IIP3 simulation in ADS Keysight

[fred3991]

Hey, everybody!

I am simulating an LNA with high linearity using Derivative Superposition Method. (This is just the core LNA without matching on NFmin and output. GAMax values are now 15 and NFMin 1.5 which I am quite satisfied with for now).
(I guess the specific method of increasing linearity is not important I just want to understand the results of the simulation).

These are the values of the simulation port and frequency.

This is the spectrum I see in the simulation.

My questions are this.
If I put the markers on the left side (M4 and M5) then calculating the IIP3 value I get a value of 18.3 dBM.
And if I put the value on the right side - then the calculated value using the ip3_in function is 22.9 dBM.

(I also check this simulation in Cadence Vitruoso to make sure that the high IIP3 value is the same in both simulators).

Can these values be considered valid? Is the value on the left or right more important?
Or is it important that they are approximately equal ?
Should I keep to these simulation results while continuing the design ? (building a full LNA circuit by adding input and output matching).
Are there any other characteristics that should be paid attention to ?

Thanks!

 

[BigBoss]

This is normal.
OIP3/IIP3 Values can be different for Left and Right IMD3 Products.(Memory Effect) For more information

Concept of Memory in RF Power Amplifiers

High Frequency Electronics

www.highfrequencyelectronics.com

 

[fred3991]

This is normal.
OIP3/IIP3 Values can be different for Left and Right IMD3 Products.(Memory Effect) For more information

Concept of Memory in RF Power Amplifiers

High Frequency Electronics

www.highfrequencyelectronics.com

let's assume I have completed the entire design and layout, with all simulations (EM and RC extraction).
And I get the same different IIP3 value on the right and left.
Will the same behavior happen again in the measurements?
When the difference is 1-2 dBm it's ok, but here for example now in the simulation the difference is almost 20 dBm!!!.

When I searched for articles about IIP3 boosting I never found any mention that the IIP3 on the left differs so much from the one on the right.

When I write an article about this LNA should I also keep silent about the bad side and just show only the good?)
I think it's a bit like cheating to be honest.

Or will everything be about the same high (or low????) in the measurements
(adjusted for modeling accuracy and the accuracy of the models in PDK)

 

[BigBoss]

let's assume I have completed the entire design and layout, with all simulations (EM and RC extraction).
And I get the same different IIP3 value on the right and left.
Will the same behavior happen again in the measurements?
When the difference is 1-2 dBm it's ok, but here for example now in the simulation the difference is almost 20 dBm!!!.
When I searched for articles about IIP3 boosting I never found any mention that the IIP3 on the left differs so much from the one on the right.
When I write an article about this LNA should I also keep silent about the bad side and just show only the good?)
I think it's a bit like cheating to be honest.
Or will everything be about the same high (or low????) in the measurements
(adjusted for modeling accuracy and the accuracy of the models in PDK)

OK, the problem is now clear.
In order to simulate or measure a OIP3/IIP3 the Output Power for each carrier ought to be approximately -20dB lower than P1dB Compression Point.
But your Output level for each tone is Too LOW. Therefore the difference seems huge.
Check P1dB Compression Level first then apply these two signals to get appropriate Output Level so than -20..-25dB lower than P1dB Compression Point.
You will see that the Difference will decrease. Because Third Harmonic Products ( 2*f1-f2 or 2*f1-f2) increase 3 times than fundamental one.
You will absolutely see almost same behavior but the difference will be more realistic/lower. You will also observe the same phenomena during measurements.