PSS with another parameter than time possible ?

[Stefan Joeres]

Hi alltogether,

do you know whether it's possible in SpectreRF to do a PSS-analysis of an
oscillator with a given frequency and to find the steady-state-solution by
varying another parameter (e.g. delay for a ringoscillator) instead of
finding the resulting frequency ?

I appreciate any hint or answer,

Stefan

 

[Andrew Beckett]

Stefan,

I don't understand what you mean. Can you elaborate? PSS finds the
steady state solution over a period of operation of the circuit. In
autonomous mode it "locks" onto the period - I'm not sure however
what you mean by "varying" another parameter - are you wanting to
do a sweep or something? I just don't understand what you're describing
here.

Regards,

Andrew.

On Mon, 10 Nov 2003 18:05:58 +0100, "Stefan Joeres" <joeres@halifax.rwth-aachen.de> wrote:

Hi alltogether,

do you know whether it's possible in SpectreRF to do a PSS-analysis of an
oscillator with a given frequency and to find the steady-state-solution by
varying another parameter (e.g. delay for a ringoscillator) instead of
finding the resulting frequency ?

I appreciate any hint or answer,

Stefan

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

 

[Stefan Joeres]

I don't understand what you mean. Can you elaborate? PSS finds the
steady state solution over a period of operation of the circuit. In
autonomous mode it "locks" onto the period - I'm not sure however
what you mean by "varying" another parameter - are you wanting to
do a sweep or something? I just don't understand what you're describing
here.

Yes, it "locks" onto the period by varying the initial conditions and
adjusting the time after which it finds the steady state solution ( = one
period ).
No I wonder whether it would be possible to change one parameter (e.g. a
control voltage which adjusts the delay of a ring oscillator) and to fix the
time after which the steady state solution should be reached.

To be more specific :
I've got a ring oscillator and I want to measure the phase noise at a given
oscillation frequency. At the moment, I need to do a transient analysis with
a sweep of the parameter controlling the delay of each stage, find the value
at which the ring oscillator oscillates with the given frequency, enter this
parameters into the variables by hand and to do the PSS and Pnoise analysis
with that parameter set - as you may know it's quite hard to get this
parameter precisely without a large number of points for the sweep.

Please ask again, if this isn't fully understandable.

Stefan Joeres

 

[Andrew Beckett]

Hi Stefan,

No, that's not possible. What you're describing is effectively a phase-locked loop.
There's a whole difference between the simulator having to solve another
unknown in the boundary conditions using shooting methods (which I think
roughly describes what it is doing...), and varying something major like
the control voltage, allowing it to settle, and monitoring when it has
locked phase in a stable way in a reasonable time. After all PLL's
generally take quite a bit of time to lock (relatively speaking). I'm not sure
there's any way of doing this in the same sort of timescales that autonomous
PSS mode uses - because it strikes me as being a much harder problem.

You might want to look at Ken Kundert's paper " Predicting the phase noise and
jitter of PLL-based frequency synthesizers" (see the front page of
http://www.designers-guide.com for a link to the paper), for
ways of characterizing your oscillator and modelling the PLL. Just an
idea; may not be what you want, but worth a look I think.

Regards,

Andrew.


On Tue, 11 Nov 2003 00:00:20 +0100, "Stefan Joeres" <joeres@halifax.rwth-aachen.de> wrote:

I don't understand what you mean. Can you elaborate? PSS finds the
steady state solution over a period of operation of the circuit. In
autonomous mode it "locks" onto the period - I'm not sure however
what you mean by "varying" another parameter - are you wanting to
do a sweep or something? I just don't understand what you're describing
here.

Yes, it "locks" onto the period by varying the initial conditions and
adjusting the time after which it finds the steady state solution ( = one
period ).
No I wonder whether it would be possible to change one parameter (e.g. a
control voltage which adjusts the delay of a ring oscillator) and to fix the
time after which the steady state solution should be reached.

To be more specific :
I've got a ring oscillator and I want to measure the phase noise at a given
oscillation frequency. At the moment, I need to do a transient analysis with
a sweep of the parameter controlling the delay of each stage, find the value
at which the ring oscillator oscillates with the given frequency, enter this
parameters into the variables by hand and to do the PSS and Pnoise analysis
with that parameter set - as you may know it's quite hard to get this
parameter precisely without a large number of points for the sweep.

Please ask again, if this isn't fully understandable.

Stefan Joeres

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