The picture below shows a particle $A$ linked to a disk through a massless rod. The rod can freely turn, with no friction, around a fixed axis that goes through the center of the disk. The system is initially at rest and starts moving because of the torque caused by the rod-particle. The moment of inertia of the disk is $0.5 \, M \, R^2$; the mass of the particle is $m$; the length of the rod is $L$.
I want to compute the speed $v_A$ when the rod is at vertical position.
I know this problem is pretty standard, and it is solved using conservation of energy of the whole system. Let me skip the details: $m g L = 0.5 \, (0.5 \, M \, R^2 + m L^2) \, \omega^2 \implies \omega = \sqrt{\frac{2mgL}{0.5 \, M \, R^2 + m L^2}}$ and therefore $v_A = \omega L = \sqrt{\frac{2mgL^3}{0.5 \, M \, R^2 + m L^2}}$
On the other hand, I feel that the energy of $A$ should also be conserved on its own, because the only non-conservative force acting on it, is perpendicular to the particle's trajectory. In this case, again skipping the details this yields $m g L = 0.5 \, m \, v_A^2$ so $v_A=\sqrt{2gL}$.
Clearly these answers are related, the first one becoming the second one as $R \rightarrow 0$. Based on what I have seen, I would take the first answer being the right one. In other words, the energy of the particle alone is not conserved, but why? What non-conservative force acting on $A$ is doing work here?