sounds like your useing an inline non return for pneumatics? these work with a spring and a rubber ball into a chamfer, when the pressure is high enought(as you suggested you think might be the problem) the rubber ball pushes against the spring and lets the air through, returning to closed when the pressure drops below its rated operating level.
youve obviously realised this might be the problem, and if your useing the part ive described like norgren or something, it wont be good enought, even tho your useing high pressure co2 the output doesent even compare, its mild especialy at 1.5 bubbles per sec.
also my experience with building reactors has been the same as yours, the only inline reactor i bodged up (bodged due to not being able to get the parts like the americans can
) used to build up air, but i think this is the due to the timespan of the photo period and it comming to an end. however it still builds up and it seems to me that it could also be the low power of most canister filters, i hear people getting wicked results useing the same filter as me yet my flow is to kak to power something, then i wonder if its just my design, to powerful a pump and you lose co2 out the outlet...... not powerful enought you wont get enought 'churn' to eat the bubbles away no matter how many choppers and balls and whatnot you use.
i still havent found/built a good reactor for myself, but im sure i will sooner or later. as im sure you will. maybe you could run your reactor from a adjustable flow pump inside the tank, and still keep the reactor outside.
hight of the canister to top of the water level can also make reactors not perform as desired IMO
by the way im not directing my point at rex griggs comment but your comment on what you have used and think could be the problem , although what he is describing sounds like pneumatic parts.. they have been recommended by the man to work with low pressures so maybe you should take him up on that.