I have a question about distortion in LNA. There are two tones (f1 and f2) fed into the LNA. At the output of the LNA, there are f1, f2, 2*f1-f2 and 2*f2-f1 terms. Ideally the amplitude of 2*f1-f2 and 2*f2-f1 should be equal. But in really normally they are not equal, can someone help me explain that? Thank you.
In general they should be equal in amplitude if the spacing between tones f1 and f2 is small.
There is only one reason that can make the 2f1-f2 and 2f2-f1 to don’t be equal (if the spacing is small), and this is related to the Memory Effect that could appear into the amplifier.
Are two kinds of Memory Effect: Electrical (due to the envelope termination on the bias filtering) and Thermal (due to the junction temperature which is modulated by the applied signal).
Thanks. Can you explain the memory effect in detail? esp. the electrical one. Thank you.
vfone said:
In general they should be equal in amplitude if the spacing between tones f1 and f2 is small.
There is only one reason that can make the 2f1-f2 and 2f2-f1 to don’t be equal (if the spacing is small), and this is related to the Memory Effect that could appear into the amplifier.
Are two kinds of Memory Effect: Electrical (due to the envelope termination on the bias filtering) and Thermal (due to the junction temperature which is modulated by the applied signal).
Memory Effect could be explained as a time lag between AM-PM and AM-PM response of the amplifier. The Electrical Memory Effect is introduced by poor gate/base and drain/collector decoupling at low frequencies causing a distortion of the envelope currents which results in IMD asymmetry.
Low frequencies mean baseband/video frequencies or the frequency spacing between two tones.
The most significant Memory Effect appears in Class AB amplifiers, with reduced conduction angle where drain/collector varies with output power. In Class A amplifiers the Memory Effect reduces.
For best characterization for Memory Effect do a two-tone IMD vs Output Power test (plotting both 2f1-f2 and 2f2-f1) for different spacing between tones (e.g. from 100kHz to 10MHz). See the difference, and tune the bias filtering for best equivalent level between tones at the spacing you are interested.
As far as I know, the equal level of the intermodulatin is caused by an equal level of the input signals (f1, f2)
If you'll change one of the input level you'll see the output intermodulation level will change.