Yes, I know that about GNURadio and the timed commands. I just
described the flowchart here.
After I generated that flowchart in GNURadio I edited the timed
commands in the generated python file, according to the code I
included in my very first message in this thread.
So the timing is definitely done properly.
I also switched to QT GUI time
sink, as you proposed, and I still observe a different phase
offset at every experiment.
I am using 3.7.12.0 version of
GNURadio.
Thanks for your help.
You might try forcing INTEGER_N tuning: