I have simulated simple cascode LNA, where the only difference is I am taking a voltage sample across D-S of M1 (figure attached) for some purpose. I ran PSS simulation for IIP3 and 1-dB point calculation. The problem is, it is showing both points to be almost the same point that is around -19dBm. Why are these values (IIP3 and 1-dB compression points) becoming same?
I have tried PSS alone and PSS+PAC together but both give the same result.
I followed these procedures.
1>f0 is 2.14GHz, second tone is selected as 20MHz away that is 2.16GHz and the power applied is -40dBm. So first order and 3rd order harmonics are taken to be 2.14GHz and 2.12GHz, or 2.14GHz and 2.18GHz, or 2.16GHz and 2.12GHz or 2.16GHz and 2.18GHz for plotting IIP3. Anything it yields is the same.
Thanks and regards,
Vinay.
I haven't used Cadence, but try to make a Pin-Pout plot of both the a fundamental and one of the IM3 tones. Check if the behavior is smooth.
Thanks Erik. I could figure out why it's giving like that. I have attached Pout vs Pin curve for both fundamental (2.14G) and IM3 component (2.12G). IM3 is not linear w.r.t Pin, hence it is taking arbitrary slope to calculate IIP3. What would be the reason that IM3 is not exactly linear with Pin?
It is normal that there is a dip around P1dB, albeit yours is a bit shallow, But for every thing below -30 dB, this looks strange. As said I haven't used Cadence, but can you tweak convergence in the solver. Also in case the solver uses a Krylov subspace method, maybe tweak its parameters as well.\
Try performing a simulation on just a diode, instead of an amplifier, if it also occurs there. If it does, then it is the simulator
- Badges
- Science topic
- Similar topics
- Correction