[GeeTwo]

We always wait until we have at least a rough robot design before we start on the control board. The past two years, having a relatively large, flat control board worked well (belly pan, under the ball in 2014 Aerial Assist, and on the rear and low, but easily accessible for 2015 Recycle Rush). For AA, we arranged so the taller cRIO was to one side and the large parts of the pneumatics system were in the corners where they would not interfere with the ball. For RR, the roboRIO was low-profile, and we did not use any pneumatics.
In 2013 Ultimate Ascent, with the suddenly reduced frame perimeter and height (if you wanted to fit under the lowest bar on the pyramid, as we did) and complexity (throwing frisbees and climbing the pyramid were completely different!), the control board wasn't a board at all, but was (as described at the time by Allister, who was in charge of the control board) like playing 4-D tetris. Not only did he have to work in a cramped space, but it changed regularly as "team foundation" (drive and shooter) moved something, which displaced the climber, which in turn moved the controls around. If we'd had a preconceived control board, it would have been guaranteed not to fit.
In 2012 we DID lay out and install the control board after the drive was done but before the manipulators. We wound up cutting it in half to make room for the pneumatic cylinder that lifted each ball into the turret. Not pretty.
Besides, you never really know what needs to go onto the board - what if the controller you're planning to use is outlawed, or not needed? Do you really KNOW a priori whether you'll need pneumatics at all?
In short, design your control board like you design your robot -- for example, if you make advance plans for three or four basic drive systems, make advance plans for three or four basic control layouts. Then, wait until you've done your game analysis, strategy selection, and at least high-level robot design before you do your control system physical design.