1. Keep the control endpoint separate from continued tuning

The control-voltage interval identifies where the documented frequency response is described, rather than an invitation to continue the same response indefinitely. AN10020 describes upper and lower control limits beyond which the frequency-voltage characteristic saturates. Moving the control voltage farther beyond those endpoints does not produce a comparable continuation of frequency change. The endpoint therefore belongs to the response description itself.

A slope taken from within the interval cannot simply be extended through that saturation region. The local relationship between a small voltage change and frequency change describes a different condition from an endpoint where further voltage produces little noticeable frequency movement. Treating one average sensitivity as a straight line without boundaries would hide this change in the behavior the note actually describes.

This is separate from deciding whether an input voltage is electrically permissible. A tuning-response boundary is not an absolute-maximum rating, and a saturated response is not evidence that an out-of-range input is harmless. Keep the response endpoints with the exact device documentation. That preserves the technical meaning without recommending a control voltage, decoding a stored part or turning general pull-range terminology into an operating approval.

2. Distinguish whole-curve linearity from local sensitivity

The shape of a complete frequency-voltage curve and its sensitivity at a particular point are different descriptions. AN10020 defines linearity through departure from an ideal straight relationship over the curve. It separately describes Kv as the local slope. One quantity concerns an overall departure; the other concerns how strongly a small control change affects frequency in the region being considered.

A curved response can therefore contain different local sensitivities even when an average slope appears convenient. The note discusses varactor-based tuning in that context, while AN10021 cautions that a favorable overall linearity specification need not exclude significant Kv variation. A whole-curve summary does not uniquely establish the local response at every control condition. The two descriptions need to remain identifiable rather than substitute for one another.

This distinction is useful when comparing documented control behavior: the same broad pull capability can leave different questions about the distribution of sensitivity across the interval. It does not require reconstructing a plot or ranking oscillator technologies. The notes’ legacy device comparisons are not transferred into a current superiority claim. The contribution here is the difference between a global curve descriptor and the local gain relevant to subsequent control behavior.

3. Follow sensitivity into closed-loop response and noise

AN10021 carries local sensitivity into a different context by representing the VCXO as a gain within its described low-bandwidth PLL. A gain inside a feedback system affects the response of the complete loop. The note connects greater Kv with faster tracking, while also retaining overshoot, peaking and proximity to instability as possible consequences. Tracking speed is therefore not the only part of that relationship.

The same gain also relates control-input voltage noise to frequency variation. Increasing sensitivity can make the loop respond more strongly to the intended control signal and to unwanted voltage at that input. This explains why a sensitivity figure should not be read solely as a convenient adjustment capability. It participates in both the control dynamics and the way control-path noise enters the timing account.

Those are architecture-bound tradeoffs, not a prediction for an unidentified PLL. A component’s Kv heading does not establish the surrounding loop, its stability margin or achieved jitter. The important distinction is between the oscillator’s local conversion gain and the closed-loop consequences of using that gain. Keeping the dependency explicit avoids treating faster tracking as an unconditional improvement or presenting the named note as a completed system evaluation.

4. Separate dynamic control bandwidth from a DC sweep

A DC sweep describes how frequency relates to slowly established control voltages. Dynamic control bandwidth instead concerns how output frequency follows a control voltage that changes over time. AN10020 describes a low-pass characteristic in that relationship. A static curve can therefore characterize the available response without describing how faithfully the oscillator follows control changes at every rate within a working system.

This oscillator control-path behavior also differs from the complete PLL response discussed above. The oscillator has its own conversion from changing control voltage to frequency; the closed loop combines that behavior with the surrounding control system. Calling both descriptions bandwidth without retaining their subject can confuse a component modulation characteristic with the dynamics of an entire feedback loop. They are connected, but their boundaries are different.

Preserving the static and dynamic descriptions makes the control record more precise. The DC interval and curve shape identify one set of relationships; modulation bandwidth identifies the rate-dependent response of the control path. Neither alone supplies a settling time, a selected loop design or a present operating state. This account uses the notes qualitatively, without their worked modulation arithmetic, numerical thresholds or a measured tracking result for stored timing devices.

Sources and further reading

Bounded AN10020 and AN10021 HTML prose only, including their legacy 380X context. No PDF reading, publication date, plotted curve, numerical tuning/linearity/Kv/bandwidth claim, loop design, present calibration or quartz/MEMS ranking is asserted.