102
Preparing electronic sensor signals for storage and transmission takes the name
of signal conditioning. Initially, the tenuous sensor signal is electronically amplified
by applying the data signal, as current or voltage, to the gate of a power transistor
across which a high power carrier signal is modulated thus achieving amplification
of the initial signal up to about a hundredfold. This output is then digitized and filtered for storage and transmission. Electronic digitization called quantization is carried out by electronic modules known as analog to digital converters (ADC). Central
to the ADC is the sample-and-hold circuit whereby the signal voltage is removed
from the sensor circuit over the interval of the preestablished instrument time set
(the clock speed), stored in a capacitor and, from there, sampled by comparator
circuits to determine the appropriate voltage bin. Some comparators sample to an
array of voltage bins, others ramp through a preestablished number of voltage bins.
In any case, the number of possible bin determines the resolution of the ADC.
Quantization and Data Precision
Quantization necessarily causes error due to the rounding off procedure which
truncates an essentially infinite number. Quantization error however decreases as
the resolution of the circuitry increases. An initial example expressed in decimal
notation is instructive: For an error of one in one hundred (1:100), a resolution of
two digits suffices (00–99) and a range of one hundred discrete values will represent the analog span observed. If digitization resolutions allow reducing the
quantization error to one in one thousand (1:1000), three digits are required
(000–999) to represent the data set. It is to be noted that the increased resolution
of one digit makes the size of the data set increase by a factor of 10. In binary,
rather than decimal notation, these resolutions are measured in bits. For perspective, a 10 bit number covers a decimal range from 0 to 1024 (2
10
). Since AD
converters today can perform quantization ranging up to 24-bit resolution, more
than 16 million (2
24
) possible discrete decimal digital bins may be achieved!
Telling Analog from Digital
Analog and digital can easily be differentiated by considering the data processing of an old-style reversing thermometer. Then, the researcher would read out
the analog signal of the mercury meniscus height against the calibrated scale
through a magnifying glass and, putting the data (a set number of digits) to
paper, would, in the act, digitize the reading. Two digits were directly read off
the scale and, with a sharp eye, a third digit could be subjectively interpolated.
Similarly, early electronic instrument readout such as CTD and XBT temperature or salinity readings consisted of an analog stylus trace on graph
paper. It was necessary to digitize the readings for both temperature and depth
by reading out against the graph paper line grids graduated in meters and
degrees centigrade or parts per thousand salinity. The last significant digit was
similarly estimated.
5 Signal Conditioning, Data Telemetry, Command Signaling and Platform Positioning…
Preparing electronic sensor signals for storage and transmission takes the name
of signal conditioning. Initially, the tenuous sensor signal is electronically amplified
by applying the data signal, as current or voltage, to the gate of a power transistor
across which a high power carrier signal is modulated thus achieving amplification
of the initial signal up to about a hundredfold. This output is then digitized and filtered for storage and transmission. Electronic digitization called quantization is carried out by electronic modules known as analog to digital converters (ADC). Central
to the ADC is the sample-and-hold circuit whereby the signal voltage is removed
from the sensor circuit over the interval of the preestablished instrument time set
(the clock speed), stored in a capacitor and, from there, sampled by comparator
circuits to determine the appropriate voltage bin. Some comparators sample to an
array of voltage bins, others ramp through a preestablished number of voltage bins.
In any case, the number of possible bin determines the resolution of the ADC.
Quantization and Data Precision
Quantization necessarily causes error due to the rounding off procedure which
truncates an essentially infinite number. Quantization error however decreases as
the resolution of the circuitry increases. An initial example expressed in decimal
notation is instructive: For an error of one in one hundred (1:100), a resolution of
two digits suffices (00–99) and a range of one hundred discrete values will represent the analog span observed. If digitization resolutions allow reducing the
quantization error to one in one thousand (1:1000), three digits are required
(000–999) to represent the data set. It is to be noted that the increased resolution
of one digit makes the size of the data set increase by a factor of 10. In binary,
rather than decimal notation, these resolutions are measured in bits. For perspective, a 10 bit number covers a decimal range from 0 to 1024 (2
10
). Since AD
converters today can perform quantization ranging up to 24-bit resolution, more
than 16 million (2
24
) possible discrete decimal digital bins may be achieved!
Telling Analog from Digital
Analog and digital can easily be differentiated by considering the data processing of an old-style reversing thermometer. Then, the researcher would read out
the analog signal of the mercury meniscus height against the calibrated scale
through a magnifying glass and, putting the data (a set number of digits) to
paper, would, in the act, digitize the reading. Two digits were directly read off
the scale and, with a sharp eye, a third digit could be subjectively interpolated.
Similarly, early electronic instrument readout such as CTD and XBT temperature or salinity readings consisted of an analog stylus trace on graph
paper. It was necessary to digitize the readings for both temperature and depth
by reading out against the graph paper line grids graduated in meters and
degrees centigrade or parts per thousand salinity. The last significant digit was
similarly estimated.
5 Signal Conditioning, Data Telemetry, Command Signaling and Platform Positioning…