Noise figure measurement needed?
No..
Measurement of NF is not necessary because you will probably have very high NF..
Thank you.
For RF power amplifiers, output noise is the primary interest. It indicates potential interference or desens to receivers, particularly in the case of duplex Tx/Rx systems like CDMA phones where excessive sideband noise crosses Rx frequency raising the receiver's input noise floor.
It does make sense for a linear RF Power Amp. What about non-linear RF PA like Class E PA?
Same applies to non-linear amps. Particular specs depends on application and regulations. GSM phones for example have Tx noise specs to avoid desensing nearby handsets and desensing a nearby basestation.
Generally, non-linear PA's have lower wideband noise output then linear PA's.
I see. Is the noise figure for non-linear PAs the same to the noise figure for linear ones in definition? What equipments should I use to test the NF? Spectrum analyzer? Network analyzer? How should I set it up?
First stop calling it noise figure. This term is used or input to output noise factor of a linear amp.
Measured with spectrum analyzer. Sometimes requiring an isolator and filters to prevent the primary output from overloading the spectrum analyzer. You have to be careful to maintain proper impedance termination on PA as it will effect output power and noise output spectral performance.
Can be expressed as dBC/Hz or absolute dBm at a specified frequency offset and bandwidth.
Now it makes sense. So I just need to measure the noise power versus different frequencies on a spectrum analyzer, right?
Yes, but be aware of dynamic range limitations of the spectrum analyzer. Don't overload it with the primary output power from the PA.
For higher power amps you put an isolator on PA output to provide proper termination, then a suckout filter after the isolator to remove the primary PA output power leaving the sideband noise to measure.
I see. The output power of the RF PA is low power. It is about 10mW.