One thing about R & D is that the project eventually moves you from physics (the “R” part) over to something that’s more engineering in nature (the “D” part). Here’s a quick example. Here’s the laser system for the caesium fountain, which is a research device. It’s spread out to be optimized for getting your hands in and tweaking on knobs, since you don’t know ahead of time what is going to work best. Lots of mirrors and other optics that need to be adjusted, and there has to be room to change things around and/or add things that might work better. You have to generate six beams for trapping (this is done at the table under the big cylinder, splitting two beams from the main table), plus a beam for optical pumping and another for probing the atoms. That’s four different beams on the table, at various (and for the MOT, adjustable) frequencies. The long paths meant that the beams would only stay aligned into the fiber couplers for, at best, weeks at a time.
Eventually you decide on a design that works, and since our production devices need to fit into a smaller space, and aren’t meant to be adjusted much (ideally, not at all) after the initial setup, you make everything smaller. Pretty much everything except the lasers (the 2 blue boxes on the left and 3 black boxes; one center-front and two right-rear) and the spectroscopy (which ends up on a different table) are compacted to fit on a much smaller table.
Everything is fiber-coupled, so it’s modular — it doesn’t matter which particular laser you use, or what you do with the beams after they leave the box. As you can see, it’s rack-mounted, and about 4u high. (You can just see another table at the top of the picture; this houses the spectroscopy/locking hardware)