If we have a continuous time signal given by $\sin(2\pi f_0t + \phi)$ the discrete time version may be represented by sampling every $T_s$ seconds (our sampling period)
Since we are sampling every $T_s$ seconds, we can replace $t$ with $nT_s$, where $n$ is the number of samples.
$\sin[2\pi f_0 nT_s + \phi] = sin[2 \pi \frac{f_0}{F_s}n + \phi]$
Which may be represented as $sin[2 \pi \frac{f_0}{F_s}(n+N) + \phi]$, where $N$ is the period of the discrete time signal
Now, if discrete time signal is periodic, value at $n$ must be equal to value at $n + N$, implying that $2\pi \frac{f_0}{F_s}N = 2 \pi k$ where k is an arbitrary integer
If you rearrange that expression, you get $\frac{N}{k} = \frac{F_s}{f_0}$. If there are no integer values of N and k that solve this equality, the n the sampled version of the signal is non-periodic.
### Techniques for signal generation
- Direct digital synthesizer (DDS)
- Phase accumulator with sin() or cos() function call/table lookup system
- Special cases
- Sines and cosines with f = Fs/2, f = Fs/4, etc.
- Digital resonator
- Impulse-excited, second order, IIR filter (what does this even mean?)
- Complex conjugate pole-pairs placed on unit circle
- Impulse modulator
- Based on scaled impulses periodically exciting FIR filter
- Lots to unpack here