The difference lies in how the reference value is arrived at. In comparison calibration, the standard and the instrument measure the same quantity together, for example the same pressure in a manifold, and their readings are compared. In direct calibration, a source or material standard delivers a known value, such as a standard mass on a balance or a signal from a calibrator, and you observe what the instrument indicates.
Comparison in practice
To calibrate a pressure gauge, the gauge and a pressure standard are connected to the same circuit, pressure is generated with a pump and both are read. The pressure does not need to be a round value: what matters is the difference between the two readings at each point. The same is true in a temperature bath where a standard thermometer and the instrument share the medium.
Direct calibration in practice
To calibrate a balance, standard masses of known value are placed on it and the indication is read. To calibrate a transmitter with a current calibrator, a value is applied from the source and compared with the output. Here the applied value is known, and the standard is the source or artifact.
What to watch in each case
- Comparison: both sensors must be exposed to the same conditions, with no gradients between them; there must be no differences in height or position that alter the quantity; and the quantity must be stable.
- Direct: the generated value must really be the nominal value, within the uncertainty of the source; and the instrument must not load or alter the source.
When it does not apply
Comparison requires that the standard and the instrument can measure the same thing at the same time, which is not always possible. Instruments that measure the dimensions of artifacts, for example, are calibrated directly with material standards, such as gauge blocks or ring gauges.
How to interpret the result
Both approaches seek the same thing: the instrument error with its uncertainty. The choice depends on the quantity, on what is practical and on which offers the lower uncertainty. A good procedure states which one was used.