1. Identify a derivative-output sensor before selecting gain
A Rogowski coil does not supply the same kind of signal as a direct resistive current measurement. TI describes its output voltage as proportional to the rate of change of primary current. The sensing element and the current waveform therefore have a derivative relationship. Recovering a waveform representing the original current requires an integration function in the signal chain.
This distinction also explains the static-current boundary. A constant DC component does not produce the changing magnetic condition on which the coil's voltage output depends. Amplifying the raw output cannot turn it into a static-DC current measurement. Before comparing sensitivity or gain, establish whether the required information is the changing-current signal or a recovered waveform with its intended timing relationship.
TI discusses both software and hardware integration, which place that recovery function in different parts of the system. A hardware integrator can use passive elements or an amplifier-based implementation. The architecture decision starts with where integration occurs and what signal the next stage expects. It should precede treating the coil and a generally labelled amplifier as a complete current-measurement interface.
2. Separate passive attenuation from active integration
A passive integrating network changes amplitude as well as phase response. In TI's explanation, the voltage available after the passive network is attenuated. That matters especially when the coil produces a weak signal at low current: an integration function may be present while leaving an inconveniently small signal for the following conversion stage.
The active example places the integrating network in an amplifier feedback path and adds a means of scaling the result. Integration and signal gain remain distinguishable functions even when one stage participates in both. Asking only for a larger gain can conceal whether the waveform recovery and low-current signal level have been considered together.
Compare architectures by following the signal through those functions. The coil supplies the derivative-related voltage; the integration path changes its transfer relationship; the scaling stage places the result within a useful downstream range. Passive attenuation belongs to that chain, rather than being an unrelated catalogue loss. This interpretation explains why weak-signal handling can change the preferred architecture without supplying a gain setting or a resistor-capacitor recipe.
3. Relate phase accuracy to response time
The integration response also carries a timing tradeoff. TI distinguishes a precision-measurement path from a fast-settling path, rather than presenting a single response as optimal for both. A measurement that depends on the relationship between voltage and current waveforms raises a phase-error question, while a rapidly responding path also has a settling requirement. A power-measurement path needs the current waveform's relationship to voltage, whereas a fast response requirement concerns when usable information becomes available. Those purposes place different emphasis on the same integration response.
The note associates a larger integration time-constant basis with reduced phase error and a different response time. Improving one coordinate therefore cannot be assumed to improve the other. A comparison needs to say which behaviour the application prioritizes instead of treating an accurate integrator label as proof of fast availability after startup.
Keep phase-error and settling descriptions alongside their respective measurement purposes. The useful decision is whether the documented response addresses those purposes together or whether different paths serve them. This does not require adopting the reference design's component values or transferring its relay and breaker results. A phase-error description must remain tied to the signal-chain relationship being measured; it does not independently establish a universal polarity convention for every coil and connection.
4. Match input loading and downstream signal range
The coil has limited output-drive capability, so the input presented by the integrator matters. TI describes an inverting stage whose input resistance depends on its gain requirements. If that input loads the coil excessively, the coil's available voltage is affected before the desired integration and amplification have produced their intended result.
The named design introduces a high-input-impedance noninverting stage to address that loading issue. This identifies a role, not a universal instruction to add the same circuit to every coil. Input loading and overall gain must remain separate comparison fields because a suitable final amplitude does not establish that the sensor was acceptably loaded along the way.
The downstream amplifier's available output swing adds a further boundary. Gain allocation has to retain the input-signal range and the loading imposed by the next stage. A design can therefore have an integration requirement, a sensor-loading requirement and an output-range requirement simultaneously. Keeping all three visible gives the coil-to-integrator comparison a concrete electrical purpose without promising accuracy or prescribing a particular implementation.
Sources and further reading
TI TIDUBY4A, revised September 2016, physical pages 3–9: derivative recovery, integration, response tradeoffs and loading prose. No universal phase-sign conclusion, numerical settings, circuit recipe or reference-design performance is transferred.