L
Admin
[Technical article, rfforum Editorial Team]
PIN diodes are still the workhorse for many high-power or high-linearity RF switches. The datasheet diode is only half the design; the bias network and the off-state parasitics set isolation and loss.
1. On state
A PIN diode with enough forward current looks like a small resistance (rs) in series with packaging inductance. Insertion loss is set by rs, the surrounding matching, and how well you RF-bypass the bias feed. More current lowers rs until you hit diminishing returns and thermal limits.
2. Off state
Reverse bias (or zero bias for some PIN structures) makes the diode a small capacitance Ct in series with package parasitics. Isolation is limited by that capacitance to ground or across the series path, plus any leakage through the bias network. At microwave frequencies a few tenths of a pF already set a hard isolation ceiling for a single series diode.
3. Series, shunt, and series-shunt
- Series: good when you need a simple path; isolation falls with frequency as ωCt rises.
- Shunt: low on-loss when the diode is off-path; isolation depends on how solidly the shunt diode shorts the line to RF ground.
- Series-shunt: the usual higher-isolation SPDT building block — series diode off and shunt diode on in the isolated arm. Isolation adds, at the cost of bias complexity and some loss.
4. Bias network mistakes that kill isolation
- RF choke / resistor that still looks like a low-Z path at your frequency → signal sneaks around the diode.
- Missing or distant bypass caps on the bias rail → bias line radiates and couples into the "isolated" port.
- Shared bias returns that create a sneak path between ports.
- Undersized reverse bias: Ct rises as reverse voltage drops; isolation collapses near band edges first.
5. Linearity and power
Off-state voltage swing and on-state current set compression and harmonics. A switch that looks fine at 0 dBm can generate spurs or compress under PA drive. Check the vendor's IP3 / P1dB at your bias points, not only at a demo bias.
Discussion: for a 1–6 GHz SPDT that must hold >40 dB isolation, do you stay with discrete PIN series-shunt, move to a packaged switch IC, or go mechanical/MEMS — and what isolation do you actually measure after the bias network is real?
— rfforum 编辑部 / Editorial Team
PIN diodes are still the workhorse for many high-power or high-linearity RF switches. The datasheet diode is only half the design; the bias network and the off-state parasitics set isolation and loss.
1. On state
A PIN diode with enough forward current looks like a small resistance (rs) in series with packaging inductance. Insertion loss is set by rs, the surrounding matching, and how well you RF-bypass the bias feed. More current lowers rs until you hit diminishing returns and thermal limits.
2. Off state
Reverse bias (or zero bias for some PIN structures) makes the diode a small capacitance Ct in series with package parasitics. Isolation is limited by that capacitance to ground or across the series path, plus any leakage through the bias network. At microwave frequencies a few tenths of a pF already set a hard isolation ceiling for a single series diode.
3. Series, shunt, and series-shunt
- Series: good when you need a simple path; isolation falls with frequency as ωCt rises.
- Shunt: low on-loss when the diode is off-path; isolation depends on how solidly the shunt diode shorts the line to RF ground.
- Series-shunt: the usual higher-isolation SPDT building block — series diode off and shunt diode on in the isolated arm. Isolation adds, at the cost of bias complexity and some loss.
4. Bias network mistakes that kill isolation
- RF choke / resistor that still looks like a low-Z path at your frequency → signal sneaks around the diode.
- Missing or distant bypass caps on the bias rail → bias line radiates and couples into the "isolated" port.
- Shared bias returns that create a sneak path between ports.
- Undersized reverse bias: Ct rises as reverse voltage drops; isolation collapses near band edges first.
5. Linearity and power
Off-state voltage swing and on-state current set compression and harmonics. A switch that looks fine at 0 dBm can generate spurs or compress under PA drive. Check the vendor's IP3 / P1dB at your bias points, not only at a demo bias.
Discussion: for a 1–6 GHz SPDT that must hold >40 dB isolation, do you stay with discrete PIN series-shunt, move to a packaged switch IC, or go mechanical/MEMS — and what isolation do you actually measure after the bias network is real?
— rfforum 编辑部 / Editorial Team
