Solid-state PIN switches come in two main topologies, reflective and absorptive, and the choice between them determines what happens to the signal that is not selected. Isolation and insertion loss are the two figures of merit that matter most in practice, followed by switching time. This article explains each of these terms in plain language, with reference to PIN switch topologies as covered by an independent English publication on microwave control components.
What are reflective and absorptive topologies in solid-state PIN switches?
A PIN diode is a diode with a thin intrinsic layer between its P and N regions. When it is forward biased it conducts and behaves like a small resistor; when it is reverse biased it behaves like a small capacitor, blocking direct current while passing radio frequencies with high impedance. A switch built from PIN diodes uses this contrast to route a signal.
In a reflective topology, the unselected port is left unterminated in effect. The PIN diode in that path is biased so that it reflects the incoming signal back toward the source. The switch itself does not absorb the unwanted power; it sends it back. This is simpler, often cheaper, and usually gives slightly better insertion loss in the selected path because there is one less termination to pass through. The trade-off is that the reflected power travels back into the preceding stage, which must be able to tolerate a poor match on its output.
In an absorptive topology, the unselected port is terminated internally in a matched load, typically 50 ohms. The unwanted signal is dissipated as heat inside the switch rather than reflected. This keeps the impedance seen by the source close to 50 ohms regardless of switch state, which protects amplifiers and other stages that are sensitive to mismatch. The cost is a small amount of extra loss in the terminated path and a slightly more complex circuit, since the termination must itself be switched in and out correctly.
A practical rule: use an absorptive switch when the surrounding circuit needs to see a constant impedance, and a reflective switch when the source can absorb the reflection and every tenth of a decibel in the through path counts.
How does isolation affect solid-state PIN switch performance?
Isolation is the measure of how much of a signal leaks into a port that should be off. It is expressed in decibels: an isolation of 60 dB means that one thousandth of the power present at the input reaches the off port. Higher numbers are better.
Isolation matters for three everyday reasons. First, a leaking signal can drive a receiver into compression or create intermodulation products when a strong transmitter shares the chain with a sensitive receive path. Second, leakage between channels breaks the separation between test paths and can corrupt measurements. Third, in a transmit chain, poor isolation can let power back into an oscillator or driver stage and pull its frequency.
Isolation varies with frequency. A PIN diode that looks like a clean open circuit at 100 MHz has residual capacitance that leaks more and more as frequency rises, so datasheets quote isolation across a band, not at a single point. Series and shunt diode arrangements are often combined in a cascade, because the leakages of successive sections multiply: two sections each giving 30 dB of isolation together give close to 60 dB, minus a small interaction term.
The driver circuit also plays a part. A PIN diode held off with a low reverse voltage stores charge that takes time to sweep out, and an under-driven diode never reaches its full off impedance. This is why the driver, its speed and its bias points, are treated as part of the switch rather than an afterthought.
What insertion loss can I expect from a solid-state PIN switch?
Insertion loss is the power lost in the selected path, again in decibels, measured from input to output with the switch in its on state. It comes from the residual resistance of the forward-biased diode, the residual capacitance of any reverse-biased diode in the path, and the unavoidable loss of the package and its transmission lines.
For a single series or shunt PIN diode at frequencies in the low gigahertz range, a through loss of a few tenths of a decibel per diode is typical. A practical multi-throw switch with driver and packaging usually lands between 0.5 and 2 dB across its band, with the loss rising toward the top of the specified frequency range. At millimetre-wave frequencies, package parasitics push the figure higher.
Two points are worth checking on any datasheet. The first is flatness: a switch that loses 0.8 dB at the bottom of the band and 1.8 dB at the top is harder to account for in a link budget than one with a flat 1.2 dB. The second is the relationship between loss and current: a heavier forward bias lowers the diode's resistance and its loss, at the cost of drive current and slower switching. Datasheets state the bias conditions for their figures, and comparing parts at different bias points is a common source of error.
Return loss, or VSWR, belongs in the same reading. High isolation and low insertion loss are of limited use if the switch reflects part of the wanted signal back into the source, so the match at each port should be checked across the band as well.
Switching time and the driver circuit
Switching time is how long the switch takes to move between states, usually quoted as the time from the control signal to 90 percent of the settled RF transition. PIN diodes are inherently fast, and much of the delay in a real part comes from the driver: the time to remove stored charge from the diode when turning it off, and the time to deliver enough current when turning it on.
The intrinsic layer of a PIN diode stores charge while it conducts. Turning the diode off means sweeping that charge out, and the speed of that sweep is set by the driver's ability to pull reverse current, not by the diode alone. This is why a well-designed driver with a fast, controlled transition can make the same diode switch in tens of nanoseconds, while a lazy driver leaves it in the hundreds.
For most receive and transmit chains, a switching time in the tens to hundreds of nanoseconds is comfortably faster than the surrounding signals. Applications such as pulsed radar, fast test sequencing or time-division multiplexing are the ones where the driver specification deserves close attention.
Where these figures are documented
The independent publication Control Line Review, referenced earlier, covers PIN switches alongside voltage-variable and digital-step attenuators and detector chains, and reads each topic the way a specification writer does: transfer curves, flatness, accuracy, tangential sensitivity and VNA calibration. For a switch selection, the useful habit it encourages is to read isolation, insertion loss, return loss and switching time as a set across the whole band, at the stated bias points, rather than as isolated headline numbers.
For the underlying device physics and general behaviour of PIN diodes, the reference material maintained by semiconductor manufacturers and by standards bodies such as the IEEE provides a sound background, and the datasheet of the specific part remains the final authority for the conditions under which its numbers were measured.
Choosing between the two topologies in practice
In a small system, the decision usually reduces to two questions. Does the stage in front of the switch tolerate reflections? If it does not, for example a fragile oscillator or a broadband amplifier without good output isolation, choose absorptive. Is the extra loss of the internal termination acceptable? If the link budget has no margin for it and the source is robust, reflective is the simpler part.
Isolation, insertion loss and switching time then act as filters among the parts that remain, with the driver circuit treated as part of the switch. Read all three across the full band, at the bias conditions you will actually run, and the choice between topologies becomes a matter of reading rather than guessing.