Re: [avr-gcc-list] Mega 128 fuses overwritten by Avaraice -p

[Kasper Pedersen]

----- Original Message ----- 
From: "Iztok Zupet" <address@hidden>


> Hello,
> 
>  During programming the flash with Avarice via JTAG ICE the fuses on the chip 
> were all overwritten with 0xff, thus disabling the JTAG and SPI programming 
> interface efectivelly.
..
>  Has anybody encountered something similar?

We saw something similar, but we don't have any solid evidence either.
Our board has 3 mega16's in a chain, and we have an in-house jtag loader
that works well with our board.
When we attached this board to the ice, it would persistently give us a message
to check the connection and signal levels*, and it too managed to kill the jtag
fuse on the third mega16 in the chain.
Also, it would state that download went well, but in reality it didn't.

We have an external watchdog on the third mega16, and IF that watchdog
is sending resets to the cpu during download or debug, things go awry.
When we disabled the watchdog, the ice worked as expected, and hasn't
killed any more mega16's.

I don't understand what's happening, because the jtag loader I wrote works
happily no matter what's going on on reset.

/Kasper

*by the way, the TCK signal out of the ice sometimes has an ugly rising edge,
depending on what operation it's doing. A 30 ns rc-filter (or just 220pF to gnd)
will do.