I came up with the following electro-mechanical model of a voltage-gated ion channel, resp. of its voltage sensor as described here and here. The voltage sensor is a bistable system with two local energy minima $E_1(V)$, $E_2(V)$, the first one indicating "closed", the second one "open":
From Bezanilla, The Voltage Sensor in Voltage-Dependent Ion Channels, p. 567.
You can interactively explore a voltage-dependent energy surface here (schematically).
I created a spring model with Algodoo which looks like this (green: the voltage sensor (= a peptide subunit), white springs: keep it in place, blue: favours the closed state, red: simulates voltage, opens the gate):
(You can download the model here and explore it interactively with Algodoo.)
To behave stochastically - i.e. the probability of being open $p(V) \sim e^{-\Delta Q(V)/kT}$ which presupposes a thermal equilibrium - there must be fluctuations that can lift the system above the energy barrier between the two energy minima.
My question concerns the source of these fluctuations in the real ion channel: are these thermal (mechanical) fluctuations (e.g. other molecules hitting or tearing the sensor) or are these fluctuations of the membrane potential?