is there a limit in how many s~/r~ connections i an have in a patch?

I didn't think so. But you can not have multiple [s~] objects with the same name but as many [r~] as you want. For the [throw~] and [catch~] objects, it's the other way around.

yes that is clear...
this is another problem

> this is another problem
Just a guess but the order of creating [s~ ] and [r~] affects how the DSP code works. Sometimes you get problems in complex patches with feedback when the evaluation order is wrong. The only way I know to fix that is to delete the [s~] and [r~] and replace them. If you have subpatches or abstractions start at the innermost level, test each part and move outwards to the global patch.

yes...you said that before.
it seems to me very unpredictable what affects what.
f.e in one case othing worked but a direct connection with inlet~ and "wires".
then as i made an abstraction with the input in it it also worked with s~/r~.
..
the othe major prob. is to find why readsf in some " loop mode" causes cp load to rise form round 70 to 90/100 and stays there. after a while of playing.