The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> pnoise with noisetype=correlations
https://designers-guide.org/forum/YaBB.pl?num=1037467194

Message started by am on Nov 16^th, 2002, 9:19am

Title: pnoise with noisetype=correlations
Post by am on Nov 16^th, 2002, 9:19am

Hi,

Let me start by clarifing my understanding of pnoise calculations:

1. pnoise with noisetype=correlations
I thought that in this case the simulator would compute:
E{X(relharmnum*fund+df) * conjugate(X(relharmnum*fund+df+cycles*fund))}

2. pnoise with noisetype=sources
In this case I expected the simulator to simply compute:
E{X(relharmnum*fund+df) * conjugate(X(relharmnum*fund+df))}

Under this assumptions pnoise with noisetype=correlations and with cycles=0 would be exactly the same as pnoise with noisetype=sources.

Well, I have written an simple behavioral model to check this (first I did an AM modulator and then a PM modulator), and in fact this is the case in my simple test circuit.

Then, I tried this with a simple VCO circuit (a simple amplifier and an LC tank). In this case I run the circuit in autonomous mode and now the values of the pnoise with noisetype=sources and pnoise with noisetype=correlations and cycles=0 are very much different.

I would also like to clarify that I was careful in the following aspects:
- in pnoise with noisetype=sources I square the result (in order to compute power) and then compare it to the other simulation
- since the output of the VCO is differential and in order to avoid errors with factors of two, I convert the differential signal to a single ended signal with a simple behavioral model.

At the end I am left with a discrepancy between the two results. I beleive the one with noisetype=sources is right.

So,
1. Am I wrong in my understanding of pnoise?
2. Is there something different in the way pnoise works in an autonomous circuit considering that my test case worked fine?

Thanks,

AM

Title: Re: pnoise with noisetype=correlations
Post by Sergei Pevchin on Nov 18^th, 2002, 2:04pm

I would suggest to try the latest version of SpectreRF. We finally fixed one problem with autonomous pac/pnoise_corr analyses.
This was released in 4.4.6.100.83 and later versions, and in
5.0 version of SpectreRF. If this will not help, we will have to look more closely at the problem.

Thanks,
Sergei.

Title: Re: pnoise with noisetype=correlations
Post by am on Nov 18^th, 2002, 7:30pm

I have tried this both with Spectre RF 4.4.6 and with 5.0
In both cases the results of pnoise with noisetype=correlations and cycles=0 is incorrect.
Furthermore, the result of pnoise with noisetype=correlations and cycles=-2 seems to be correct (the cross correlation seems to be correct for the case of a VCO, where we do not have AM noise close in).

Even more strange is the fact that I can not find out what is the correct carrier phase considered for the correlations (I need this to separate AM and PM noise).
The results make sense when I run it in driven mode (I have a very simple code that breaks the AM and PM noise based on the USB, LSB and the correlation between USB/LSB), but in autonomous mode the phase reported in pss-fd does not make sense (it actually looks like if the phase carrier phase was normalized to +90 degrees).

AM