Potter T.D., Colman B.R. (co-chief editors). The handbook of weather, climate, and water: dynamics, climate physical meteorology, weather systems, and measurements
Подождите немного. Документ загружается.
04106 Wind Monitor-JR=MA is the sixth standard model in the Wind Monitor
line (Fig. 5).
Late in 1995 we had to do some extensive testing of all products, including all
models of the Wind Monitor and associated signal conditioning and display
modules, to determine conformance with the European requirements regarding elec-
tromagnetic emissi ons and susceptibility to electromagnetic interference known as
the ‘‘EMC Directive’’ (Electro-Magnetic-Compatability). Most products met the
requirements without modification, but a few circuits required some modification,
and in most installation s the use of shielded cable is required. After it was deter-
mined that the sensors met these requirements, we were required to affix a ‘‘CE’’
label (from the French ‘Conformite
´
Europe
´
an’) to the sensor and also add a
‘‘Declaration of Conformity’’ to the instruction manual for each shipment destined
for Europe.
13 SERIAL OUTPUT
After many months of development yet another standard model of the Wind Monitor
was added to the product line. This new model incorporates an absolute optical
encoder in place of the potentiometer for the wind direction transducer. Because
of the very limited space available within the body of the Wind Monitor, it was
necessary to de sign and fabricate a custom encoder that utilizes a unique scanning
technique to determine azimuth angle to better than one degree. The operation of the
encoder and the signal conditioning are controlled by a microprocessor in the junc-
tion box. There are three selectable output protocols, which are changed by moving a
small jumper on the microprocessor circuit card. One output is specially developed
by the National Center for Atmospheric Research (NCAR) in Boulder, Colorado, for
its portable weather stations. A second output is a standard used in the marine
industry d eveloped by the National Marine Electronics Association (NMEA). The
third output is our own, which we refer to as the ‘‘RMY’’ (R. M. Young) protocol.
The outputs are a serial data stream, which can either be polled or continuous. In
addition there are cali brated analog voltage outputs. This seventh standard model
was introduced in December, 1995, as Model 09101 Wind Monitor-SE (serial
output) (Fig. 6).
Icing continues to be a major limitation on Wind Monitor operation. The Wind
Monitor is currently being deployed in many remote areas where there are occasional
heavy rime ice co nditions and limited power for any effective form of heating.
We believe that the most promising potential solution is some form of passive
surface modification. We have been investigating a couple of different surface treat-
ments that we are hopeful will reduce the ability of clear and rime ice to adhere to the
propeller and other outer surfaces of the sensor. Several researchers have reported
varying degrees of success with a spray-on material called Vellox, which was devel-
oped specifically to reduce ice adhesion. This material seems to work quite well for a
period of time following its application, but it rubs off easily and requires reapplica-
tion at regular intervals to be fully effective. Another form of permanent hydropho-
776 COMMERCIAL RESPONSE TO MEASUREMENT NEEDS
bic surfa ce treatment is currently being investigated but results have so far been
inconclusive and further testing is still needed. We are also investigating a new
epoxy paint that is specially formulated to prohibit ice adhesion. If we are able to
find a surface treatment that is practical and effective, it may result in the introduc-
tion of yet another model of the Wind Monitor.
The continually evolving needs of customers have dictated many changes as well
as the ongoing development of a series of similar but quite different instruments. By
this process products are continuously being improved and adapted to new demands.
BIBLIOGRAPHY
Anemometer Comparison Project (1986). A Report to the Atmospheric Environment Service
of Canada, RVI=50.09, Resource Ventures Incorporated, Charlottetown, Prince Edward
Island C1A 3W2.
Crawford, K. C., F. V. Brock, R. L. Elliott, G. W. Cuperous, S. J. Stadler, H. L. Johnson, and
C. A. Doswell, The Oklahoma Mesonetwork—A 2 Pt Century Project, Preprints, Eighth
International Conference on Interactive Information Processing Systems for Meteorology,
Oceanography, and Hydrology, Atlanta, American Meteorological Society, pp. 27–33.
