Component Instance
Understanding Component Instance
In RF and microwave design tools, every element you place on a schematic, whether a lumped capacitor, a packaged transistor, a microstrip line, or an entire imported S-parameter file, becomes a component instance. The distinction between a model and an instance is one of the most important concepts in electronic design automation. A model is a template; it describes how a class of device behaves through equations, lookup tables, or measured data. An instance is one specific use of that template, given a unique name and connected into the larger circuit. In the same way a class in software is instantiated into many objects, a SPICE or harmonic-balance model is instantiated into many circuit elements, each with its own state.
Instance name, model reference, and parameters
Every component instance is built from three essential parts. First is the instance name, also called the reference designator, which must be unique within its scope so the solver can address it without ambiguity. Second is the model reference, a pointer to the shared library definition or device model that supplies the underlying physics. Third is the parameter set, which lets a designer override defaults for this placement only, for example setting a specific capacitance, line length, or bias point. When a circuit simulator parses a netlist, it expands every instance into a set of equations, stamps them into the system matrix, and solves the whole network simultaneously. Changing one instance parameter affects only that placement, never the shared model or its other instances.
Instances in hierarchical and parametric design
Component instances become especially powerful inside hierarchical designs. A subcircuit, such as a single amplifier stage or a matching network, can itself be instantiated many times to build a multi-stage chain or an array. Each higher-level instance carries its own copy of the parameters passed down to the elements inside it, so one well-characterized building block can be reused with different values throughout a layout. This is what makes parametric sweeps, Monte Carlo yield analysis, and optimization practical. The tool varies an instance parameter, re-evaluates only the affected elements, and reports the response. In RF practice this might mean sweeping the length of one transmission-line instance to tune a match, or varying the value of a single bypass capacitor instance to study stability without disturbing the rest of the network.
Why the model and instance distinction matters in RF
At microwave frequencies the parasitics and package effects captured in a device model dominate performance, so reusing a single validated model across many instances keeps a design consistent and trustworthy. If a foundry updates its transistor model, every instance referencing it inherits the correction automatically, while each placement keeps its own bias and geometry. Conversely, accidental duplication of instance names, or silently sharing parameters that should be independent, is a common source of simulation error. Disciplined naming, clear parameter passing, and careful tracking of which instances reference which models are core skills for anyone assembling RF front ends, converters, or integrated assemblies in a simulator.
Counting Instances Across Hierarchy
Ntotal = Ntop × Nsub
Where Ntotal = total number of leaf instances of the model in the flattened netlist, Ntop = number of times the parent subcircuit is instantiated at the top level, and Nsub = number of instances of the model inside one copy of that subcircuit. The simulator flattens the hierarchy before solving, so a model placed twice inside a stage that is used in a four-stage chain yields Ntotal = 4 × 2 = 8 independent leaf instances, each with its own solved currents and voltages.
Model versus Instance Reference
| Attribute | Belongs to Model | Belongs to Instance |
|---|---|---|
| Defining equations or measured data | Yes | No |
| Default parameter values | Yes | No |
| Unique reference designator | No | Yes |
| Parameter overrides for one placement | No | Yes |
| Node and port connections | No | Yes |
| Per-element solved currents and voltages | No | Yes |
Frequently Asked Questions
What is a component instance?
A component instance is a single placed, uniquely named copy of a library model in a circuit netlist or schematic, with its own parameter values and node connections. The shared model defines the behavior, while the instance defines this particular usage. One model can be instantiated many times across a design, and the simulator solves each instance independently.
How does a component instance differ from a model definition?
The model definition holds the reusable equations and default parameters, while the instance is one concrete placement of that model with a unique reference designator, specific parameter overrides, and its own connections. Editing an instance parameter changes only that placement; editing the model changes every instance that references it.
Why must each component instance have a unique name?
The unique instance name, or reference designator, lets the simulator and netlist address each element separately. Without it the tool cannot assign distinct node connections, apply per-placement parameter overrides, or report per-element currents, voltages, and power. Duplicate instance names are a common cause of netlist errors.