Distinguish converter subtraction from the selectable input pair
The ADS111x converter core measures the difference between its positive and negative analog inputs. That subtraction defines the measured quantity. The ADS1115 additionally has an input multiplexer that establishes which terminals supply those two roles. The converter's differential operation and the MUX's available selections are therefore connected but different descriptions: one explains the measurement, while the other identifies its input pair.
In a single-ended ADS1115 selection, the multiplexer supplies ground internally as the converter's negative reference. The positive terminal is then measured in that ground-referenced coordinate. TI separately describes use of AIN3 as a common point for supported measurements. That arrangement changes the comparison reference from ground to the selected common terminal. It does not turn the MUX into an arbitrary pairing of all available pins.
The selected reference determines what a positive or negative difference means. A voltage can be below the common point while remaining above system ground. The device's input count alone leaves this relationship unspecified. This coordinate question is also distinct from channel-observation timing: knowing which pair is selected explains what is measured, whereas simultaneous or successive observation describes when different measurements occur.
Keep signed differences inside absolute terminal conditions
The sign of a differential result describes the ordering of two input voltages. If the positive-input voltage is lower than the negative-input voltage, the measured difference is negative even when both terminals remain positive relative to ground. A signed measurement consequently does not require a negative terminal voltage. The measurement reference and the terminals' absolute reference to ground must remain separately identifiable.
TI's single-ended-input discussion states that the named devices accept positive analog voltages relative to ground. The input-protection discussion separately identifies diodes to the supply rails and the conditions that avoid turning them on. Differential arithmetic does not change those physical input structures. A terminal outside its permitted conditions is therefore a different matter from a valid negative difference between two terminals inside their operating range.
The MUX adds a further consequence to an overdriven input. TI states that overdrive on one ADS1115 input can affect conversions taking place on other inputs. Selecting another channel does not make the physical condition at the first terminal irrelevant. This links the selected measurement to the surrounding input environment without converting the protection description into an external clamp design or assuming that a plausible result proves every terminal is operating correctly.
Separate PGA scaling from available analog span
The PGA full-scale setting defines the scaling used by the converter. Its mathematical measurement span and the voltages physically permitted at the analog terminals have separate meanings. TI explicitly retains the analog-input limits when discussing full-scale range. Selecting a larger span therefore changes the measurement scale without raising the supply-dependent allowance at a terminal.
The supply can leave part of the configured span unavailable. When the permitted terminal voltages cannot reach the entire scale, some of the mathematical range has no corresponding usable input condition. This is an interaction between the configured measurement and the physical voltage environment. The full-scale name alone does not establish that every point in that span can be applied in the actual supply context.
Reference selection also affects which parts of the span are useful. A ground-referenced single-ended measurement and a measurement relative to a positive common point allow different relationships between positive and negative differences. TI qualifies the AIN3 common-point arrangement separately, including its lack of the stated common-mode noise attenuation. This prevents the extra differential range from being mistaken for a universal common-mode-rejection result. The same PGA setting can accompany different usable measurement coordinates.
Compare common-mode and differential input loading
The ADS111x analog front end repeatedly charges and discharges internal capacitors during measurement. The source supplies charge during that activity. TI explains that the average loading appears resistive to the external circuit, even though the underlying mechanism is switched-capacitor operation. A static-looking impedance description therefore summarizes an ongoing charge demand rather than describing an input that draws no current.
The datasheet distinguishes common-mode and differential effective impedances. Its common-mode description concerns the average current drawn from inputs moved together; its differential description concerns the current associated with a difference between them. The full-scale setting enters these loading relationships. A value or label for one excitation condition should retain that condition rather than being assigned indiscriminately to both kinds of source interaction.
This matters when the driving source has appreciable output impedance. The converter's loading can affect the measurement, while an active buffer introduces its own noise, offset and gain contributions. The interface question therefore includes both charge delivery and any additional conditioning stage. It remains separate from the acquisition-window budget of a different SAR converter: the named delta-sigma input discussion concerns its continuously switched loading and mode-dependent source relationship.
Sources and further reading
ADS111x SBAS444E qualitative prose on physical pages 12–15 and 29; page 23 coding was read but is not adopted. No register bits, code table, numerical range, protection recipe or measured accuracy is supplied.