PHASE NOISE Analysis in Spectre

[michael_skoufis]

I am trying to run phase noise analysis for a voltage controlled coupled
ring oscillator implemented using differential buffers. Technology is
TSMC 0.18 micron. The problem is that PSS does not converge. As far as I
understand this happens because simulator thinks oscillations don’t ever
start. However according to my transient analysis this is not true. I
have run a lot of simulations on the schematic and I'm confident it works
just fine. My settings for the PSS are the following:

Beat frequency --- 1GHz
Output Harmonics --- 7
Accuracy Default ---moderate
tstab --- 1ns

I have selected “Oscillator” and specified the oscillator and reference
nodes as the output nodes of the last-stage differential buffer.

Then I selected Pnoise Analysis in the ADE window. I set Sweeptype to
relative and Relative Harmonic to 1. Frequency Sweep Range starts at 10K
and stops at 10M. Sweept type is Linear with a step size of 10K. Maximum
sideband is set to 7. Output is set to Voltage and for the Positive and
Negative outputs I have again selected the two outputs of the differential
buffer. Input Source is none.

Simulation runs for a number of loops and then it fails without converging
(for PSS). Any help is appreciated.

Thanks, Mike
Southern Illinois University

 

[Thomas Ussmueller]

Hello,

does the pss or the pnoise simulation not converge? If the pss does not
converge, have a look at the options for the simulation and try playing
around with the steadyratio paramter (for details see the SpectreRF user
guide).

Best regards
Thomas


michael_skoufis wrote:

I am trying to run phase noise analysis for a voltage controlled coupled
ring oscillator implemented using differential buffers. Technology is
TSMC 0.18 micron. The problem is that PSS does not converge. As far as I
understand this happens because simulator thinks oscillations don’t ever
start. However according to my transient analysis this is not true. I
have run a lot of simulations on the schematic and I'm confident it works
just fine. My settings for the PSS are the following:

Beat frequency --- 1GHz
Output Harmonics --- 7
Accuracy Default ---moderate
tstab --- 1ns

I have selected “Oscillator” and specified the oscillator and reference
nodes as the output nodes of the last-stage differential buffer.

Then I selected Pnoise Analysis in the ADE window. I set Sweeptype to
relative and Relative Harmonic to 1. Frequency Sweep Range starts at 10K
and stops at 10M. Sweept type is Linear with a step size of 10K. Maximum
sideband is set to 7. Output is set to Voltage and for the Positive and
Negative outputs I have again selected the two outputs of the differential
buffer. Input Source is none.

Simulation runs for a number of loops and then it fails without converging
(for PSS). Any help is appreciated.

Thanks, Mike
Southern Illinois University

 

[michael_skoufis]

Thanks for your response. I got it to work. I am not sure what the
problem was. I played with tstab and made some more changes in my
schematic.

Mike