Zero calibration is carried out automatically every time the instrument is switched on, or, as is possible in some madur instruments, every time the auto-calibration of the oxygen sensor is performed (usually used for continuous measurements over a couple of hours - in connection with a gas cooler/drier ).
Span calibration, for toxic sensors, is recommended to be performed at least every six months. Basically, it is dependent on the expectations in regard to accuracy one has and the amount of use the analyser sees. Some companies perform a span calibration every other week. Therefore they are able to achieve accuracies around +/- 2%. An auto-calibration for the oxygen sensors to 20.9% O2 in air is performed every time the instrument is switched on. Or, as is possible in some madur instruments, at pre-set intervals to ensure good readings during continuous measurement sessions.
Calibration is an important factor in the accuracy of an analyser reading. The instrument refers to the calibration results as the standard value at every measurement and has no other way to compensate for changes in the characteristics of the sensors. These do remain reasonably stable over fairly long periods, but a certain drift and tendency to a reduced signal with time is unavoidable. The quality of the calibration gas is also important, as is its homogeneity.
Calibration should be carried out at a set temperature and pressure. The signal from the sensors is defined at 20°C, so a calibration at or near this temperature is to be recommended. Better instruments, including those from madur electronics, have a temperature sensor near the electrochemical gas sensors to allow these effects to be compensated electronically.
Electochemical sensors have a fairly wide range of tolerance on the output at a set concentration, so they cannot be used without calibration. These tolerances do not have anything to do with the quality of the sensors, they are rather a result of the complicated chemical processes involved. Having a rough idea of the calibration value to be expected will allow an experienced service technician to immediately see when a sensor is nearing the end of its useful operating life.
If an analyser is used for extended periods of time it will probably prove necessary to repeat the zero calibration at regular intervals to neutralise any changes that occur with time. Particularly the zero point will tend to drift to a certain degree with time. This can not always be compensated electronically since certain factors are random and a drift that produces a negative result will be automically treated as zero, since negative concentrations are obviously impossible. For this reason, stationary equipment will generally have a programme feature to do this automatically. This will show up as a sudden zero on the results, but such factors can easily be removed later by software.