Figure 6 (Top left) Model 05103 Wind Monitor (standard model), August, 1984. (Bottom
middle) Model 09101 Wind Monitor-SE (serial output), December, 1995. (Top right ) Model
05106 Wind Monitor-MA (marine model), April, 1993. See ftp site for color image.
13 SERIAL OUTPUT 777
Frietag, H. P., M. J. McPhaden, and A. J. Shepard (1989). Comparison of equitorial winds as
measured by cup and propeller anemometers, J. Atmos. Oceanic Technol. 6, 327–332.
Hayes, S. P., L. J. Magnum, J. Picaut, A. Sumi, and K. Takeuchi (1991). TOGA-TAO: A moored
array for real-time measurements in the tropical Pacific Ocean, Bull. Am. Meteor. Soc.,
72(3), 339–347.
Jones, D. W., D. B. Hatton, D. A. Jenkins, and A. P. Scott (1992). A Field Comparison of Some
Wind Sensors, Report No. 49, WMO Technical Conference on Instruments and Methods of
Observation (TECO-92), Vienna, pp. 347–350.
Michelena, E., and J. Holmes (1983). A Rugged, Sensitive, and Lightweight Anemometer Used
by NDBC for Marine Meteorology, Preprints, Fifth Symposium on Meteorological
Observations and Instrumentation, Toronto, American Meteorological Society.
Michelena, E., and J. Holmes, The Meteorological and Oceanographic Sensors Used by the
National Data Buoy Center, Proceedings MDS 9186 Marine Data Systems International
Symposium, New Orleans, Marine Technology Society, pp. 596–601.
Payne, R. E. (1988). The MR, a meteorological sensing, recording and telemetering package
for use on moored buoys, J. Atmos. Oceanic Technol. 5, 286–297.
Phillips, C., D. Bumham, L. Jacobs, and D. Hazen (1992). 1991 LLWAS Anemometer Test
Program, Final Report DOT=FAA=W=92-1, DOT-VNTSC-FAA-92-6, Research and
Special Programs Administration, John A. Volpe National Transportation Systems
Center, Cambridge, MA 02142-1093, September. (Document available through N.T.I.S.,
Springfield, VA 22161).
Qualid, G., P. Gregoire, M. Gilet, B. Hoegger, and A. Heimo (1993). WMO Wind Instrument
Intercomparison, Preprints, Eighth Symposium on Meteorological Obser vations and
Instrumentation, Anaheim, American Meteorological Society, pp. 274–278.
Weller, R. A., and D. S. Hosom (1989). Improved Meteorological Measurements from Buoys
and Ships for the World Ocean Circulation Experiment, Proceedings Oceans 1989, Seattle,
IEEE, pp. 1410–1415.
Weller, R. A., D. L. Rudnick, R. E. Payne, J. P. Dean, N. J. Pennington, and R. P. Trask (1990).
Measuring near-surface meteorology over the ocean from an array of surface moorings in
the subtropical convergence zone, J. Atmos. Oceanic Technol. 7, 85–103.
Weller, R. A., M. A. Donelan, M. G. Briscoe, and N. E. Huang (1991). Riding the crest: A tale
of two wave experiments, Bull. Am. Meteor. Soc. 72(2), 163–183.
WMO (1997). Wind Instrument Intercomparison 1992–1993, Meteo-France=Swiss Meteoro-
logical Institute, World Meteorological Organization, Commission for Instruments and
Methods of Observation, Final Repor t, October.
778
COMMERCIAL RESPONSE TO MEASUREMENT NEEDS
CHAPTER 39
COMMERCIAL RESPONSE TO
MEASUREMENT SYSTEM DESIGN
ALAN L. HINCKLEY
1 INTRODUCTION
The commercial opportunity to produce multiple measurement systems began when
measurement instrumentation was first used to document observations. While the
following contains examples specific to Campbell Scientific’s experience in
measurement system development, the information presented reflects the direction
taken by the measurement indus try as a whole.
