This small project was based on a small RF Power Board I bought off of EBAY. The board uses an AD8362 chip, which reports to be able to turn complex waveforms into RMS and yield the power referenced in Decibel Milliwatts very accurately over a 50Hz to 3.8Ghz bandwidth. The unit works for equipment that utilize 50 ohm outputs. The board was approximately $20 so I gave it a try. You have to supply the power, input connectors and cables and a digital voltmeter that can read millivolts. I supplied these things and built this project. The next step was to test and understand how to use it. I found that if you started with a known output (say -20dBm) into the 50ohm, and then changed the output to another value (say -30dBm), you could subtract the voltage measured at the start from the later voltage at -20dBm, then divide the result by .05v and get a very accurate accounting of the dBm shift (-10dBm in this example). I then decided to graph the results of these shifts over a 40dBm range for various frequencies. I found that the output was extremely linear for any given frequency, but that the voltage outputs were different at different frequencies. To be able to use this unit to measure dBm output in a more direct way meant I was going to have to understand my output vs frequency characteristics. I started to develop various curve fits for my test results over the test frequencies, eventually settling for a 6th order polynomial equation that best fit the data. Then I tested my equation over my test band of frequencies and random dBm values, and I found it to be accurate to better than 4% (better than 2% in many cases). This video chronicles the effort.