I haven't figured out what they use in the small medical 'flow meters'. I'm guessing either a needle valve with preset detents (like used with a Thorpe tube, but set at the factory), or a fixed orifice with a variable backpressure. Either way, or even if it's a different way, I doubt the ~2psi downstream backpressure affects the rate much, since the upstream pressure is likely 60psi or greater. I can't see them getting accurate/precise single digit psi changes with a diaphragm that small. That said, the Thorpe tubes are a much more accurate meter when dealing with variable downstream pressures. If someone combines a Thorpe tube flow meter with one of the cheap medial flow meters with/without a stone attached, we would know for sure how much downstream backpressure affects them.
One thing about injecting the O2- the bulk of the diffusion happens at the surface (provided the flow is high enough). From memory, it is ~90% compared to the diffusion that occurs as the bubbles rise. There are limits and variables, of course, but the basic guidelines are to have a moderate patch of bubbles breaking the surface. As long as the currents aren't so violent as to collapse the bubbles in the stream (as mentioned earlier), but enough to cause some turnover, the premise holds.
I would have to search for the paper, but it was an involved study related to aerating settling ponds (yes, ponds of poo). The critical factor was the 'skin effect' as the bubbles popped on the surface. Smaller bubbles were still better than larger ones.
