Personally, I still think the station should go above Sunrail's, somewhere near Landstreet Drive along MCO's northwestern edge, with DMU/EMU-based peoplemover connecting the station to the main terminal itself (directly above the ground transportation & rental car center, initially running on FEC's tracks with Sunrail's trains & staff to keep the cost sane & something FDOT/Orange County/MCO can stomach).
IMHO, direct accessibility to the rental car center and ground transportation is important... but in the real world, passengers *expect* to ride a peoplemover to get from rail to airport terminal. People get upset and massively turned off by shuttle buses, but think nothing of transferring from one train to another for "the last mile". Train-to-mover is the expected norm. Train-to-mile-long-walk sucks. Train-to-bus is just plain ghetto, and will instantly turn off a huge number of potential customers.
Now, let's look at the flip side: people getting dropped off, picked up, driving their own car to/from the station, or transferring to/from Sunrail.
The most time-sensitive passengers are going to be those heading to/from Sunrail, because even under the most ideal London-like headways imaginable, the FEC and Sunrail trains *might* be on 10 or 15-minute headways someday. Throw a long (10+ minute) trip via bus or van between stations (if FEC were in its own spur at the terminal, and Sunrail were along the edge), and you've just forced those passengers to budget an extra 30 minutes of travel time if they want to make sure they don't miss their connection. Epic fail, angry customers.
Yes, people going from Train to Plane are time-sensitive too... but to be honest, I really don't think there will be a lot of people LITERALLY going from Plane to Train & vice-versa. Rail passengers care about MCO because it means cheap 24/7/365 rental cars and abundant ground transportation options. Think about it for a minute... would you EVER buy a nonrefundable train ticket for a departure at some specific time after your flight was *supposed* to land, knowing what we all know about air travel delays due to weather, security, or whatever? Of course not. You land when the plane does, make your way to the station, and crawl onto the next departing train, whenever it happens to be. Train-to-Train transfers are more time-sensitive, because a well-run rail network really IS precise enough to let you get away with 5-minute transfer windows when planning a trip (in Japan, at least).
Now, think about the passengers getting dropped off, picked up, or parking their own cars. Why, exactly, should those cars go anywhere *near* the main terminal? When you compare the cost of adding new capacity to the roads into and out of MCO, and enlarging its garages, to the cost of just giving the rail passengers THEIR OWN parking facility, it makes more sense to just keep them separate. Siphon them off 528 at Boggy Creek Road, and keep them segregated from air travelers. God knows, there's no real shortage of land around MCO. Build a huge, sprawling surface parking lot for FEC & Sunrail around the new station, then gradually replace it with infill construction and new parking garages in the future (like they did at Metrorail stations south of downtown Miami).
To cut costs, the MCO "peoplemover" (really, a conventional DMU or EMU) could share tracks with the FEC mainline (with spur down into the terminal itself), and be operated by Sunrail using Sunrail's trains & maintenance facilities. Interference with FEC intercity trains would be a non-issue, because "98%" of them are always going to end up stopping at MCO's station, so any peoplemover-DMU sharing the 2 miles of track between Sunrail and the spur into MCO would be sharing tracks with intercity trains that are going to be accelerating or decelerating *anyway*.
Someday, when there's enough passenger traffic to justify the cost, the peoplemover could be given tracks of its own and be severed from FEC's mainline. In the meantime, sharing tracks, trains, and maintenance facilities with FEC & Sunrail would probably save FDOT & Orange County/MCO at least a half billion dollars in up-front construction & operating costs. Yeah, it might be kind of gross sledgehammer-like overkill to run a DMU back and forth between two stations ~2 miles apart all day, but it makes sense when you consider that the alternative is a billion-dollar peoplemover (with its own staff, proprietary equipment, and maintenance facility) like the one Dade County had to pay Fujitsu to exercise monthly on its test track in Japan so it wouldn't get weak & flabby in storage. Just share the tracks, use compatible trains, pay Sunrail to dedicate 1 or 2 of their trains to the task, and pocket the savings.
Anyway, in summary, I think putting the FEC station itself in either downtown Orlando (presumably Amtrak's existing station) or semi-adjacent to the airport terminal itself would be a mistake. People literally heading to/from the terminal would STILL have one hell of a hike to get between the two (especially if it were built south of the existing terminal), and it would inconvenience everyone ELSE using the train in perpetuity -- adding 5-10 minutes to the trips of anyone NOT going directly to the airport, and 15-30 minutes to the trips of anyone transferring between FEC and Sunrail. And Sunrail would STILL be cut off and isolated from the terminal (or worse, detoured and delayed by a trip of its own into the main terminal, making Sunrail useless for travel between Kissimmee and downtown). The only viable way Sunrail itself could ever run directly into the terminal would be if they ran it as two lines... one that ran from Deland to MCO, and another that ran from Kissimmee to Deland (without the airport detour, with the assumption that the few people taking it from Kissimme to MCO would transfer to a southbound Sunrail train at the first station north of the MCO spur).
Likewise, running FEC trains to the existing Sligh Blvd. Amtrak station downtown would be a mistake, too. Running trains there INSTEAD of stopping at the airport would be suicidal -- poor rental-car availability, station that just SCREAMS "run down", and a location miles away from where most people who'd be taking the train from Miami to Orlando actually want to *go*. Running trains there in ADDITION to the airport would also be a mistake, because it will create the expectation of eternal service to downtown Orlando, and subject FEC to political pressure (since ultimately, they're depending on FDOT to get them the ROW for their new tracks between Tampa & Cocoa) to keep running trains there, even though it's 10 miles out of the way for trains heading to I-Drive, Disney, and Tampa. It's far better to just not run trains there at all, and encourage Florida to get Sunrail running from a nice, convenient, shared station connected to the airport and downtown via direct rail link instead.
Come to think of it, I can think of another possible "5-year plan" to launch service by 2014, without totally screwing their long-term viability or wasting lots of cash. They could build new tracks along 528 from Cocoa to the existing tracks that run from the power plant to MCO & Sunrail, then limp along those existing tracks for now to the southern access road into the airport. There, they could build a temporary station (concrete platform, canopies, trailer bathrooms, gravel parking lot) adjacent to the airport's access road to use for 3-5 years, surrounded by lots of renderings of their REAL airport station that would be under construction a few miles away. Instead of having to rush things through and build a half-assed FDOT-style compromise that makes nobody happy and costs 16 times as much as it should, they can launch service quickly with the new temporary station to the south, then take a few years to plan and build the new tracks and station along the northwestern edge of the airport *right*.
As an added bonus, having solved the immediate problem of getting passengers to MCO and its rental car center, they could hold off on the last segment of new track along 528 to the airport until they knew beyond doubt what route they're going to take to I-Drive, Disney, and Tampa & have the exact easements nailed down and guaranteed for them by FDOT, instead of building 200 million worth of tracks, then running the risk of having FDOT say, "whoops, you can't go that way after all... you'll have to either do something awkward to use the tracks you already built, or abandon the new station you just built and the last 5 miles of track leading up to it, and build a new station along a new route instead..."