As a commercial producer of more than 10 different measurement systems, a
great deal of practical experience has been gained in their design and in the repre-
sentativeness and uncertainty of their measurements. Over 75,000 systems have been
sold since the first battery-powered datalogger, the CR5, was shipped in August,
1975. Although this commercial success has given growth and economic stability,
the greatest gain has been what we and our customers have learned about improving
unattended environmental measurements.
Experience has taught us that measurement system design is optimized when it is
based on theory and observation. The degree of confidence one can have in a system
design is the degree to which the theory and observation agree. A measurement
system may be designed and operational, but without some theoretical basis sources
of uncertainty cannot be investigated (Campbell, 1991).
New designs should include a method by which the machine can detect and
respond to its most probable failures—by observing and recording the errors and
failures of current systems, better ones may be produced. Experience with similar
designs that have been produce d and suppor ted in quantity is a most valuable asset
in the design of new systems. Two principles that lead to good system design are (1)
Handbook of Weather, Climate, and Water: Dynamics, Climate, Physical Meteorology, Weather Systems,
and Measurements, Edited by Thomas D. Potter and Bradley R. Colman.
ISBN 0-471-21490-6 # 2003 John Wiley & Sons, Inc.
779
making sure that the needs of the customer are met and (2) recognizing that there is
an optimum high level of quality at which the total costs of desig ning, producing,
selling, and servicing a product are minimized.
From manually measured merc ury thermometers and staff gages, to chart recor-
ders with spring-powered clocks, to microprocessor-based dataloggers, the tools for
making good measurements have evolved. The se new tools are more accurate,
operate longer in harsher environments, and provide many more data retrieval
options.
While the modern measurement system eliminates many of the uncertainties
associated with manual measurements, others remain and new ones have been intro-
duced. Using the measurement of temperature as an example, the modern systems
have solved, within budget limits, the problem of insufficient human resources to
measure the temperature in remote areas over extended periods of time. Most errors
associated with manual data transcription and data transcription delays have been
eliminated. Remaining are siting and exposure errors associated with the new
temperature sensor s and radiation shields. Inaccuracies in the liquid-in-glass thermo-
meters are replaced by inaccuracies in the electronic temperature sensor (thermistor,
resistance temperature detector (RTD), or thermocouple). The eye-sight-induced
parallax error associated with manual measurement of a liquid-in-glass thermometer
has been replaced by the datalogger’s measurement error. Introduced are the effects
of temperature changes on the meas urement system’s performance. While they do
occur in the human observer, memory failure, run-down power supplies, and light-
ning damage are general ly considered new uncertainties. Also introduced but often
unresolved are a new set of ‘‘phantom’’ events like the rain caused by geckos (a
tropical lizard) playing ‘‘teeter-totter’’ on the tipping bucket of a rain gage or the
instantaneous half-meter of snow caused by an elk that lies down under the acoustic
snow depth sensor.
The rest of this chapter will discuss some important features designed into the
modern measurement system that improve its performance and reduce the uncer-
tainties associated with unattended measurements. Where possible, the uncertainties
will be defined and quantified. This information will facilitate the selection of good
equipment and provided needed metadata (characterizing information) about the
new measurement systems. Where they are helpful, anecdotes illustrate problems
that have been overcome and failures that have led to greater understanding.
2 MEASUREMENT SYSTEM
Today’s microprocessor-based measurement systems range from dedicated units with
fixed sensor configurations, measurement rates, and reporting intervals to program-
mable syst ems with flexible measurement electronics. Dedicated systems have
advantages in certain applications but, in general, the ability to accommodate
different sensor types without additional signal conditioning as well as the ability
to perform on-site calculations offers advantages for all but the most routine
applications. Only electronic digital systems that are capable of operation without
alternatin g current (ac) power are considered in this chapter.
