hi there, I once called for the attention that these objects were buggy, as they were giving inverted values.
Anyway, since they were both screwed up in the same way, the conversion back would inverted back and make everything alright, or at least sound correct.
Well, for the pd extended 0.43 I see this has been corrected, at least for the control data objects.
As for the signal objects, [cartopol~] was corrected, but not [poltocar~]. This means that it now can screw up your sound processing a lot and make some patches - that were actually "ok" before - sound bad.
the way to work around this problem is to invert the imaginary part (multiply it by -1) coming out of [poltocar~]