Hi Guys,

The usual non linearity problem is the tiny voltage required to turn on the transistor or diode that is used to do the measuring. Commonly seen as a 0.6 or 0.7 volt drop across the base emitter junction of a silicon device and about 0.3 volts across a germanium device.

In my early ham radio days I used to use what were called "Zero Bias" diodes to rectify low levels of RF often from the output of directional couplers, often called SWR meters. Even so those were quite non linear at very low levels. Trying to use a transistor to amplify the voltages from the coupler first made things even worse.