Include ADC Protection Components in the Accuracy Budget

An input protection network can keep a fault away from an ADC while degrading normal measurements. Series resistance, clamp leakage and added capacitance belong in the signal-chain model. Reviewing only a protection device’s headline voltage rating leaves both accuracy and actual fault behavior unresolved.

Define normal and fault conditions separately

List the valid input range, source impedance, sampling schedule and temperature range. Then define the electrical fault conditions the design is intended to withstand, including duration and source capability. A short transient and a sustained misconnection are different cases and may require different protection methods.

Texas Instruments’ ADS124S08 protection application report discusses protection component choices and their measurement effects in a specific RTD interface. Use such examples to identify mechanisms, then verify the limits and behavior of the actual ADC and protection components selected for the product.

Calculate normal leakage errors

Trace where each leakage current flows and which resistance converts it into an input error. As a simple illustrative calculation, 10 nanoamps through 10 kilohms produces 100 microvolts. Compare that with the system error allowance before deciding that a leakage specification is negligible.

Use relevant voltage and temperature conditions from component documentation. Typical room-temperature leakage may be inadequate for a worst-case assessment. In differential measurements, assess mismatched paths as well as common behavior; nominal symmetry does not guarantee equal leakage or resistance.

Check dynamic behavior

Include the protection resistance and capacitance in acquisition and filter settling analysis. A larger resistor can limit current while slowing the input response. If channels are multiplexed, test transitions between representative extremes rather than only steady measurements.

Also define how long a measurement may be invalid after an allowed overload. A clamp can stop conducting before the rest of the signal chain has recovered. Software should not immediately interpret every returning ADC code as a valid process value without evidence supporting that timing.

Verify the complete protection path

Identify where diverted fault current goes and whether that destination can accept it. A supply rail used as a clamp destination may rise if its regulator cannot sink the current. Include the unpowered state because a connected external source can remain active while the board is off.

Perform controlled validation using equipment and limits appropriate to the defined faults. Record both survival criteria and post-event accuracy. A board that still communicates may have suffered a measurement shift. Keep the fault specification, current-path analysis and normal error budget linked so a later diode or resistor substitution receives both kinds of review.