questions on current noise simulation

Hi, guyes,

I ran into problem when I simulate input inferred equivalent noise with AMI 0.5um CMOS process with cadence spectre. I followed noise analysis note in https://www.ece.tamu.edu/~spalermo/ecen474/noise_sim_notes.pdf and set the flicker noise simulation parameters:

[For NMOS model, add the following to the end of your model :
KF=0.276E-25 &
AF=1.53 &
)
For PMOS model, add the following to the end of your model :
KF=0.466E-26 &
AF=1.61]

In noise analysis, I set iprobe3 as output and Iin as input current source. Iin DC value is set at 10nA for DC bias; Iin AC magnitude is set at 1 for noise simulation.
To display the input referred current noise, I selected RESULT->DIRECT PLOT->EQUIVALENT INPUT NOISE. However, I got quite strange input referred current noise result: Noise in high freq. domain is higher than that in low freq. domain. In theory, should low freq noise be higher than high freq noise because of flicker noise?
Anyone can help me out?

I got quite strange input referred current noise result. Noise in high freq. domain is higher than that in low freq.

Click to expand...

What's strange with this result? Any MOSFET has it, review your text books. Gate capacitance converts voltage noise into current noise.

what is the transfer function of the circuit? Is it working correctly? Input referred noise can look unusual sometimes as it is taken by applying the input-output transfer function to the output noise in reverse. If the frequency response of the system is unusual then the output noise can look sensible but input referred can look strange.

Keith

FvM said:

What's strange with this result? Any MOSFET has it, review your text books. Gate capacitance converts voltage noise into current noise.

Click to expand...

Because of flicker noise, should low freq noise be higher than high freq noise?

- - - Updated - - -

keith1200rs said:

what is the transfer function of the circuit? Is it working correctly? Input referred noise can look unusual sometimes as it is taken by applying the input-output transfer function to the output noise in reverse. If the frequency response of the system is unusual then the output noise can look sensible but input referred can look strange.

Keith

Click to expand...

Sound reasonable.

Is there any problem in my noise analysis setting?

Because of flicker noise, should low freq noise be higher than high freq noise?

Click to expand...

Voltage noise yes, current noise no. See a typical MOSFET noise characteristic:

FvM said:

Voltage noise yes, current noise no. See a typical MOSFET noise characteristic:

Click to expand...

Thanks. I understand it a little bit. Define input referred voltage noise as Vnin, and input capacitance as Cin, input referred current as Inin.

Inin=Vnin*j*w*Cin

Thus, high freq Inin component is larger than low freq Inin component.

BTW, I find a different image about input referred noise plot from

https://www.planetanalog.com/document.asp?doc_id=528133

. Any comments?

Any comments?

Click to expand...

pA range noise densities belong to bipolar devices, not MOSFET.

hyleeinhit said:

Any comments?

View attachment 99031

Click to expand...

Now this is a current noise vs. resistance plot, not noise vs. frequency. Apples & pears!

erikl said:

Now this is a current noise vs. resistance plot, not noise vs. frequency. Apples & pears!

Click to expand...

Opps! My mistake....Thanks for your reminder!