The Dillon procedure will get you close. I'd suggest that you try both methods, evaluate them, and report back. I won't bad-mouth the Dillon recommendations on this forum, and I truly don't know which metod will produce better results - or if the difference can be measured. When using either method, you should cycle the press several times to allow the bearing to "seat", then repeat the setup.
You might also consider the Dillon alignment procedure, which uses an alignment boss and an otherwise unpopulated tooling plate to obtain perfect alignment at Station 2.
I sense that many of us are seeking improved performance of the 650 (and other presses) by reducing or eliminating the accumulated tolerances that occur in a mass-produced machine. I'm on board with that, and hope to hear from other contributors on the subject.