Separating Analog Switch Off Isolation From Channel Crosstalk

An analog switch can block DC conduction while still coupling an AC signal through parasitic capacitance. Another channel can also couple into the selected path. Off isolation and crosstalk describe different unwanted paths, so choose the correct measurement before deciding whether a switch meets a multi-channel signal requirement.

Draw the aggressor and victim paths

The TI guide to selecting signal switches distinguishes off-state signal coupling from coupling between channels. The specified values depend on frequency and the test arrangement.

For off isolation, identify the input and output of the disabled path. For channel crosstalk, identify the driven channel and the separate observed channel. Keep control-to-output transients as another category rather than mixing them into a steady sinusoidal isolation result.

Match impedance and operating state

Record source impedance, receiving load, switch supply, signal bias, and every channel’s state. A value measured with a low impedance load can differ substantially from the result at a high impedance amplifier input.

Include the inactive channels’ real terminations. Leaving them floating during a bench test can produce a different network from the assembled system, where sensors or cables remain attached.

Measure across the required spectrum

Apply a known aggressor and measure the unwanted signal at the defined victim node. Express the ratio with a clear sign convention. For example, a hypothetical 1 V aggressor producing 1 mV at the victim corresponds to a voltage ratio of minus 60 dB.

Check the instrument and fixture floor using an appropriate reference arrangement. Direct coupling between cables or probes can become larger than the switch’s contribution and limit the measured rejection.

Connect the result to functional error

Determine whether the coupled signal remains out of band or can be sampled, rectified, or mixed into the desired measurement. A small high frequency voltage can matter more in one receiving circuit than another.

Test relevant combinations of selected and unselected channels, including the largest realistic aggressor amplitude. Keep the switch inside its supported signal voltage range so an overvoltage path is not mistaken for ordinary capacitive coupling.

Save path definitions, impedances, bias conditions, channel states, frequency sweep, and measurement floor. A satisfactory selection demonstrates acceptable unwanted signal at the actual receiving circuit. A favorable off-isolation number at one frequency does not independently establish low crosstalk between every channel, nor does a low DC leakage specification describe the same AC behavior. Include the inactive channel connections in the saved measurement schematic for reproducibility.