KenF:
This would make more sense than the setup the OP currently has. Alternatively a simple NPN transistor, a zenner diode and resistor would make the arduino totally redundant and be far more stable.
using a zener diode like that would be a LINEAR power supply, and would cause a tremendous amount of wasted power. The variation in the supplied voltage and the required load voltage gets eaten up by the resistor... and dissipated as waste heat! Not something you really want in an alternative energy setup such as used with LED lights and wind powered generators. Did ya think to wonder why I was going through all the trouble with making a switching regulator over a linear regulator? Thanks for the suggestion though.. however useless it is...
And your suggestion that measuring the output voltage and making adjustments to the duty cycle rather than measuring the source voltage and applying the correct duty cycle baffles me... This seems like absolutely the most backward approach to the problem.
I can see the use for measuring the output as a feedback mechanism to fine tune the duty cycle.. but using it as the main means for acquiring the correct duty cycle to use?? backwards!! This poses all types of problems.. the load may be exposed to excessively large voltage pulses while the system is attempting to adjust itself to find the correct duty cycle... Whereas if the source voltage is simply measured, the correct duty cycle can be calculated and applied without issue. Output voltage can be measured, and if it's off by a bit, duty cycle can be slightly adjusted accordingly.
Does anyone else see anything wrong with what kenF and nick gammon are suggesting?