to 1C00 the led flashed happily. So I guess 1C00 is the correct value.

albertsm · Joined: 09 Apr 2004 Posts: 5921 Location: Holland

the more memory a processor has, the higher the boot location. so logic demands that it is impossible that M88 and M168 have the same boot location. Because the M168 has twice the memory as the M88.
the MCS boot loader uses 1024 words. and when you look in the data sheet for the M88 you find a table. 27.8.14 which states that for the M88 the boot reset location is $C00. That is a word address since all instructions use 16 bit/
Now consider what this means : C00=3072 in dec. In bytes it would be 6144. We use an 8K processor so the boot space would be 8192-6144=2048 bytes. and this is 1024 words. that match the table and should match the fuse byte we set.

when you look in the table for M168 (27.8.1.15) for the same 1024 words you get : 1C00. that is no surprise. And you can do that math again to see it is correct.
For the M328 the location is 3C00

These locations match the values in the bootloader.bas example.

MadAl · Joined: 26 Jul 2010 Posts: 48

Correct and verified. My program is just too big to fit and is overwriting the boot sector apparently, but that is a totally different (i.e. my) problem.

albertsm · Joined: 09 Apr 2004 Posts: 5921 Location: Holland

you need mega328 ? how many and which casing?
otherwise send an email to support. as far as i can see there are plenty available.

MadAl · Joined: 26 Jul 2010 Posts: 48

I need ATMEGA328P-MMH 28 pins QFN casing

albertsm · Joined: 09 Apr 2004 Posts: 5921 Location: Holland

i see ATMEGA328P-AUR but not MMH.

SZTRAD · Joined: 30 Dec 2019 Posts: 165

Hi
if you need them badly, they have them here.

https://www.win-source.net/microchip-technology-atmega328p-mmh.html

https://www.gtelec.com/index.php?option=com_searchproduct&view=detailswithsku&virtuemart_product_sku=ATMEGA328P-MMH

http://rfq.rxicsource.com/h5/#/pages/rxicsource/quote?part=ATMEGA328P-MMH

Tough times for manufacturers and developers.

Here's a general overview of official vendor stock.

https://octopart.com/atmega328p-mmh-microchip-77760217#PriceAndStock

RS