Distinguish zero-scale offset and full-scale gain correction
Offset and gain error describe different departures from a converter's intended transfer relationship. ADI discusses zero-scale offset and full-scale gain calibration separately. Correcting the zero-related contribution does not describe the same adjustment as correcting the scale of the conversion. A requirement for calibration should identify which error contribution it expects the converter to address.
The note also describes separate coefficient roles in producing a corrected conversion result. The offset-related correction and gain-related correction participate differently in that process. Their existence is more specific than a general statement that the ADC has been calibrated. It explains why one correction cannot simply be treated as proof that the other part of the transfer relationship has also been addressed.
This distinction remains separate from output coding. Coding identifies how a digital word represents the input range; calibration concerns correction factors applied to conversion results. Changing a numerical interpretation is not the same operation as generating a correction from a known input. The useful requirement keeps the error contribution and coefficient role explicit before asking where the known calibration condition is established or what external circuitry it includes.
Identify the internal known-input boundary
Internal calibration establishes a known condition within the converter's own documented path. ADI's internal zero-scale discussion describes an internally established zero-input relationship for the selected analog inputs. That lets the converter address its internal offset context without treating the external signal path as though it supplied a known system input.
For internal full-scale calibration, ADI’s AD7124-4 and AD7124-8 discussion describes an internal source related to the selected gain and reference. The same paragraph qualifies the allowed power-mode context and explains when the resulting coefficient remains applicable after a mode change. Those conditions belong to this named capability statement; they should not disappear when the feature is shortened to internal gain calibration.
Keep the location of the known input separate from the error type. Internal zero and internal full-scale describe different known conditions, while both retain the boundary of the documented converter path. This is not a claim that every ADC provides the same collection of modes. The explanation here remains within the saved AD7124 paragraph for internal full-scale and makes no AD717x comparison or universal capability promise.
Locate the system-calibration boundary
System calibration establishes the known input through an external path. ADI explains that system zero-scale calibration can include the offset contribution of external circuitry together with the ADC. The location at which the known condition enters therefore determines what part of the signal chain participates. A calibration described only as internal does not automatically include those additional contributions.
The system full-scale discussion makes a corresponding distinction for gain. Applying the known full-scale condition through the module input can include the front-end gain relationship alongside the converter's own contribution. This is a boundary explanation, not a statement that every sensor imperfection becomes corrected. The signal path traversed by the known input remains the subject of the calibration claim.
This gives the application a precise question to answer: which part of the measurement path must the correction cover? A converter-level known input and a system-level known input can serve different requirements even when they produce coefficients used by the same ADC. Preserve that coverage difference without copying a connector short, supplying an applied voltage or turning a named evaluation-board example into a wiring procedure for another system.
Retain coefficient applicability to channel and configuration
A coefficient belongs to the channel and configuration for which it is applicable. AN-1464 discusses the selected channel, gain and reference in its calibration context. These are not incidental labels around an otherwise universal correction. Changing the conditions that defined the relationship can create a new applicability question, even when the converter's product identity has not changed.
The AD7124 internal-full-scale discussion illustrates that specificity. It retains the same gain and reference when describing continued use of a coefficient across its stated power-mode change. That is a documented relationship, not permission to transfer any coefficient between arbitrary modes. The broader discussion likewise makes channel selection relevant rather than presenting one coefficient set as automatically covering every input path.
Finish the requirement by identifying the error contribution, known-input location, covered signal path and configuration to which the correction applies. These layers explain the calibration method without prescribing when to run it or which registers to write. They also keep calibration distinct from filter latency and acquisition architecture. A corrected conversion relationship does not independently establish a settled result, sensor linearity or verified accuracy for a complete measurement system.
Sources and further reading
ADI AN-1464 bounded saved HTML intervals on offset/gain, internal zero, AD7124-4/-8 internal full-scale, system calibration and coefficient applicability. No AD717x capability comparison, calibration voltage, wiring, register sequence, schedule or achieved accuracy is provided.