Designer’s Guide Community

Dave F.

Ex Member

Identifying Noise Sources
Jun 15^th, 2015, 11:31am

Hi,

I am trying to understand the noise that I am seeing in my circuit. I am having difficulty reducing the noise, and it seems much higher than what would be expected after reading various textbooks.

I am now trying to mess with the model parameters to zero out all noise sources to see which type is contributing the most (thermal, flicker, shot, ...). I have zeroed out every noise-related parameter that I can find, but my noise is almost entirely unchanged. Am I missing some parameters or some fundamental aspect as to how the simulator works?

I am using Spectre. These are the parameters I have modified so far:

+ fnoimod = 0.0
+ tnoimod = 0.0
+ ef = 1.0
+ noia = 0.0
+ noib = 0.0
+ noic = 0.0
+ em = 0
+ ntnoi = 0.0
+ lintnoi = 0
+ af = 1.0
+ kf = 0.0
+ tnoia = 0.0
+ tnoib = 0.0
+ rnoia = 0.0
+ rnoib = 0.0

I am running transient noise and regular noise

Thanks

Back to top

IP Logged

Geoffrey_Coram

Senior Fellow

Offline

Posts: 1998
Massachusetts, USA

Re: Identifying Noise Sources
Reply #1 - Jun 18^th, 2015, 10:46am

There should be a way to have Spectre tell you the noise contributions not only from each device but from each noise source within the device; eg, you should be able to see if it's the channel thermal noise, or the noise of the resistors (rd, rs, rg), or the flicker noise.

I also think you can turn off the noise of a specific device, something like is_noisy=no (perhaps as an instance parameter?).

Back to top

If at first you do succeed, STOP, raise your standards, and stop wasting your time.

IP Logged

sheldon

Community Fellow

Offline

Posts: 751

Re: Identifying Noise Sources
Reply #2 - Jun 26^th, 2015, 5:55pm

Quanticles,

What you are looking for is the noise summary in ADE. It allows
you to search the noise sources to identify the primary contributors.
The noise summary only works for small signal analysis, noise and
pnoise. As Geoffrey says, the noise summary will report on the
noise contributors within the device. Of course the effects need
to be included in the models.

Sheldon

Back to top

IP Logged