Taking into account some likeness of moderate deviations (MD) andcentral limit theorems (CLT), we develop an approach, which made agood showing in CLT, for MD analysis of a family $S^\kappa_t=\frac{1}{t^\kappa}\int_0^tH(X_s)ds,t\to\infty$ for an ergodic diffusion process$X_t$, provided that $0.5 $\frac{1}{t^\kappa}\int_0^tH(X_s)ds=$corrector$+\frac{1}{t^\kappa}$ ${M_t}$martingale. and show that, as in the CLT analysis, the corrector isnegligible, and the main contribution in the MD brings the family '$\frac{1}{t^\kappa}M_t, \ t\to\infty. $'' Starting from Freidlin,[7], and finishing by Wu's papers [33]-[37], in the MD studyLaplace's transform dominates. In the paper, we replace thistechnique by 'Stochastic exponential'' one, enabling to formulatethe MDP conditions in terms of 'drift-diffusion'' parameters and$H$. However, a verification of these conditions heavily depends ona specificity of a diffusion model. That is why the paper is named'Examples ...''.