For the fast tour ....
Without knowing the details of the reference design you saw, the two frequency ranges (at the input) of interest of any PLL are:
Lock range (Acquisition range) Hold-in range
[Other frequencies of interest in a PLL include the phase rate (the rate at which the input frequency can change) the PLL can follow and loop frequency (the filter characteristics of the loop filter in the frequency domain) - these two are closely related in terms of PLL performance].
At the output we care about the ouput VCO range (which may be directly fed back or may be sent through dividers as is done in frequency synthesis).
Both *input* ranges are a function of the VCO (the VCO Fout/Vin function determines the output frequency range of the VCO), the phase detector and the loop filter (and divders if present), but typically Hold-in is wider than lock, although not always.
To address your questions:
If the frequency is higher or lower than the acquisiton range (assuming it is unlocked), PLLs will do one of a number of things, depending on whether there is a signal present or not, and whether it is within the lock/hold-in range of the input
a. Sweep across their frequency range (Sawtooth applied to VCO V[control] as the phase detector gives a varying output (common if F[in] is outside the lock range of the PLL, but depends on the characteristics of the phase detector)
b. Sit at a single frequency (except for jitter) - common for many PLLs in the absence of a signal.
As to 'How does it know the signal is within that range' - the range of frequencies a PLL can cover is determined by the overall characteristics as noted above.
Cheers
PeteS