SGP is assuming J2000 for both solvers so that is not the right setting. If you change it to J2000, I think you will find the it agrees with PlateSolve2. SGP is expecting J2000 results and if your scope is not centering on that location you have some other problem. Again, check your scope time and location coordinates as well as your computer time. Those are involved in the conversion from J2000 to JNow. SGP uses a standard ASCOM library to do this conversion and it is well tested by many users.
An other (unlikely) possibility is the ASCOM driver for your scope is reporting it is a J2000 mount.
SGP is not the issue - there is something wrong about how your mount or driver is set up.