transistor noise in TRAN simulation

[Alexey Borodenkov]

Sorry, if this question is stupid, but anyway...

Is the transistor noise included in the transient simulation??
If yes, than which kinds of noise (thermal, flicker etc)?

Regards, Alex

 

[Richard Griffith]

Alexey Borodenkov wrote:

Sorry, if this question is stupid, but anyway...

Is the transistor noise included in the transient simulation??
If yes, than which kinds of noise (thermal, flicker etc)?

Regards, Alex

No simulators that I know of add nosie to transient simulations. This
would make the transient simulation non-repetable without specifying a
noise seed value.

Noise analysis usually means computing the transfer function (AC
analysis type) from all noise sources to the desired output node. Gain
from input to output nodes is then used to calculate input referred
noise and noise figure. Each noise source gets its "gain" to the output
node noted and the type of the noise source (thermal,flicker,etc) is
noted to allow plots vs freqency and tabulate information.

In spectreRF the PSS dynamic operating point is used to compute the
noise transfer function as a function of time point within the periodic
interval. This can cause single frequency noise points to be mapped back
over several harmonics to a single point.

 

[Christian Auner]

Richard Griffith wrote:

Alexey Borodenkov wrote:
Sorry, if this question is stupid, but anyway...

Is the transistor noise included in the transient simulation??
If yes, than which kinds of noise (thermal, flicker etc)?

Regards, Alex


No simulators that I know of add nosie to transient simulations. This
would make the transient simulation non-repetable without specifying a
noise seed value.

Some Mentor products can do a transient noise analysis. I don't know how
well this works. In any case you would have to run a number of simulations
(like MC) to get a meaningful result.

[...]

Chris

 

[Paul Muller]

Hi Alexey,

as Christian mentioned, Eldo has transient noise simulation mode, which
contains thermal noise and if the device parameters are set, also
flicker noise. It can be useful to analyze folding of flicker noise
components, e.g. in downsampling filters, but is not so trivial to
set-up and especially interprete.

You don't need to run MC-like multiple simulations, but your simulation
time anyway gets quite long due to the extra calculations at each
simulation point.

If you have MGC products, have a look at the Eldo doc, there's a full
chapter about it. If you don't, except flicker noise folding, which can
be tough to calculate, you should be able to do almost everything with
frequency domain simulations.

Paul

 

[Andrew Beckett]

A better approach may be to use something like spectre's tdnoise analysis (part
of pnoise in SpectreRF).

There are issues with any transient noise based approach - see my previous
append:

http://groups.google.com/groups?q=transient+noise+andrewb+group:comp.cad.cadence&hl=en&lr=&ie=UTF-8&group=comp.cad.cadence&selm=dp118vcp36t6rn3i0bv4o9dgpqfvt50rn1%404ax.com&rnum=1

Regards,

Andrew.

On Tue, 25 May 2004 08:39:53 +0200, Paul Muller <paul.muller-at-epfl.ch> wrote:

Hi Alexey,

as Christian mentioned, Eldo has transient noise simulation mode, which
contains thermal noise and if the device parameters are set, also
flicker noise. It can be useful to analyze folding of flicker noise
components, e.g. in downsampling filters, but is not so trivial to
set-up and especially interprete.

You don't need to run MC-like multiple simulations, but your simulation
time anyway gets quite long due to the extra calculations at each
simulation point.

If you have MGC products, have a look at the Eldo doc, there's a full
chapter about it. If you don't, except flicker noise folding, which can
be tough to calculate, you should be able to do almost everything with
frequency domain simulations.

Paul

--
Andrew Beckett
Senior Technical Leader
Custom IC Solutions
Cadence Design Systems Ltd

 

[Paul Muller]

Andrew,

thank you for this info on tdnoise, I was not aware of this feature!
This seems indeed to be more interesting than transient noise analysis.
BTW, as an additional comment to Alexey, MGC doesn't make a big deal out
of transient noise analysis, which seems to show that it may not be so
useful...

Paul

Andrew Beckett wrote:

A better approach may be to use something like spectre's tdnoise analysis (part
of pnoise in SpectreRF).