Hi,

I am running pnoise simulations using spectre on chargepump.

I get a noise profile of chargepump which has flicker noise+thermal noise floor.

analytically also it should be a scaled version of  current source noise  with duty cycle of the square wave as shown in.. http://www.designers-guide.org/Forum/YaBB.pl?num=1389088736/1#1


since this is already 2 sided noise PSD ( which is in the form of 4kTgm )  I think we just need to integrate the noise psd (in the pll bandwidth) and multiply with PLL transfer function to output while estimating rms jitter.


However I have noticed many designers and journal papers using a factor of after integration.


Please clarify whether a factor of 2 is needed or not and why ?


Thanks,
yvkrishna