proudly sponsored by
|
archiving of SOWACS proudly sponsored by |
|
To the list:
Three terms, resolution, repeatability and accuracy, related to calibration
of electronic devices sometimes become confusing. My understanding and use
of them is described below:
Resolution: The count of scientifically significant digits, whether
decimal or binary, of a measurement indication. Examples of six digit
resolution would be 345678, 34567.8 and 003456.78. The degree of resolution
does not necessarily impart any value to the reading indicated.
Flash Analog to Digital conversion resolution is typically described as the
number of bits which can be resolved by the convertor. The system
measurement resolution may be much less, depending on how the convertor has
been implemented. Good practice dictates employing as much of the
conversion range as possible without encountering overflow conditions.
Repeatability: Repeatability specifications reflect possible variations
in the performance of the sensor over the range of operating and
environmental conditions expected to be encountered.
A measurement value sometimes depends on the stimulus "path" taken to
achieve the current reading. This degradation in repeatability is called
hysteresis.
Sometimes sensor readings change due to aging or variations in parameters,
such as temperature or chemical reactions, which are not being measured.
Accuracy: Reflects the true value of a sensor. A statement of accuracy
implys comparison with a traceable standard. The US National Institute of
Standards and Traceability (NIST), as well as other government agencies,
provides calibration procedures and maintains such standards.
Statements of accuracy, either absolute or relative, should be supported by
calibration procedures employing measurement standards which are maintained
and certified according to NIST recommendations. If specific procedures are
not available directly from NIST, they shold be developed by appropriately
qualified professionals according to best industry practice.
National and international commerce depends on statements of accuracy
followed by actual performance of goods and services according to these
statements. Legal sanctions may be imposed for making false statements of
accuracy.
Statements of resolution, repeatability and accuracy reflect three
interdependent qualities. Ideally, resolution exceeds accuracy and
repeatability exceeds resolution. In practice, resolution can exceed
accuracy by many magnitudes which can lead to false confidence in
measurements.
Consider a beginning physics student who, having just read the room
temperature using a mecury thermometer calibrated in degrees Celsius,
converts to Farenheit with a 12 digit calculator. Beware the ides of
digital technology!
Having made the above premise, I make the following statements as supplement
to earlier comments about the measurement of absolute water concentrations
in soil:
The HydroTek sensor provides a continuous, monotonic, analog output signal
over the specified range. The resolution of an attached metering instrument
completely determines the resolution of any measurement reflected.
Thermal electrical noise levels, typically a few microamps, limit the
resolution of the sensor output.
The factory calibration of HydroTek sensors, repeated using NIST
instrumentation and noncertified buckets of water, indicates average short
term variations on the order of 0.1% or less.
Due to the variety of environmental conditions and length of time required
for substantiation, no claims are made for long term repeatability.
The previous statement concerning measurement of absolute accuracy follows:
>
>In considering installation and calibration methods for the HydroTek
sensor, it has become apparent that a single data sheet will not suffice.
Possibly, a single, meticulous calibration procedure will not even suffice.
Because soil moisture sensors cannot differentiate between soil and air,
volumes which can be occupied by water depend on compaction. Any activity
which disturbs compaction will probably alter the concentration of water.
Sensor geometries can be developed to minimize this disturbance, but will
probably never be small compared to the geometries of disturbance.
>
>Presently, I am of the opinion that truly accurate results (better than 1%)
can only be achieved by rigorously following an installation method designed
specifically for the soil type at the installation, and then comparing the
readings initially and periodically with laboratory evaluated samples for
several seasons.
>
>After gaining confidence in a specific region with a particular sensor
using specific methods, only then would I recommend unauthenticated
placement and use.
>
>Of course, if only relative readings matter, i.e., for direct control
applications, absolute calibration is not required as long as the sensor
provides repeatable, monotonic readings which are uneffected by other
factors such as temperature, somewhat beyond the range of interest. Then
the irrigation system can be designed to be self calibrating with the
sensors installed.
>
>M. G.
>