I believe I've found a discrepency between tabread4~ and tabosc4~. If you use them to synthesis a signal from the same table they will go out of phase with each other. The problem is more apparent with a table with lots of harmonics.
I tested this on PD-Extended 0.41
Not sure how much of a problem this really is or whether this has been fixed in later version.
I've included a patch demonstrating the discrepancy.
Anonymous
demonstrates discrepency between tabread4~ and tabosc4~