[Al Skierkiewicz]

Whoa Boys,
I stop looking at the boards for a few minutes and things go a little crazy on me. The PWM output (or input for that matter) is not able to be accurately read on most multimeters as Kevin specified above. The input circuitry for most meters, Fluke general line included, is designed to measure AC signals below 400Hz in most cases. The conversion electronics are designed to give reasonable RMS datum on nearly pure sine waves. With the Victor output being a PWM square wave at a much higher frequency than 400 Hz, very few meters will give any indication that would be linear. Two methods that would give you accurate data would be to measure the duty cycle with a scope or with a PWM to average convertor. The second may be easy to build. Check out the ARRL Radio Amatuers Handbook for full wave bridge power supplies. Build up the bridge and capacitor filter, (two components from Radio Shack, bridge rectifier and 100 microfarad capacitor) and add a little load resistance (say 470 ohms at 1/2 or 1 watt.) in parallel with the cap. Measure the voltage across the resistor for different PWM inputs and you should see what you are expecting to see in linearity. (Give the output a chance to settle a few seconds before making the measurement.) The bridge will also allow you to measure reverse currents as well since the bridge will automatically make the switch for you. Remember that this test is only looking for representative numbers, not absolute, dead nuts, accuracy. It is just easy to build and get valid, representative data.
Mike, the RMS value of a square wave of varying pulse duration can be calculated for each pulse by averaging the voltage over the pulse repetition, i.e. the duty cycle. If the duty cycle is 40% then a 12 volt pulse would average 4.8 volts, a 10% pulse would be 1.2 volts, etc.(average is the secret word here!) You can, of course, write an integral that describes the action or even use Fourier transforms for other than pure sine waves. Since RMS is theoretically intended to give the same "heating power" of an AC signal as the equivalent DC voltage would. (pure sine wave or not) True reading RMS voltmeters as Kevin suggested above actually make this measurement. The 0.707 * PEAK Voltage only holds true for pure sine waves and is the solution to the integral for a pure sine wave. Add a little distortion to the signal and you can throw your calculations out the window.