Measurements, Testing, and Calibration Network Analysis Informational

How do I measure the isolation between ports of a switch or multiplexer accurately?

Isolation is the attenuation between two ports of a switch or multiplexer that should be disconnected. It is measured as the transmission coefficient between the off-path ports: Isolation (dB) = -20×log10|S21_off| = |S21_off| expressed as a positive number. For example: if S21_off = -60 dB, the isolation is 60 dB. Accurate isolation measurement requires: (1) Sufficient VNA dynamic range: the VNA must be able to measure signals 60-100 dB below the reference level. VNA dynamic range = maximum output + sensitivity = typically 100-130 dB. For 80 dB isolation measurement: need at least 90 dB dynamic range (10 dB margin). To maximize VNA dynamic range: reduce IF bandwidth to 10-100 Hz (reduces noise floor by 20-30 dB compared to 10 kHz IFBW), increase source power (0 to +10 dBm if the switch can handle it), and average multiple sweeps (10 averages improves noise floor by 10 dB). (2) Proper calibration: full two-port SOLT calibration with cables and adapters in place. The corrected directivity and crosstalk of the calibration limit the minimum measurable isolation. After calibration: if the residual crosstalk is -90 dB, measurements of isolation > 90 dB are unreliable. (3) External leakage elimination: at high isolation levels (> 60 dB), signals can leak around the DUT through: external coupling between input and output cables (ensure cables are physically separated and do not run parallel), shared ground paths (use the DUT ground return, not a common bench surface), and electromagnetic coupling through the DUT enclosure (if the switch housing has poor shielding, RF leaks around the switching element, limiting the measurable isolation even if the internal switch element is perfect). (4) Proper termination: all unused ports must be terminated in 50 ohms. An unterminated port creates reflections that can couple through the multiplexer and reduce the apparent isolation.

Category: Measurements, Testing, and Calibration

Updated: April 2026

Product Tie-In: VNAs, Calibration Kits, Cables

Switch Isolation Measurement

Isolation measurement is one of the most demanding VNA measurements because it requires detecting very weak signals (the leakage through the off-state path) in the presence of the VNA noise floor and external coupling.

Parameter	SOLT Cal	TRL Cal	eCal
Accuracy	Good	Excellent	Good-very good
Standards Needed	4 (S,O,L,T)	3 (T,R,L)	1 (module)
Bandwidth	Broadband	Band-limited	Broadband
Setup Time	5-10 min	10-20 min	1-2 min
Best For	Coaxial, general	On-wafer, waveguide	Production, speed

Calibration Procedure

(1) Equipment: VNA with sufficient dynamic range (100+ dB). Low-loss, well-shielded cables. Torque wrench for connectors. 50-ohm terminations for all unused ports. (2) Calibration: perform full two-port SOLT calibration at the switch reference planes. For highest accuracy: use the DUT connectors (not adapters) during calibration. After calibration: verify that the S21 noise floor (with ports disconnected) is at least 10 dB below the expected isolation. If the noise floor is -90 dB and the switch isolation is specified at 80 dB: adequate. If the switch is specified at 90 dB: the noise floor must be reduced. (3) Bias and control: apply the correct bias voltages and control signals to set the switch to the off-state. For PIN diode switches: ensure the correct reverse bias is applied (typically -5V to -12V). Insufficient reverse bias degrades isolation. For FET switches: ensure the gate voltage is in the pinch-off region and all FETs in the off-path are properly biased. (4) Measurement: measure S21 between the input and the isolated output port. Set VNA IFBW to 100 Hz or lower for isolation > 60 dB. Use 16-32 averages. Record isolation across the full operating frequency range (isolation often degrades at higher frequencies due to parasitic coupling).

Performance verification: confirm specifications against the application requirements before finalizing the design
Environmental factors: temperature range, humidity, and vibration affect long-term reliability and parameter drift
Cost vs. performance: evaluate whether the application demands premium components or standard commercial grades
Interface compatibility: verify impedance, connector type, and mechanical form factor match the system architecture
Margin allocation: include sufficient design margin to account for manufacturing tolerances and aging effects

Error Sources

Switch isolation typically decreases with increasing frequency because: (1) Parasitic coupling: the off-state capacitance (C_off) of the switching element (PIN diode, FET, MEMS) creates a parasitic path. The impedance of C_off decreases with frequency: Z_C = 1/(2pi×f×C_off). For a FET switch with C_off = 0.1 pF: at 1 GHz: Z_C = 1592 ohms → isolation ≈ 20×log10(1592/50) = 30 dB. At 10 GHz: Z_C = 159 ohms → isolation ≈ 10 dB. At 40 GHz: Z_C = 40 ohms → isolation ≈ -2 dB (the switch is essentially transparent). (2) Package coupling: electromagnetic coupling between the input and output package leads increases with frequency. Even if the switching element has high isolation: the package can limit the achievable isolation to 50-70 dB at high frequencies. Solution: use shielded packages, minimize lead length, and use series-shunt-series switch topologies for higher isolation.

Common Questions

Frequently Asked Questions

What isolation do I need for my application?

Receiver protection (T/R switch): > 30-50 dB (to prevent transmitter leakage from damaging the receiver LNA). Test & measurement (switch matrix): > 60-90 dB (to prevent signal leakage between measurement channels from corrupting results). FMCW radar (TX/RX isolation): > 40-60 dB (to minimize transmitter leakage into the receiver chain, which limits maximum sensitivity). Antenna switching (diversity, MIMO): > 20-30 dB (sufficient to prevent significant coupling between antenna paths). Filter bank switching: > 60-80 dB (to prevent out-of-band signals from leaking through the off-path filter).

How does isolation differ from insertion loss?

Insertion loss is the attenuation through the ON-state path (desired to be as low as possible, typically 0.5-3 dB). Isolation is the attenuation through the OFF-state path (desired to be as high as possible, typically 20-80+ dB). They are both |S21| measurements but in different switch states. Common switch specifications include both: for example, a SP4T switch might specify 1.5 dB IL, 60 dB isolation at 6 GHz. The ratio IL/Isolation defines the effective on/off ratio of the switch (58.5 dB in this example).

Why does isolation vary between different off-state ports?

In a multi-throw switch (SP3T, SP4T, etc.): the isolation between the common port and each off-state throw depends on the physical layout and coupling paths. Throws that are physically adjacent to the active (on-state) throw often have worse isolation because: (1) The on-state signal can couple through the shared common junction to adjacent throws. (2) Electromagnetic coupling between adjacent throw arms is stronger than between distant arms. (3) The bias/control lines for adjacent throws may have mutual coupling. Always specify and measure isolation for each throw individually, and also measure the throw-to-throw isolation (coupling between two off-state throws), which can be different from the common-to-throw isolation.

Keep Reading