Transient Noise Simulation gives the same transient results even though the noise spectrum of the noise sources are different.

Dear delgsy,

delgsy said:

Basically, I want to check the transient noise at VOUT if I provide different noise source (I6).
I used Noise/Freq pairs = 2, 1st pair = 1p*1p@10Hz, 2nd pair = 1p*1p@10MHz (left), 2nd pair = 1p*1p@100MHz (right).
In transient noise analysis setup, I used Fmax = 100GHz.
Both simulation gives exactly the same results.

I am really having trouble providing some thoughts on your question and will apologize upfront for that!  Your post leaves me with several questions:

1. You note you want to check the transient noise at VOUT when you provide different noise sources. However, you have only provided us one of the noise sources which I believe is I6. What is the noise profile of the other source?

2. Your noise source I6 has the same power density for each breakpoint in frequency. Why not just specify one frequency and its spectral power density will be applied through your noisefmax setting?

3. As I mentioned in your Forum post yesterday, your simulation time of 1 us will limit the visible noise below 1 MHz. Any differences in the input noise sources spectral density below 1 MHz will not be evident.

4. You have a very large resistor R0 shown as 1e9 ohm and a parallel capacitor C0 of unknown value. Is R0 noisy? In any case, these two components will filter any noise components with frequencies above 1/(2pi()*R0C0). Hence, your two noise sources are low-pass filtered by the following first-order filters depending on your value of C0:

Hence, if C0 is 0.1 fF, the transient noise frequencies you are observing have components up to about 1.6 MHz. With only a 1 us sample time, leading to an effective high-pass filter of 1 MHz, you are are only observing noise between about 1 MHz and 1.6 MHz.

5. Have you examined the spectral density of your output noise? I might suggest you review an effort I recently did to help explain some apparent misunderstandings of a transient noise simulation at URL:

https://www.dropbox.com/s/7vxzliml4uz6518/psd_study_sml_090622v1p1.pdf?dl=0

and a follow-on to one the Forum poster's questions at URL:

www.dropbox.com/.../two_sided_one_sided_fft_sml_090922v1p1.pdf

Perhaps if you have time to either correct my misunderstanding(s) or provide added data, I might be of more help - sorry!

I have a lot of experience with transient noise simulations and their analysis does require some effort. However, in my experience, with the results have always been explainable and reasonably accurate.

Shawn

Dear delgsy,

It struck me after I wrote my response as I thought about your noise plots and the magnitude of the noise evident in each, that 1 Gohm resistor may indeed be noisy. As such, its noise will dominate the noise you are adding with your noise files. I decided to plot the Johnson noise due to the 1 Gohm resistor and assumed your C0 was 1 fF. Note its magnitude far exceeds the spectral density of the noise you are adding.

Shawn

Dear delgsy,

It struck me after I wrote my response as I thought about your noise plots and the magnitude of the noise evident in each, that 1 Gohm resistor may indeed be noisy. As such, its noise will dominate the noise you are adding with your noise files. I decided to plot the Johnson noise due to the 1 Gohm resistor and assumed your C0 was 1 fF. Note its magnitude far exceeds the spectral density of the noise you are adding.

Shawn