780 COMMERCIAL RESPONSE TO MEASUREMENT SYSTEM DESIGN
All digital measurement systems must include (1) the measurement electronics to
convert sensor signals to digital values, (2) either electronic storage media to collect
the data on-site or telecommunications hardware to transmit the digital values, or
both, (3) sensors that provide an electronic signal, and (4) mounting hardware,
protective enclosures for the electronics, and power supplies (Tanner, 1990). In
many of today’s systems, a stand-alone environmental datalogger performs the
measurement, processing, and on-site data storage functions as shown in Figure 1.
The datalogger, the cornerst one of a data acquis ition system, has specific design
requirements determined by the application. Our first fully digital datalogger, the
CR21, was designed to meet the weather station needs of the agricultural and
environmental studies disciplines (Schimmelpfennig et a l., 1979). The datalogger
design features required by these applications include:
Sensor Compatibility Di rect connection and measurement of sensor signals
without external signal conditioning circuitry reduces cost, complexity,
measurement error, and power requirements. The ability to resolve signals to
the required precision affects the choice of sensor.
On-Site Processing Field processing reduces data storage requirements,
scales sensor signals to engineering units, and provides logic decisions for
control applications.
Field Observation of Measurements Field verification of sensors requires the
ability to continuously observe instantaneous measurements, in engineering
units, from a display or printer.
Input Transient Protect ion Environmental datalogging is vulnerable to major
hardware damage caused by large lightning-induced transients entering the
system on sensor leads. Protection hardware such as transorbs and spark gaps,
and proper grounding procedures are required to minimize damage.
Hardware Microprocessor Reset Unattended, processor-based instrumenta-
tion should reset the processor, restoring normal execution in the event it is
altered d ue to input transients or intermitten t component failure. User-entered
programs should exist in write-protected memory, minimizing the possibility
that the processor can overwrite the program should an abnormal execution
state occur.
Figure 1 Generalized data acquisition sequence.
2 MEASUREMENT SYSTEM 781
Low Power Consumption Operation from direct current (dc) power supplies
(batteries) and low current drain are required. An average power consumption
of 50 mW (12 V at 4 mA) allows 2.5 months of operation on eight, alkaline D
cells (7.5 Ah).
Operation in Adverse Environmental Conditions Operation at high tempera-
ture and humidity are the main concerns in agricultural applications (at least
until snow algae becomes the next food craze), but environmental dataloggers
must operate and maintain measurement specifications over a minimum range
of 20
Ctoþ55
C. Solar heating can raise datalogger enclosure temperatures
20
C above air temperatures. Tight enclosures and desiccant provides the
simplest and most cost-effective means of preventing water from condensing
on the electronics.
Remote Comm unication Capability Telephone was the only remote commu-
nication method available for the CR21. Since that time, data telemetry has
become increasingly important. Models requiring climatic data are run daily to
predict water use, crop development, disease and pest growth, etc. Synoptic
models utilizing 15-min data are run several times a day. These applications
require the timely transfer of data via radio, telephone, or satellite.
Logic and Control The ability to compare values or time, with programmable
limits and make decisions, provides powerful control capability. Samplers can
be switched on or off, sensors with high current drain can be powered only
during the measurement.
Good system design requires that the sensors be properly sited and mounted and
that the electronics be protected and grounded. The accuracy and response times of
the sensor needs to match those of the datalogger and the needs of the investigator.
The data exchange between the measurement electronics and the data
retrieval=storage hardware needs to be easy, reliable, and accurate.
Sales of our seventh measurement system, the Basic Data Recorder (BDR), did
not meet expectations. Only 2000 BDRs were sold compared to sales of 35,000 of
the datalogger built just before it. The BDR met the specific design requirements
spelled out in a government bid, but it lacked the measurement versatility, telecom-
munication options, and software support required by our ‘‘typical’’ customer. Had
the datalogger been built to fill both requirements, it would have been more success-
ful. In spite of the failure, valuable lessons were learned. A new ‘‘table-based’’ data
storage method was developed. Tighter government radio frequency (RF) noise
requirements led to improved RF testing, shielding, and eventually to European
market ‘‘CE Compliance.’’ Both have been invaluable in the development of
today’s dataloggers.
