[SOLVED] PSS & Pnoise simulation for sample and hold circuit

[je01911]

Hi guys

I am trying to simulate the PSS & Pnoise of simple sample and for circuit for thermal noise simulation.



The clock signal for gate voltage swings from 0 to 2.4V (for rail to rail sampling)

The input source (PORT0) generates sine signal from 0 to 1.2V

Accoring to kT/C noise theory, I expect to noise power as much as 37.692nV²/Hz because of 100fF load capacitance.

However, the simulation results shows just 0.3464fV²/Hz at flat zone (white noise by thermal) like



I can't understand why the value computed by theory is different with simulation results in aspect of thermal noise.

If you knows the reason, please help me.

 

[crutschow]

Does your simulator actually calculate kT/C noise? I don't think many simulators do.

 

[pancho_hideboo]

Accoring to kT/C noise theory,
I expect to noise power as much as 37.692nV2/Hz because of 100fF load capacitance.
However, the simulation results shows just 0.3464fV2/Hz at flat zone (white noise by thermal) like

Integrate regarding frequency.
Then you can get same value for both static case(dc-noise) and sampled case(pss-pnoise).

See https://www.designers-guide.org/Forum/YaBB.pl?num=1258339986/#13

 

[je01911]

Thank you for your reply!

However, I have some question.

The mean of "Integrate regarding frequency" is like this?



However, In my case, the integrated noise power is changed according to range of integration.

First, in my opinion, the integrated noise power is limited at some frequency because of low pass filter of switch and load capacitor. However, the integrated noise power is increased constantly. Is it right?

Second, if it is right, How much I should set the range of frequency when I integrate the noise ?

 

[helpmejerry]

Hi je01911,

I tried to measure kT/C noise in the circuit you provided but got strange results and it did not converge.
I have attached a simple low pass filter, where we have two switches and a capacitor. The clock frequency is 10Mhz so the equivalent resistance should be approx 1Mohm with a 100fF cap. With a 1pF cap the pole should be at aprox 160kHz. I used complementary switches and non overlapping clock signals.



The noise is aprox 20nV/sqrt(Hz) so it is close to the theoretical calculation.
Hope this helps.

BR Jerry