I was finding exactly the same behavior in my own program… so seeing this thread, I sought to confirm it. So first, the math for constant acceleration:
x = x0 + v0t + 1/2a*t^2
If we assume a point mass starting at 100m (x0 = 100), and initial velocity of 0m/s (v0 = 0), a constant acceleration of 9.81m/s^2 (a = 9.81), and a final resting position at ground (x = 0) then we can solve for t.
0 = 100 + 0.5*(-9.81)*t^2
t = sqrt(200/9.81) = 4.515
So, creating a new, blank project, I added a sphere to the project. The default sphere with no scaling has a radius of 50cm. Therefore, I position the sphere at 10050cm (100.5m) above the plane. Upon hitting a key, the sphere falls and generates a hit event 4.56s after starting. Pretty close to the theoretical. I then repeated the experiment in my game where my flying object feels very “floaty”… and I get exactly the same result. So I guess there is nothing wrong with the physics engine.