3 DATALOGGER
A datalogger is made of pieces and processes that must fit and function together to
make accurate measurements on low power under extreme environmental condi-
782 COMMERCIAL RESPONSE TO MEASUREMENT SYSTEM DESIGN
tions. Seemingly simple design details like the style of a connector or the circuit
board cleaning process are critical to the long-term perfor mance of the unit. For
example, the switch from a rosin-core solder flux to a water-soluble flux midway in
the life of the first datalogger flooded the service department with repairs, threaten-
ing the existence of the new company. The dataloggers were failing because the flux
remaining on the boards would adsorb water under humid conditions, become
conductive, and cause shorts. Water-soluble fluxes were not as good as they are
today and our procedure for completely cleaning it from the circuit boards had not
yet bee n perfected.
Most modern datalogger uncertainties can be separated into two categories, fail-
ures or routine inaccuracies. Most failures result in the loss of part or all of the
digital data. Only a few compromise the quality of the recorded data. Routine
inaccuracies are the day-to-day errors in the datalogger’s measurements that affect
the quality of the recorded digital data.
Measurement system failures can be caused by:
Failure of one of the components or circuit boards wi thin the datalogger.
Environmental conditions such as lightning, humidity, heat=cold, dust, floods,
fire, wind, hail , ice, ultraviolet (UV) degradation, humans, animals, birds,
insects, mold, etc.
Human error in the wiring or programming of the datalogger.
Failure of one of the power system components (battery, regulator, solar panel,
etc.). Sometimes a datalogger component fails in a mode where it draws
excessive current running down the battery prematurely.
Failure of a sensor or sensor cable.
Failure of the data storage=telemetry system.
Failure of the protective enclosure or mounting hardware.
Inaccuracies of the measurement system that are routine in nature are caused by:
Inaccuracies inherent in the voltage measurem ent, pulse measurement, or clock
circuits over the temperature range of the datalogger. These uncertainties are
quantified by the accuracy specification given by the manufacturer and should
include the temperature range over which it is valid.
Inaccuracies in a sensor ’s measured parameter induced by another environ-
mental variable (e.g., temperature-induced inaccuracies in a relative humidity
measurement).
Changes over time of components in the voltage measurement, pulse measure-
ment, and clock circuits–recalibration issue.
Changes over time in the sensor’s response to the parameter being measured—
recalibration issue.
Fortunately, well-designed dataloggers rarely fail and their routine inaccuracies
are relatively small and measurable. Many of the above-mentioned failures and
3 DATALOGGER 783
inaccuracies are prevented or min imized by good datalogger design and manufactur-
ing, system maintenance, and user training. Some are minimized by paying more for
units with greater accuracy or reliabil ity. Other failures, polar bears, for example, are
simply Murphy’s law at work. While it is not comprehensive, the following discus-
sion details some of the more important features designed into a good environmental
datalogger.
Low-Power Microprocessors
The microprocessor and quartz crystal clock are the brain and the heart of the
datalogger. The processor does the computations and controls the operational
sequences while the clock controls the timing and to some extent the speed. As
evidenced by the cooling fans built on top of the Pentium processors in today’s
personal computers (PCs), most processors use a large amount of power. In a
datalogger, which needs to run for 6 months on the power in eight alkal ine
D cells, it is important to use processors that require very little power.
In the early 1970s, a new logic circuit technology called CMOS (complementary
metal-oxide semiconductor) was introduced. The first CMOS chips released were
simple AND, OR, and NOR gates. About 2 years later, RCA introduced the 1802,
the first CMOS-based microprocessor. The power consumed by a CMOS processor
is low while changing states ( processing), but it drops to less than 1 mA when it is
inactive or ‘‘quiescent.’’ Low-power system design requires that power consumption
be minimized by placing CMOS or equivalent processors in the quiescent mode in
between events.
