Looking at the data sheet for a 1N4733A from Fairchild, the min Zener voltage is listed as 4.845 volts. Does that mean that I should design my voltage dividier such that the maximum expected input will produce less than 4.8v across R2, ensuring the Zener won’t break down within my measured range?
As long as there is series resistance, there’s nothing wrong with some current through the Zener. 10V across 1K would be 10mA, and 10mA at 5V is 50mW. So, the Zener would be working well within it’s range.
The “breakdown” voltage is where the Zener switches-on and does it’s thing. Below breakdown, it’s essentially not in the circuit. At breakdown, it’s limiting the voltage, which is it’s purpose. If you never hit the breakdown voltage, it may as well not be there.
I’d design the voltage divider for 12V or 14.4V or whatever the “nominal” voltage is, and then use the Zener for anything “unexpected”. If you need to measure 18V, then the voltage divider should be designed for that, and make sure the Zener doesn’t kick-in until you exceed whatever voltage you want to measure accurately.
Alternatively, you can use a regular diode (or a Schottky diode) connected “backwards” to the Arduino’s 5V regulated power supply. [u]This page[/u] shows the Zener method, and another method with two regular diodes to protect against excessive positive voltages, or negative voltages. (I’ve used the regular-diode method.)
In any case, you need at least one resistor to limit the current through the protection diode(s) and you can use either method with a voltage divider (2 resistors).