1. Explain charge transfer before naming the voltage ratio
A charge pump transfers energy by changing how capacitors are connected over successive switching phases. TI describes the LM2662/LM2663 inverter using internal switches and external capacitors for charge transfer and storage. One interval establishes charge on the flying capacitor; another transfers it to the output-storage role. The sequence, rather than the general product-category name, gives the conversion its meaning.
The negative output in that explanation depends on the particular connection relationship. TI describes an ideal no-load result under assumptions that remove switch loss and capacitor ESR. Those assumptions are part of the explanation, not a declaration that every practical output is an exact regulated copy with reversed sign. Charge transfer and output regulation are different properties.
Start the component requirement with the intended conversion function and the evidence for that mode. This separates the underlying switched-capacitor architecture from the desired rail. A device described as a charge pump may offer relevant functions, but the category alone does not establish which output polarity, connection arrangement or regulation behaviour belongs to the application under review.
2. Separate inversion from positive doubling
The same named devices also have a positive-doubler application. TI explains that this function changes some application connections and gives it its own startup discussion. It is therefore a different use of the charge-transfer hardware, rather than a second name for the negative-output arrangement or a promise that mode selection is identical across all charge pumps. The positive-doubler startup discussion specifically concerns establishing the oscillator's supply conditions. That requirement belongs to the selected operating arrangement, even though the steady-state output is described through a multiplication function.
Inversion and positive multiplication answer different rail requirements. An application seeking a negative supply needs the sign relationship made explicit. An application seeking a larger positive rail needs the applicable multiplication mode and its conditions instead. Comparing a detached magnitude could hide that distinction even where the devices belong to the same converter family.
Startup also belongs to the selected mode. The doubler prose describes how the oscillator obtains the conditions needed to begin operation, showing why the steady unloaded ratio does not explain the complete operating transition. Keep that architectural difference visible without copying its external circuit. This preserves the distinction between choosing a conversion function and specifying the connections or startup components of an actual design.
3. Read load droop through finite output impedance
The unloaded ratio is not the whole loaded-output description. TI represents the practical inverter with an ideal-source contribution and a finite output resistance. Internal switch resistance, switching frequency, the flying and output capacitors, and their ESR all participate in that output-impedance basis. Loading consequently introduces a separate question from identifying the ideal conversion polarity.
This model explains why a capacitance headline alone cannot complete a comparison. Increasing capacitance can reduce one contribution while switch resistance and capacitor ESR remain. The note explicitly describes a point beyond which further increases in capacitance or frequency become ineffective at reducing the remaining dominant terms. The controlling mechanism can change without the conversion mode changing.
Read output-impedance information together with its operating context when defining a candidate requirement. This connects the expected loading to the conversion mechanism instead of treating the ideal unloaded ratio as a regulated loaded value. It also distinguishes capacitor charge-transfer limitations from resistive contributions. No component value, output-drop calculation or exact load capability is derived from the simplified explanation here.
4. Keep clocking policy and ripple/current tradeoffs mode-specific
Clocking policy is another mode-dependent selection coordinate. TI states that a higher oscillator frequency can permit smaller capacitors for an equivalent output-resistance and ripple basis while increasing supply current. The comparison therefore includes an interaction among frequency, capacitance, ripple and operating current, rather than an assumption that increasing frequency improves every property at once.
The inverter and positive-doubler modes do not have identical clock options. The named datasheet permits an external-clock approach in inverter mode but explicitly excludes driving the oscillator externally in voltage-doubling mode. A general synchronizable-converter requirement must therefore be checked against the actual mode, rather than copied from another application of the same device.
Preserve the selected conversion function when comparing clock and capacitor information. Otherwise, a clock feature from one mode can be combined with an output description from another, creating a combination the evidence does not support. The useful decision is whether the documented operating policy addresses the required rail and signal environment together. It does not supply an oscillator setting, capacitor choice or a guaranteed ripple/current result for the customer's circuit.
Sources and further reading
TI LM2662/LM2663 SNVS002E, physical pages 10, 12, 16–17: named-mode charge-transfer, output-impedance and clocking prose. Divider/cascade design, pin connections, component recipes and numerical performance are not supplied.