Hi all,

I have performed tran+pss+pnoise analysis, on a cross-coupled LC oscillator, with two levels of accuracy setting Hard-1 and Hard-2 as per pnoise primer.

I have two sets of spot (@100kHz offset) noise summary data. In these sets I see that the noise contribution calculated for the switching transistors is as follows:

1. Results for Hard-1 setting

Device   Param Contribution          % of Total

/I13/M5     id       3.27963e-12           19.87    
/I13/M6     id       3.062e-12             18.55    
/I13/M0     id       2.06276e-12           12.50    
/I13/M1     id       1.93074e-12           11.70    
/I13/M0     fn       1.63057e-12           9.88      
/I13/M6     fn       1.24191e-12           7.52      
/I13/M5     fn       9.09556e-13           5.51      
/I13/M1     fn       6.11352e-13           3.70      

2. Results for Hard-2 setting

Device   Param Contribution          % of Total

/I13/M5     id       3.27783e-12           18.87    
/I13/M6     id       3.06363e-12           17.63    
/I13/M0     fn       2.28545e-12           13.15    
/I13/M0     id       2.06522e-12           11.89    
/I13/M1     id       1.92947e-12           11.11    
/I13/M6     fn       1.23788e-12           7.12      
/I13/M5     fn       9.10614e-13           5.24      
/I13/M1     fn       8.26599e-13           4.76      


As we can see the contribution by all components is different in both cases.

My question is How much difference is considered to be numerical error and can be neglected? In your opinion is the difference between the two results significant? Can I consider the pnoise result to be accurate?

FYI. The tran+pss+pnoise analysis were completed successfully without any error messages.

Thanks for your reply in advance.

Regards,

Kriyang