The good news: after some days of rewriting, debugging and struggeling with sdcc-bugs, the communication-part of the firmware appears to work very stable and doesn't create strange packets any longer.

but it was a bloody deed. I had to go through assembler-code to find out if the compiler switches memory-banks right (sometimes he doesn't, which can lead to "nice" results: e.g. debugging output of variables prints the wrong variable, ...). i also had to learn that interrupts could overwrite my local variables ...

yet, this part is working now. also the motors are syncing fine, thanks to dan's hints and after some reading and measuring and changing the firmware to get infos, I found out that one of my board had a wrong connection: two-lines were placed to close.
also the jumper for the max-sensors were set wrongly and the plug of the connecting-wire for the sync had a slack joint.

not enough I destroyed my extruder again :(
I had to open it for cleaning and by doing so I broke the nichromwire. %&$. as i can't find that wire in austria, I have to order a new one ... oh well.