photodetector_simple - CML Compiler Model

The photodetector_simple photonic model was created to model photodetector (PD) elements. It supports photodetector elements with up to two optical inputs, as well as multiple (up to two) electrical anode ports. It provides an ability to model photodetectors which include both shot and thermal noise, as well as saturation effects. This model can also include second and third order harmonic responsivities that arise from nonlinearities in the system.

The photodetector_simple photonic model supports statistical modeling. Users can choose an arbitrary number of statistical parameters and define their influence on the responsivity, electro-optic bandwidth and dark current of the photo detector. For information on statistical CMLs, see Statistical CMLs.

Note: If harmonic responsivity data is included, statistical support will be disabled.

Lumfoundry Templates: Photodetector, Photodetector (Statistical), Harmonic Photodetector

Quality Assurance Test: photodetector_simple QA

Statistical Modeling Support: responsivity, bandwidth, dark current

Supported Parameters: None 

Tuning Support: None

Platform and Model Support

[[snippet||51412253389715]]

Model Information

• This model supports thermal noise, shot noise and power saturation. 
• This model does not support pcells.

Electrical Equivalent Circuit for INTERCONNECT model

The 3-dB bandwidth of the photodetector is determined by two independent mechanisms: the carrier transit-time limit and the electrical bandwidth of the PIN diode. In the INTERCONNECT optical model, these effects are represented by separate electrical low-pass filters (LPFs). When the model is used in INTERCONNECT or INTERCONNECT as an Engine in OptoCompiler, both LPFs are enabled and together define the photodetector's overall frequency response.

The INTERCONNECT model can also be used in conjunction with Cadence Spectre for electro-optical co-simulation. In this configuration, the LPF associated with the PIN diode electrical bandwidth is automatically disabled to avoid duplicating the bandwidth limitation. Instead, the electrical response of the photodetector is modeled using a physical RC equivalent circuit generated by CML Compiler and simulated directly in Spectre. The transit-time LPF remains active within the optical model, while the RC equivalent circuit accurately captures the electrical bandwidth behavior of the device.

This section describes the use of an electrical equivalent circuit for the case of Spectre-INTERCONNECT co-simulation. The below figure shows a schematic of the electrical equivalent circuit for a photodetector, which can be created by running CML Compiler with the “--interconnect_virtuoso” command line option. 

The figure below illustrates the photodetector implementation used for electro-optical co-simulation, which combines the INTERCONNECT optical model with its electrical equivalent circuit. The junction voltage across Cj is used to bias the optical INTERCONNECT model. During co-simulation, Spectre simulates the electrical equivalent circuit, capturing the photodetector's electrical bandwidth and loading effects, while INTERCONNECT simulates the optical behavior of the device. To avoid double-counting the electrical bandwidth limitation, the internal LPF that represents the PIN diode electrical bandwidth within the INTERCONNECT model is automatically disabled. The transit-time LPF remains active and continues to model the carrier transit-time response of the photodetector.

Electrical equivalent circuit for Photonic Verilog-A model

As shown below, the photonic Verilog-A photodetector model integrates both the photonic device model and its electrical equivalent circuit within a single compact model. The electrical equivalent circuit captures the photodetector's electrical characteristics, including bandwidth and loading effects, while the photonic model describes its optical behavior. Unlike the INTERCONNECT co-simulation flow, where the electrical equivalent circuit is implemented separately from the optical model, the Verilog-A implementation combines both representations into a single model that is solved simultaneously during simulation.