gain expansion query

Hi All,

We are observing a peculiar problem with our MMICs. we have designed 3 stage x-band LNAs using pHEMTs, model from a reputed GaAs foundary, simulated on MwOffice. While testing the chips we observed that the overall gain has reduced to the tune of about 4-5 db from that simulated.
Also at a particular P(input), the gain expands and at the peak the gain is the same as simulted. This expansion is not seen in the simulation.
Can someone guide on the possible reasons of the same and the ways to eliminate these problems.

Thanks and regards

Definitely you have a bias problem. Gain expansion appear mostly in nonlinear class amplifiers. You need to increase the bias current, to put the amplifiers in a more linear behavior (close to class A) where you can meet the simulated gain, and the gain expansion will be minimal.

thanks vfone,

we are using self bias. And as such we are driving the stages at Idss/2. actually what we observed is just the reverse. When we decrease the bias of first and second stage, the gain seems to flatten out.
The design is at all stages 2V, performance improves at biases of 1.2V (first stage), 1V (second stage) and 2V (last stage)
Pl keep guiding.
Thanks and regards.

In your case having lower bias on driver stages improves the gain enhancement because it will not drive harder the last stage, and accentuating the self-bias in this stage. That means you need more current to the last stage.
As I said a linear class for amplifiers will solve the issue, but you have to make a compromise to meet the desired quiescent current.
There are more things that can affect this problem, but anyway all are resume of playing with bias currents of all stages.