I took a couple of designs from here and tried them. I found the flow to be too slow with the 1" tubing if the "final T(or X)" was too high. Even using two 1" pipes didn't seem good enough. Using the hydraulic engineering I took in school: (twice
) I think the portion that runs out of the tank above the waterline needs to be smaller than the rest of the system. The way I tired mine was to put a t near the waterline, doubling the pipe area increasing the effeciency of the siphon forces. Velocity is important, and too big of pipe above the waterline reduces velocity.
In the test setup at steady state, I can dump a half gallon in the 10gal tank and it will return to steady state in about 16seconds, which is about 1/3rd the time it took when there was two complete pipes.
I also set up the pipes geometry so I could easily tune the hieght of the X-piece to my pumps flow just by rotating the pipes feeding it. The bad side is that the whole thing sticks out about 8" from the tank. (maybe 2" of that could be fixed using some 45s on the last up-pipe.