Checking Feedback Continuity During Multiplexer Break Before Make

A programmable gain amplifier can meet its steady gain specification and still produce a large excursion whenever the gain changes. If a multiplexer selects the only feedback connection, its break-before-make interval temporarily removes negative feedback. The amplifier then responds to offset, bias current, and the signal difference with its open-loop behavior. Recovery may take longer than the switch transition itself.

Draw the intermediate circuit

Draw three schematics: old channel connected, every channel open, and new channel connected. Include feedback capacitors, parasitic capacitance, input source impedance, and any permanent resistor. A capacitor may preserve feedback for a fast transition but cannot establish a DC operating point indefinitely. Analog Devices distinguishes break-before-make from overlapping connections in its switch and multiplexer selection guide. Neither switching style is automatically suitable for every feedback network.

For example, consider changing an inverting stage between two feedback resistors while its input remains at a nonzero voltage. During an open interval, current through the input resistor charges the summing-node capacitance. Even if the output never reaches a rail, the next channel begins with a displaced node voltage. Repeating the test with zero input can understate the disturbance encountered in normal use.

Choose a bounded transition

A permanent feedback resistor can maintain a defined loop during switching, but it also changes every programmed gain and the noise gain. Calculate the parallel combination in each state. An overlapping switch can avoid an open loop yet momentarily combine two resistors or connect incompatible sources. Disabling the downstream measurement during the transition may be sufficient when the output is allowed to move and recovery is bounded.

• Capture the control edge, amplifier output, and downstream sampling event together.
• Exercise both directions between the largest and smallest gains.
• Repeat at both input polarities and near the allowed output limits.
• Measure time to the actual error band, including any overload recovery.

Make the acquisition delay evidence based

Use the slowest measured valid recovery together with component and operating-condition margins to define when data may be accepted. A fixed delay based only on the multiplexer data-sheet switching time omits amplifier behavior. Distinguish a brief charge-injection step from loss of feedback by changing input amplitude and by testing an alternative permanent feedback path.

Relate the result to switch charge injection and amplifier overload recovery. Record the intermediate circuit with the gain table so a replacement switch is reviewed for transition behavior as well as resistance and voltage range.