Advanced Instruments Inc.
Accuracy & Calibration
Single Point Calibration: As
previously described the galvanic
oxygen
sensor generates an electrical current
proportional to the oxygen
concentration in the sample gas.
Absolute Zero: In the absence of
oxygen
the sensor exhibits an absolute zero,
e.g. the
sensor does not generate a current
output in
the absence of oxygen. Given these
linearity
and absolute zero properties, single
point
calibration is possible.
Pressure: Because sensors are
sensitive
to the partial pressure of oxygen in the
sample
gas, their output is a function of the
number
of molecules of oxygen 'per unit
volume'.
Readouts in percent or PPM are permissible only when the total pressure of the sample gas being analyzed remains
constant. For optimum accuracy, the pressure of the sample gas and that of the calibration gas must be the same (in
reality, within 1-2 psig).
Temperature: The rate of diffusion of oxygen molecules into the sensor is controlled by a thin Teflon membrane
otherwise known as an 'oxygen diffusion limiting barrier'. All diffusion processes are temperature sensitive, therefore,
the fact that the sensor's electrical output varies with temperature is normal. This variation, however, is relatively
constant (2.5% increase per ºC increase in temperature).
A temperature compensation circuit employing a thermistor offsets this effect with an accuracy of better than +5%
(over the entire Operating Temperature Range of the analyzer) and generates an output function that is virtually
independent of temperature. There is essentially no error in measurements if the analyzer calibration and sampling are
performed at the same temperature or if the measurement is made immediately after analyzer calibration. Lastly, a
small sample/ambient temperature variations (within 10-15º) produce < 2% error in measurements.
Accuracy:
In light of the above parameters,
the overall accuracy of an analyzer is affected by two types of errors:
1) those producing 'percent of reading errors', as illustrated by Graph A below, such as +5% error in temperature
compensation
circuit due to
tolerances in electronic components and 2) those producing 'percent of full scale errors',
illustrated by Graph B, such as +1-2% linearity errors in readout devices, which are generally very minimal due to
today's advancements in technology and the fact that these errors are 'spanned out' during calibration. Graph C
illustrates these 'worse case' specifications that are typically used to develop the analyzer's overall accuracy statement
of < 1% of full scale at constant temperature and pressure or < 5% over the operating temperature range. The error
in QC testing is typically < 0.5% prior to shipment of analyzers.
Example: As illustrated by Graph A, any error due to the tolerances in the circuit, will increase with increasing oxygen
concentration if the analyzer calibration is done at lower end of the range, e.g., calibration with 20.9%, any error
would be multiplied by a factor of 4.78 (100/20.9) when used for measurements near 100% oxygen. Conversely, an
12
