Re: [avrdude-dev] JTAG ICE3 + XMEGA32A4U = works! Now the DFU..

[Ing. Daniel Rozsnyó]

On 05/07/2013 02:36 PM, Joerg Wunsch wrote:
> As "Ing. Daniel Rozsnyó" wrote:
>
> > So.. I did managed to finally detect and program my device.
>
> Congratulations!  What's been the problem before?  Just bad wiring?

The VTG was not wired previosly. I was worried to connect that 2.5V 
(measured) pin to my 3.3V VCC... but they I tried.

Atmel's documentation is poor - there is only this:

http://www.atmel.no/webdoc/jtagice3/jtagice3.connecting_pdi.html

But no mention about what the VTG is (input or output, what voltage it 
accepts or supplies).



> > Does anybody know what fuse settings are required for the DFU
> > bootloader?
>
> Xmegas don't need that much fusing as classic AVRs.  All the clocking
> stuff can be adjusted from software.  So I guess the only thing you
> have to touch is BOOTRST (bit 6 in FUSEBYTE2).
>
> Isn't there an appnote for the DFU bootloader?

Yes, it just says to flip that bit (in text, not giving exact way how to 
do it), so in x32a4u case that is:

-U fuse2:w:0xBF:m

I can read that back too.

For programming the bootloader I take the precompiled HEX from ATmel, 
and flash it to "flash" section, all 36KB goes okay, so I suppose it is 
in correct place.

Also readback of -U boot:r:boot.bin:r gives 4022 byte of code.

Now goes the tricky part:

The http://www.atmel.com/Images/doc8429.pdf states PC3 as activator for 
my A4U bootloader. But that does not work - I could not enter the 
bootloader by that means. (I have my apps on/off button wired there). My 
test application starts immediately.

So I have tried to flash the 36K code with chip erase (-e) to end up 
with an "empty" MCU... and in that state, the bootloader detects it and 
will activate.

So half-win so far... with application present, I can not enter 
bootloader mode anymore.

Is it possible to detect, why the bootloader fails? E.g. see from 
dis-assembly if it really checks for PC3 being low ?