Weird phenomenon when measuring the P1dB of a LNA

Hi guys,

I have this very weird phenomenon on my measurements of the P1dB of a LNA i designed as you can see on the figure below:

I have used the same method to measure other LNAs and i got the expected P1dB with no problems beside this one.

Thanks in advance

It looks like your "linear gain" value is smaller than the actual small signal gain ?

In fact, measurement of Power Gain vs. Output Power is more suitable to define P1dB.Then the corresponding Output Power@G1dB is found
Such curves above are a bit confusing.

What you observed is due to the self biasing of the transistor. Increasing the quiescent current the phenomenon should disappear.

Yes, the gain i considered as linear gain is the one at -15dBm

Here is a graph of Gain Vs Input power.




I'm going to look into that, thank you

If the Gain increases with Input Power, there are some problems such as :

-Unwanted Feedback
-Temperature Drift/Thermal Issues
-Optimum Load Impedance Changes-if applicable-( Output Power is relatively low, this probability is also low )
-Measurement Errors

Can you post the circuit diagram and its layout too ?

The name of the phenomenon is, gain expansion.
The reason is only the bias point of the device. Usually appears in devices biased in Class-AB.
As was stated, just increase the quiescent current (near the class-A bias) and the gain expansion will disappear.

Also, an LNA with a gain of 5dB seems hardly worth the effort.

The Output Power is relatively low,I don't think the Gain Expansion occurs.It generally occurs in higher Power Amplifiers.

I actually was expecting a gain of 10dB


I have measured Vce for different Pin, and it actually changes with Pin. which apparrently happens for gain expansion

Here is the design and the transistor i'm using is the BFU760F

Where is the emitter connection to ground on your board? Or, perhaps show the PCB that you used.

Yes, gain expansion is mainly observed in power amplifiers, because only the power amplifiers are biased in class-AB.
LNAs always should work in Class-A (by definition).