An FM signal is given by
$$ s_{FM}(t) = A_c \cos [\omega_C t +k_f \int _{-\infty}^t m(\alpha)d \alpha] $$
Let $ \int _{-\infty}^t m(\alpha)d \alpha] = a(t) $
Therefore, $ s_{FM}(t) = A_c \cos [\omega_C t +k_fa(t)]$
Now, for NBFM, $k_f$ is very small such that $|k_fa(t) << 1|$
We can then approximate $ s_{FM}(t)$ by
$ s_{FM}(t) \approx A_c[\cos \omega _c t - k_fa(t)\sin w_ct] $
We can generate an approximate NBFM at point (a). At point (b) we get the signal
$ x_b(t) = A_c \cos [n\omega_C t +nk_fa(t)]$. Where n is the multiplying factor.
At point (c) the signal is passed through a mixer and we get the signal
$ x_c(t) = A_c \cos [(n\omega_C - f_{crystal}) t +nk_fa(t)]$
After passing the signal through the frequency multiplier, we finally get the following signal at (d).
$ x_d(t) = A_c \cos [n_1(n\omega_C - f_{crystal}) t +n_1nk_fa(t)]$ , Where $n_1$ is the multiplying factor.
At point (d) since $k_f$ is multiplied by $n_1n$ the overall value of $k_f' (=nn_1k_f)$ increases. Thus, we are able to convert NBFM into WBFM by effectively increasing $k_f$ to $k_f'$.
I wold like to know if this explanation is correct or not?