The opportunities presented by this new microprocessor technology were recog-
nized with great excitement by Eric Campbell, one of the founders of the company.
His enthusiasm was such that he and his brother Evan built their younger brother
Wayne a Christmas present, a CMOS-based ‘‘alarm clock’’ with speakers and light-
emitting diodes (LEDs) that went off at 6 a.m. Christmas morning. The CMOS logic
chips were first utilized in the CA9 path averaging laser anemometer and in the first
datalogger, the CR5. Shortly after its release, the 1802 was added to the CR5
datalogger making it the first, or one of the first, microprocessor-based dataloggers.
Three subsequent dataloggers—the CR21, CR7, and 21X—also utilized the
low-power processing of the 1802. Quick recognition of the benefits of the 1802
processor placed the company right in the middle of an adventurous revolution in
measurement system design.
Customer requirements and technology are constantly changing. Requirements
for incre ased measurement and computational speed led to the use of two different
versions of Hitachi’s 6303 CMOS microprocessor in the next four dataloggers. The
increased capability of this processor enabled subsecond measurement rates on a
small number of channels or one-second rates on all channels. This speed is suffi-
cient to meet or exceed most measurement needs in the weather, climate , and water
fields. New microprocessors are introduced into new or upgraded dataloggers to
meet changing customer requirements or because evolving technology has made
the older parts unavailable. However, each microprocessor has its own quirks. Test-
784 COMMERCIAL RESPONSE TO MEASUREMENT SYSTEM DESIGN
ing on the 6303 processor showed that its internal ‘‘overrun timer’’ was not reliable,
so other methods had to be developed to work around the bug. Good system design
requires extensive testing of a new processor to become familiar with its strengths
and weaknesses.
‘‘Watch Dogs’’
Microprocessors specifically and digital circuits in general are quite susceptible to
static electricity. Quite often, however, the static will ‘‘bomb’’ the processor or latch
up a chip or change an addres s without causing permanent damage. Because light-
ning and other forms of static electricity are a very real part of the datalogger’s
environment, good system design requires that a ‘‘watch-dog’’ circuit be added to the
system to reboot it when it crashes. A watch dog is a hardware count-down timer
circuit that reboots the processor when the processor bombs for more than a preset
time. It is like an automated ‘‘Ctrl–Alt–Del.’’ The watch dog may also include a set
of software checks that reboot the system if its system parameters are not within
set limits. Once a watch-dog reset occurs, the processor reboots and special software
is executed that attempts to restore the system to proper operation without the loss of
data. The watch dog also assists in the development of datalogger software because it
detects when a software bug causes the system to crash. The watch dog is a critical
design requirement for all environmental dataloggers.
Clock
The clock needs to be inexpensive and low power but accurate enou gh that it can be
left unattended for up to a year in places like Antarctica without a significant shift in
time. The datalogger’s clock is a quartz crystal that ideally oscillates at a constant
frequency. In low-power less expensive crystals the frequency varies slightly as a
function of temperature. The raw quartz crystal’s accuracy of 4 min per month is
improved in the datalogger to 1 min per month by measuring the clock’s tempera-
ture and correcting the clock based on an algorithm permanently stored in the
datalogger. While satellite applications require even greater clock accuracy, clocks
that stay accurate to within 1 min per month meet most requirements.
Self-Contained Measurement Circuitry
Most of the sensor s used in the fields of weather, climate, and water output a voltage
or change resistance in response to the parameter they are sensing. The rest eith er
output pulses or digital data via a serial protocol. Well-designed environmental
dataloggers measure multiple voltage, pulse, and serial signals without the need
for additional external signal conditioning, thus reducing the cost, complexity,
errors, and power requirements associated with the extra circuitry.
3 DATALOGGER 785