Noise analysis shows different transfer function compared with AC or XF analysis

Dear delays,

delgsy said:

I always consider the gain from the noise anlaysis is the same as the transfer function.
But it seems that they are defined differently.
As can be seen in the picture below.
AC and XF analysis show the same result, but they are different compared with the gain from noise analysis.
did I miss anything?

Well, I don't think you are missing anything as you noticed there appeared to be a difference in your two sets of results and the asked the question!

In reality, the noise gain is not always the same as the signal transfer function gain. Why? The noise gain expresses the transfer function of both the input noise and any other noise source of the device under test. For example, in an op-amp, the noise sources might include the signal source and the offset voltage (usually modeled as a voltage source in series with the input). There are a lot of references for noise gain available. However, I've always found the information from Analog Devices to be well written and often referenced. You might consider the introduction from Analog Devices at URL:

www.analog.com/.../mt-033.pdf

to provide a better and more extensive answer to your question than I have in my response!

Shawn

Dear delays,

delgsy said:

I always consider the gain from the noise anlaysis is the same as the transfer function.
But it seems that they are defined differently.
As can be seen in the picture below.
AC and XF analysis show the same result, but they are different compared with the gain from noise analysis.
did I miss anything?

Well, I don't think you are missing anything as you noticed there appeared to be a difference in your two sets of results and the asked the question!

In reality, the noise gain is not always the same as the signal transfer function gain. Why? The noise gain expresses the transfer function of both the input noise and any other noise source of the device under test. For example, in an op-amp, the noise sources might include the signal source and the offset voltage (usually modeled as a voltage source in series with the input). There are a lot of references for noise gain available. However, I've always found the information from Analog Devices to be well written and often referenced. You might consider the introduction from Analog Devices at URL:

www.analog.com/.../mt-033.pdf

to provide a better and more extensive answer to your question than I have in my response!

Shawn

Actually, this isn't correct Shawn. The gain in the noise analysis is from the specified input source to the output node - it's what is used for computing input-referred noise (by dividing the output noise by the gain) and also to compute noise figure. Given that noise analysis is essentially based upon an xf analysis (with many more sources added for all the noise sources), if you've picked the same source in the xf results as you did in the noise analysis setup, I would expect it to give the same result.

However, I see from the graph that you are comparing pnoise gain with the gain from xf and ac analyses. That's not necessarily the same, because that is performing a PSS analysis (so you have a periodic steady state rather than a DC operating point) and then giving you the gain from a specific frequency to the output frequency - whereas XF/AC are computed around the DC operating point and give you the transfer function at the same frequency. You don't appear to be comparing like with like. That said, I just plotted the noise transfer function from the direct plot form and rather bizarrely this gets annotated as "pnoise gain" - so maybe you are just comparing the noise result after all. However, I suspect there is some difference in how you have the noise and xf analyses setup - perhaps you (delgsy) can share the analysis statements from the input.scs?

Andrew