My bias, after years of putting together historical battle scenarios: Why ever have to use rules with artificial 'fall off the map' rules and limits? Practically, of course you may need them even with a larger board size.
Tony, I have come up with something similar to your thinking. I have a set-up template that I use to take a 4x6 or 6x8 table and turn it into a number of 'table areas.' (I got the basic idea from some creative type in your group.) The rest of the table I use to put good looking terrain on (a town, a castle, a river, a seashore with ships, etc.). I've been using this with historical battles like Telamon the SoA Battle Day game (a very confusing battle facing multiple directions.)
Nice if you have the space. Most may struggle with that. One key thing v3.0 did was to challenge player deployment. It is harder to have a great deployment. I interpret this as saying v3.0 has the first bound or two as the actual final movement of troops into their attack/defence posture. With big boards you could lose that.
Remember, the board edge can represent standing orders, logistical issues, doctrine and the battle plan just as much as actual geography.
DBA is a subtle game. Easy to mess with. Hard to significantly improve.
... Remember, with the roads as is, you can pretty much force your attacker on a 30" board to deploy in a quarter that is packed with bad and rough, due the relatively more space to place multiple features in the same quarter.
I don't understand how the road forces the attacker to choose the side with the "bad quarter"? Doesn't the defender still have a binary choice with a road? Have I missed something?