I think that the data look like this, for each pair in an ET condition.

ID S ET T1 T2 T3
1 P CA  x  y  z
1 S CA  a  b  c

I don't know how to run it with Mixed, but I can suggest how
you can get the interesting results where the pairing is considered
essential - Simplify (1) by subtracting S from P to test their difference
and interaction with ET; (2) by averaging S and P to test the main
effect of ET.

Time usually deserves simplifying, too, into Mean and Linear.
- These analyses might give you something to compare against
while you are fiddling with options in Mixed.