Monte Carlo Yield Analysis

Real components have manufacturing tolerances. A resistor labeled “10k” might be anywhere from 9.5k to 10.5k (5% tolerance). Monte Carlo analysis simulates your circuit hundreds of times with randomized component values to see how often the design meets spec.

Goal

Determine whether an RC lowpass filter with 5% resistors and 10% capacitors reliably hits a 10 kHz bandwidth target.

Workflow

Define the circuit

Start with a known-good netlist. You can generate one with create_from_template:
```
{
  "template_name": "rc_lowpass",
  "params": {"r": "1.59k", "c": "10n"}
}
```
Read the generated .cir file content for the next step.

Run Monte Carlo

Use the monte_carlo tool with component tolerances:

{
  "netlist_text": "* RC Lowpass\nV1 in 0 AC 1\nR1 in out 1.59k\nC1 out 0 10n\n.ac dec 100 1 1meg\n.end",
  "n_runs": 100,
  "tolerances": {
    "R1": 0.05,
    "C1": 0.10
  },
  "seed": 42
}

This runs 100 simulations, each time varying R1 by up to +/-5% and C1 by up to +/-10% using a normal distribution.

Analyze results

The tool returns per-run .raw file paths. For each run, use measure_bandwidth to get the -3dB frequency, then compute statistics across all runs:
- Mean bandwidth
- Standard deviation
- Min / max range
- Yield: percentage of runs where bandwidth falls within your spec window (e.g., 8 kHz to 12 kHz)
Interpret

With 5% resistors and 10% capacitors, expect roughly +/-15% variation in cutoff frequency (tolerances add in quadrature for an RC product). If yield is too low, consider tighter-tolerance components or a design with more margin.

Key takeaway

Monte Carlo analysis answers “will this design work in production?” rather than “does it work with ideal components?” Running 100 iterations typically takes under 30 seconds since each AC simulation is fast. For high-confidence yield estimates, use 500+ runs.