Re: Precision 3D Printer ... Revisited
+1 on opposite corners. But consider the rotational axes as well... There are 6 axes to constrain here, and you want the threads to only constrain the one that they're supposed to move in - i.e. linear z. Given the tight tolerances you're aiming for, may as well have a good crack at them all.
X,Y, rZ - (over)constrained by linear rods
Z - constrained by one threaded rod (with 2 you have to keep them aligned well, as it's technically overconstrained)
rX, rY - perhaps split these up a bit differently, imagine two diagonal axes from corner to corner. Assuming you have the diagonal corner threads, rotation about the line between the two smooth-rod corners would be constrained by the two threads (their z-constraint is tangent), but rotation about the line between the two nuts is not really constrained by the smooth rods (their free direction is tangent to the rotation). I'd recommend having the bearings and the nuts at different heights from the bed to fix this - e.g. nuts on top, bearings below, 10-20mm spacers each. No additional $ parts needed, and only downside is it might eat up a bit of z travel.
Smoothieboard via Octoprint on RPi