How do I design an impedance matching network for an RF plasma chamber?

Designing an impedance matching network for an RF plasma chamber requires transforming the varying complex impedance of the plasma load (typically 2-20 ohms resistive with a large capacitive reactance of -50 to -500 ohms) to the 50-ohm output impedance of the RF generator, while adapting in real time to impedance changes during plasma ignition, process transitions, and recipe changes. The most common topology is the L-network (series-shunt configuration) using two variable capacitors: a series (tune) capacitor and a shunt (load) capacitor. The L-network can match any load impedance inside the corresponding Smith chart circle. Variable capacitors are typically vacuum-variable (10-1000 pF range, 5 kV voltage rating, stepper motor driven) for medium to high power applications, or electronically tuned (varactor or PIN diode switched capacitor banks) for fastest tuning speed. The auto-tuning algorithm continuously monitors forward and reflected power (via a directional coupler) and adjusts the two capacitor values to minimize reflected power, using PID control or gradient-descent optimization. Tuning speed must be fast enough to track plasma impedance changes: typical settling time is 50-500 ms for motor-driven vacuum capacitors, or < 1 ms for electronic tuning. The matching network must handle the full RF power (up to 10+ kW) with low insertion loss (< 5% power lost as heat in the network).