AnsweredAssumed Answered

Tips for using OscPort oscillator analysis

Question asked by JERRY_MARKETOS on Jun 16, 2004
Tips for better chances of success when using the OSCPORT
automated nonlinear closed loop oscillator analysis:

Oscillator circuits can be tricky to analyze in closed loop form.
Sometimes, the oscillation frequency in closed loop form can shift
due to impedance interaction of the two ports that were previously
used to analyze the circuit in open loop form.  Also, the higher the
loaded Q of the oscillator, the more difficult it generally is to analyze
steady state large signal conditions.  (The easier it is to miss the
frequency with steady state oscillation conditions).

The OscPort module provides an automated means for finding
closed loop detection of oscillator circuit operation.  To best apply
the OscPort, here are some tips:

1) Choice of node to locate OscPort:  It's important to locate the
OscPort at a node which exhibits certain impedance conditions
at the oscillation frequency.  (The Real portion of the port's Zin
is negative and the Imag portion of the port's Zin is zero).  There
can exist situations where the closed loop conditions change
sufficiently to invalidate predictions that may have been made in
the open loop analysis - this is part of the challenge of oscillator
circuit simulation.  The actual location of the OscPort may end
up being somewhere on the input of the circuit or somewhere
on the output, in order to satisfy the criteria above.

2) Add a nonlinear (Harbec) analysis to the "Analysis" folder.
In the "Harbec Options" window, "Oscillator" tab, set the minimum
and maximum search frequency, and number of points for small
signal analysis.  Note that searching too narrow a band, or searching
an insufficient number of data points...  can result in not finding
oscillation conditions.  You may need to try different frequency
intervals, or increase the number of data points in a given interval.

3) Check the checkbox at bottom for "Display spectrum and waveform graphs".
This will cause the Harbec module to create the oscillator analysis equations
and output measurements.

4) Click the button "Find initial oscillator port frequency".  The Harbec module should
find a frequency and display it in the dialogue window.  If a frequency is not found,
close the Harbec options dialogue box and re-run the Harbec analysis by right-clicking
the Harbec analysis in the file tree and choosing "recalculate".

At first running, a set of output equations and graphs will be created.
Open the graph "Oscillation criteria for HB"
(note if the filename used for the Harbec analysis is too long, the graph may
come out with a short name such as Graph1).

The graph shows a sweep of the OscPort impedance.  In order to detect
oscillation, the OscPort is looking for the presence of three conditions:
a) The phase angle of the impedance at the OscPort must be sufficiently near zero
b) The impedance signature contains a relatively sharp change
c) The Real portion of impedance magnitude is <0  (negative)

5) If the circuit does not demonstrate oscillation conditions in the predefined frequency
band, move the OscPort to a different node.  The OscPort may be placed anywhere
in the feedback path from output to input, or within nodes associated with the resonator.
For high Q oscillators, make sure to use enough frequency data points to avoid missing
the sharp resonance conditions.

6) When the frequency of oscillation is found, proceed to the next step:  nonlinear analysis.
At this point, click the radio button "use oscillator solver" on the "oscillator" tab of the nonlinear
analysis.  This activates the Newton algorithm of the oscillator analysis, which seeks to solve
for Real and Imaginary portions of the first Harmonic content to be zero.

Automated analysis of an oscillator circuit involves conflicting requirements.
On one hand, we desire high sensitivity to very small changes (in Z, I, V, etc.)  On the other
hand, the numeric processing associated with the higher sensitivity promotes
stronger likelihood of the existence of natural numerical errors.  This tends to
increase with circuit complexity.