Edited by Leland Teschler

PIN diodes are named for a region of undoped intrinsic semiconductor lying between their p and n-type material. This area eliminates capacitance effects between the n and p-type areas that can otherwise lower device impedance at higher frequencies.

A simple PIN diode switch is turned on by applying a dc bias that makes the diode conduct. This lets an RF source pass a signal through the diode to a load. Applying a negative dc bias turns off the diode by widening the diode depletion layer and thus blocks the RF signal. Multithrow switches can be constructed by connecting multiple diodes to a common junction and forward biasing only those in the conducting paths.

An alternative connection scheme puts the PIN diode across the input line in parallel with the load. Here forward bias makes the diode conduct and thus reflect RF energy back to the source. Negative bias lets RF energy bypass the diode and get to the load.

One problem with PIN-diode switches is that they dissipate power when biased on. So RF switches incorporated into integrated circuits are more likely to use MOSFETs rather than diodes as switching elements. The principle of operation is the same: Application of a dc voltage turns the switching element on or off. The fabrication technology used is generally silicon-on-insulator or similar techniques that limit the energy dissipated during switching.

PIN-based switches are complicated to design for use in the microwave range. Thus, RF switches handling signals on the order of gigahertz typically use GaAs switching elements such as MESFETs (metal semiconductor FETs), basically FETs with a Schottky junction for a gate.

There has been recent development work toward RF switches that employ MEMS technology. The typical structure is that of a thin metallic film suspended over a dielectric material that serves as a contact. Application of an electrostatic field pulls the membrane down to touch the bottom piece. (Strictly speaking, a MEMs switch is electromechanical because it has a moving part.) The advantage of this approach is low switching energy and potentially lowcost processing via straightforward IC techniques.