For the first question: yes you are correct, provided indeed the centres of the planet gears stay fixed.
As for the other question...
The problem is that the needs you are specifying for planetary gearing is in conflict with each other, i.e. the problem is over-constrained. You can define a speed ratio between two types of gears (e.g. sun and planet) but not for three gear types; the third gear will already be determined in order to satisfy the gear ratio between two gear types while also satisfying geometric constraints: that is, the annulus (ring) gear must have a number of teeth equal to that of the sun gear plus twice that of a planet gear; gear pitch radius is proportional to number of teeth as teeth size on meshing gears must be the same.
There are initially six unknowns here: angular velocities of the sun, planet and annulus gears; and the number of teeth of the sun, planet and annulus gears:
$$\omega_s, \omega_p, \omega_a, N_s, N_p, N_a$$
Six unknowns require six independent equations for a unique solution.
Kinematics, or the relations of speeds of gears, must be valid, and so this enforces 1 of the 6 equations:
$$\omega_s N_s + \omega_p N_p = \frac{1}{2} \left( \omega_s N_s + \omega_a N_a\right)$$
Geometry must be satisfied; i.e. the sun and planet gears must be able to fit inside the annulus gear: equation 2 of 6:
$$N_a = N_s + 2 N_p$$
Now, you wish to constrain the motion so that the centres of the sun and planet gear do not move; equation 3 of 6, noting pitch radius is proportional to tooth number:
$$\omega_s N_s = -\omega_p N_p$$
Equations 4 and 5 come from specifying speed ratios:
$$\omega_p = -2 \omega_s$$
$$\omega_a = -\omega_s$$
At this point, it's tempting to think: oh the problem's under-constrained, we still need one more equation. However, whenever you substitute equations into each other, two variables will cancel out in the process: one of the angular velocities and one of the tooth numbers. This is okay: it is the ratios of speeds between gears that matter, not the absolute speeds; same for teeth numbers.
With two unknowns dropping out, this leaves us with 5 equations and 4 unknowns: the problem is over-constrained. This means that, apart from pure coincidence, contradictions will arise whenever you try to solve the equations. An example would be like trying to solve the following:
$$x + y = 3$$
$$x + 2y = 5$$
$$2x + y = 5$$
Two of the equations give a solution, the third creates a contradiction.
How do you resolve this over-constraining? Remove one of the five equations! You can't remove the kinematic or geometric equations since they are fundamental laws for planetary gears. Instead, remove the constraint where the centres of the planet gears must be fixed, or enforce a speed ratio between only two gear types. Then you get four equations, four unknowns.