Mark, you are still confusing the Earth's daily rotation with the Earth's orbit. This is the sole source of your confusion. Imagine what things look like from a reference frame that is not rotating with respect to the remote stars and with origin at the solar system barycenter. The International Celestial Reference Frame, for example. The Earth-Moon barycenter follows a nearly elliptical path in this frame. The time it takes for the Earth-Moon system to complete one orbit in this frame is one sidereal year.
Now flip your point of view to an Earth-centered frame, but keep the same axes. This is an Earth-centered inertial frame (a bit of a misnomer; this is an accelerating but non-rotating frame). From this perspective, it is the Sun that completes one orbit about the Earth in one sidereal year.
What about a tropical year? The Earth's rotation axis is tilted with respect to the Earth's orbital angular momentum vector. Suppose the only bodies in the universe were the Sun and Earth (i.e., no Moon or Jupiter to confuse things) and suppose the Earth's rotation axis was unchanging. With these simplifying assumptions, the line connecting the Earth and the Sun would lie completely on the ecliptic plane twice a sidereal year -- and those equinoctal points would be fixed points in the solar system barycentric frame. The between successive vernal equinoxes, the tropical year, would be one sidereal year.
The Earth's rotational axis is not constant; it instead undergoes a slow precession. Because of this precession, a tropical year is a bit shorter (~20.5 minutes) than a sidereal year. Given that a full orbit (one sidereal year) is 360 degrees by definition, 20.5 minutes corresponds to about 50 arcseconds.