Using an LDO is ineffective for a current sourced PV unless perfectly impedance matched between source and load.
Using PWM does not match naturally impedances and thus cannot achieve maximum power transfer (MPT). It must using a reactive L impedance in a continuous mode to store the energy at a matched impedance. The PV impedance is known from the slope peak operating points which drop in current faster than voltage, thus rising Req.
A PWM modulates the impedance of a load by Z/d with duty cycle d (0 to 1)
Your load is between a current source PV with a computed MPT ESR of 18V/14A = 1.4 Ohms
Your battery ESR is low (tbd) and depends on C charge rate max.
If only pulsing this loop resistance , it is possible to regulate the open circuit voltage Voc down to 18V but with a certain power loss in the net ESR of all parts including PV load cap.
A more efficient solution uses a flyback series reactor of inductance L (Buck) such that when switched at a fundamental frequency of f creates an impedance of 1.4 Ohms minimum and rises with changing duty cycle from 50% as the fundamental voltage at f reduces and relative harmonic energy increases from asymmetry of pulse. (Fourier spectrum) yet avg. power increases with d.
Thus it does not present a perfect matched impedance at MPT to the source PV impedance to satisfy the ideal maximum power transfer theorem.
By choosing f and L with variable d for PWM, one can sense the source voltage output from PV and load current I to compute this impedance on a cycle-cycle basis or an average basis depending on what control Design is used. a flyback Zener is suggested to make switching off faster but lossy, or a Schottky diode for CCM mode or a Dual FET half bridge for Ideal diode characteristics.
Since the output battery charge must be regulated for CC and CV and cutoff, there must be a designed control mechanism for input MPT current and output battery CC current with Voltage regulation now on both LV and battery.
Now you have an MPT and battery charge controller in one system design.
A similar analysis may be done in the time domain rather than my frequency
-domain matched-impedance analysis that I did. This may be Continuous or Discontinuous using CCM and DCM modes of switched reactor with amp-seconds of current transferred from input to output using LTI formula of V=LdI/dt for some switched interval ON and Off via the other conductor ( diode or FET).
You may try to understand this logic then search for simpler design rules to choose a PWM flyback PMT Battery charger or consider the losses in Efficiency from not achieving a matched impedance which rises as solar power input declines. Zpv = Voc/Isc *Pmax/Pin for some power ratio or “solarity” of PV input power. This is my approximation of a PV current source. ( for better or worse, it works)