Working on understanding the impact of the new AP driver on the flip point caused me to think about some issues with respect to this topic. Perhaps these thoughts will be useful.
1) The geographic data in your planetarium program, the mount driver, and ACP must be the same in order to get even close to the same transit values.
2) Even when the data is the same, the transit time as provided by the planetarium program probably won't match the flip point calculations whether provided by the driver or ACP. Even being as exact as possible with the data, my experience is transit times are off 30 plus seconds. Therefore, the PRE- and POST-FLIPMARGINs need to be big enough to deal with these discrepencies.
3) Ultimately, you don't really know exactly when the mount will flip. You have ACP's view if using a V1 driver (based on the flip point setup data), or you have the driver's view if using a V2 driver. Whether either of those views are exactly right, you never really know because you never send continuous slew commands to the mount to see exactly when it does flip. With respect to using a GEM in practice, in my view, one needs to view the flip point as an estimate, not as an exact point. Therefore, give some tolerance - probably at least 120 seconds - when setting up your system with ACP (e.g. PRE- and POST-FLIPMARGINs).
4) If your mount does track beyond the flip point (usually the meridian), I would suggest taking advantage of that feature. Not only does it make imaging as efficient as it possibly can be around the meridian, the setup becomes much simplier as well. One doesn't need to compensate for the tolerance around not knowing the exact flip point. The only real issue that needs to be understood is how far beyond the meridian can the mount go before it hits something. As long as you set your PREFLIPMARGIN within that physical constraint (as a negative number), the flips will happen with no issues at all.
As a practical suggestion, just set your PREFLIPMARGIN to the negative value of your longest normal exposure and set POSTFLIPMARGIN to 0.
FWIW.
Jim