We can get the mixing action by using the nonlinear characteristics of an active device.

The input-output characteristic of a 2-port network with weak nonlinearity is represented by a polynomial as

(1)

If , then the nonlinearity of the device introduce inter-modulation products in the output at frequencies , where p,q are integers and (p+q) is the order of non-linearity.

Due to mixing action the desired components should be at , where p=q=1. So second order nonlinear terms are required for mixing action. Higher order terms introduce distortion.

A nonlinear device with square law behavior is well suited for mixing operation.

Consider a second order polynomial,

(2)

Higher frequency components (RF, LO and their harmonics) are removed by filtering at the output. A device with square law behavior is well suited for mixing operation.

A long channel MOS transistor have nearly square law behavior is a good candidate for mixing operation. Even devices with other nonlinear characteristics does the same job, but introduce higher amount spurs (for example diode, BJT, etc.,).