Archive of the former Yahoo!Groups mailing list: Homebrew PCBs

This sort of thing was done for a long time to make plates for fabricating integrated circuits for quite a while. (ICs use a photosensitive resist spun onto the wafers.)  Optics had to be precise to set the exact reduction and be accurate across the field.  However, the area exposed was generally quite small, less than 0.5 inches on a side.  The images were stepped and repeated across the plate to the size of the wafer.  The exposures to make the wafers were done with contact printing.  However, defects were created when the plate touch the wafer causing bad die and yield loss.  Later a collimated light beam as use through a plate to make non-contact printing.  Now instead of full wafer exposure plates, the images are in a relative small plate (reticle) with a few to several die in a cluster.  These images are made in redundant groups with the most perfect (defect free) version is selected for production use.  The cluster is projected and stepped across the wafer to in consecutive exposures – a   bit slower, but better final yields.

The idea of doing a reduction, 2:1 or 10:1 as we did back in the day is possible, but would require very good optics with a wide undistorted projection field - bigger than your biggest board.  That would present a challenge that grew exponentially with board size.