The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Measurements >> Phase Noise and Jitter Measurements >> [help!]question about time domain noise simulation of a divider
https://designers-guide.org/forum/YaBB.pl?num=1235706435

Message started by lonemy on Feb 26^th, 2009, 7:47pm

Title: [help!]question about time domain noise simulation of a divider
Post by lonemy on Feb 26^th, 2009, 7:47pm

Hello all:

i have a basic question regarding time domain noise simulation of a simple divider.

I have a digital divide-by-2 with the following settings.

input : 50 MHz square wave
output : 25 MHz

pss setup:

beat frequency : 25 MHz
number of harmonics : 50
accuracy defaults : moderate
additional time for stabilization : 45n

pnoise setup:

start : 1
stop : 50 MHz
maximum sideband : 20
output voltage : positive output node : /net061 (the output)
negative output node : /gnd!
input source : none
noise type : time domain
number of points : 1
add specific points : 11.5n, 51.5n, 91.5n (the output is 25M, so i choose three threashold crossing time)

but the simulator reminded me: deleted time point 5.15e-08 in pnoise analysis that exceeds periodic interval
deleted time point 9.15e-08 in pnoise analysis that exceeds periodic interval

After the simulation finishes, I go to Results -> Main Form -> tdnoise -> output noise -> total noise -> sweep:spectrum
-> magnitude -> plot
i cant get a correct plot, but like this:

could you tell me where is my mistake?

thanks very much!

-lonemy

Title: Re: [help!]question about time domain noise simulation of a divider
Post by Andrew Beckett on Feb 27^th, 2009, 2:16am

The PSS analysis is analysing a single period of the PSS fundamental (25MHz).

Consequently you cannot specify time points outside of the period - since the period is 40ns, specifying a time of 51.5n or 91.5n makes no sense. In fact, what would it tell you? If you've got a steadystate solution, the noise at 51.5n and 91.5n would be the same as it is at 11.5n (remember that this is a small-signal noise analysis, not a transient noise analysis, where the noise sources are computed randomly at each timestep).

So your mistake was to specify times exceeding the periodic interval, as the simulator message stated.

Regards,

Andrew.

Title: Re: [help!]question about time domain noise simulation of a divider
Post by lonemy on Feb 27^th, 2009, 4:21am

Andrew.thank you very much!
i simulated the circuit again and modified nothing except the blank of add specific points(i deleted 51.5n and 91.5n).
but i can't get correct plot neither(the result changed nothing ).
i think there must be some thing wrong with my simulation settings or process.
could you tell me where my mistakes are?
thank you for your time.

Title: Re: [help!]question about time domain noise simulation of a divider
Post by Frank Wiedmann on Feb 27^th, 2009, 5:20am

You should use the following settings:

pss setup:

beat frequency : 25 MHz
number of harmonics : (what you want to display, does not influence pnoise results)
accuracy defaults : moderate
additional time for stabilization : 45n

pnoise setup:

start : 1
stop : 12.5 MHz (half the "beat frequency")
maximum sideband : (depends on the bandwith of your circuit, see http://www.designers-guide.org/Forum/YaBB.pl?num=1233801262/1#1)
output voltage : positive output node : /net061 (the output)
negative output node : /gnd!
input source : none
noise type : time domain
number of points : 0 (you don't need the point at timeindex=0)
add specific points : 11.5n (the threshold crossing time in the pss analysis)

If you are using a recent version of Spectre, you should use noise type jitter instead of time domain.

Why do you think that your result is incorrect?

Title: Re: [help!]question about time domain noise simulation of a divider
Post by lonemy on Feb 27^th, 2009, 6:28am

Dear Frank:

thanks for your answer.

with your settings, i simulated my circuit again.

i guess this result was correct.

thank you again!

-lonemy

Title: Re: [help!]question about time domain noise simulation of a divider
Post by Andrew Beckett on Feb 27^th, 2009, 6:54am

Thanks Frank - I neglected the fact that lonemy was sweeping past half the PSS fundamental - I should have spotted that.

Andrew.