The waveguide_simple photonic model is best suited for waveguide models which have no parameters that effect the mode properties (effective and group index, loss, or dispersion). Some applications of this photonic model include:
- Straight waveguides with a fixed cross-section
- Arced waveguides with a fixed radius and cross-section
Note that the cross-sectional properties (width, height) cannot be parameterized in this model, since they would change the effective and group indices, as well as the loss and dispersion. Furthermore, this model is not appropriate for waveguide arcs which allow the user to modify the arc radius, since curvature alters the mode properties as well. A photonic model which can handle such parameters is wg_parameterized.
The waveguide_simple photonic model supports statistical modeling. Users can choose an arbitrary number of statistical parameters and define their influence on the effective index, group index, and loss of the waveguide. For information on statistical CMLs, see Statistical CMLs.
Lumfoundry Templates: Waveguide Straight (Fixed), Waveguide straight (Statistical)
Quality Assurance Test: waveguide_simple QA
Statistically Enabled Parameters: neff, ng, loss
Supported Parameters: waveguide length, temperature
Tuning Support: No.
Interoperability with Cadence Virtuoso:
- Circuit design flows using INTERCONNECT model: Yes.
- Circuit design flow using photonic Verilog-A model: Yes.
Model Information
- Temperature dependency is only supported by neff using dneff/dT parameter.
-
Mode properties are defined as below:
Given a waveguide with effective index as a function of angular frequency, we can express the propagation constant in a taylor expansion:
$$\beta (\omega) \equiv {\omega \over c} = \beta_0 + \beta_1 (\omega-\omega_0) + {1 \over 2} \beta_2 (\omega-\omega_2)^2 + \cdots$$
$$\beta_m \equiv {d^m\beta \over d\omega^m}\bracevert_{\omega=\omega_0}$$
The waveguide's group index and dispersion are defined in terms of the first and second order terms of this expansion:
$$n_g = c\beta_1 \approx n_{eff}-\lambda {dn_{eff} \over d\lambda} $$
$$D = - {2\pi c \over \lambda^2 } \beta_2 = {d\beta_1 \over d\lambda} \approx -{\lambda \over c} \cdot {d^2n_{eff} \over d\lambda^2}$$
-
Time domain settings
number_of_taps_estimation is a digital filter setting that affects the time domain model. The frequency domain model is unaffected."disabled" : The time domain model will attempt to capture the following frequency dependence of the propagation constant, beta, and the loss$$beta(f) = 2*pi*neff(f0)*f0/c + 2*pi*ng(f0)/c*(f-f0)$$
$$loss(f) = loss(f0)$$where f is the frequency, f0 is the center frequency of the simulation. The propagation constant and loss will be most accurate near f0 and will deviate the most at the edges of the simulation bandwidth. The accuracy will improve as the simulation sample rate is increased because the higher simulation bandwidthresults in a smaller time step, dt. The accuracy will also improve for longer waveguides. In some cases, where the group delay of the specified waveguide length happens to be an exact multiple of dt, beta and loss (as described above) will be fully accurate across the entire simulated bandwidth.This is the recommended setting as it most accurately captures the most important effects of photonic waveguides in time domain simulation: the phase delay (neff), group delay (ng) and loss which are related to tuning, FSR, Q, IL and other key parameters in many devices.
"group delay" : The time domain model will attempt to capture the frequency dependence of neff, ng, dispersion and loss exactly as they are calculated in a frequency domain simulation, over the entire simulated bandwidth. As a result, this may compromise the accuracy of neff and ng (including at the center frequency), and can result in undesirable effects such as amplification of the optical signal at some frequencies. This setting is therefore not recommended. For best results, a high simulation sample rate may be required. Also, the accuracy will improve for longer waveguides and this setting may be desired for ultra-long waveguides (in devices without feedback) where dispersion could be important.