An Automatic Depth Control System for the
On-the-Go Soil Strength Sensor
Except where reference is made to the work of others, the work described in this
thesis is my own or was done in collaboration with my advisory committee. This
thesis does not include proprietary or classified information.
Clifford Warren Smith, Jr.
Certificate of Approval:
Randy L. Raper, Co-Chair
Affiliate Professor
Biosystems Engineering
John Y. Hung, Co-Chair
Professor
Electrical & Computer Engineering
Lloyd S. Riggs
Professor
Electrical & Computer Engineering
George T. Flowers
Interim Dean
Graduate School
An Automatic Depth Control System for the
On-the-Go Soil Strength Sensor
Clifford Warren Smith, Jr.
A Thesis
Submitted to
the Graduate Faculty of
Auburn University
in Partial Fulfillment of the
Requirements for the
Degree of
Master of Science
Auburn, Alabama
December 17, 2007
An Automatic Depth Control System for the
On-the-Go Soil Strength Sensor
Clifford Warren Smith, Jr.
Permission is granted to Auburn University to make copies of this thesis at its
discretion, upon the request of individuals or institutions and at
their expense. The author reserves all publication rights.
Signature of Author
Date of Graduation
iii
Vita
Clifford Warren Smith, Jr., son of Clifford W. Smith and Patricia L. Smith, was
born January 23, 1971 in Northfork, West Virginia. He graduated in 1989 from Mount
View High School in Welch, West Virginia. He entered the United States Army in
1989 and served in Desert Shield/Desert Storm in 1990-1991. He entered Alabama
Southern Community College in 1998, where he received an Associate of Science
degree in May 2000. He entered Auburn University in 2000, where he received a
Bachelor of Engineering degree in Electrical and Computer Engineering in May 2003.
He entered the Graduate School at Auburn University in August 2003.
iv
Thesis Abstract
An Automatic Depth Control System for the
On-the-Go Soil Strength Sensor
Clifford Warren Smith, Jr.
Master of Science, December 17, 2007
(B.S., Auburn University, 2003)
(A.S., Alabama Southern Community College, 2000)
89 Typed Pages
Directed by John Y. Hung and Randy L. Raper
Hardpans are areas of soil compaction formed by vehicle traffic or natural pro-
cesses. These compacted areas beneath the topsoil can hinder successful crop produc-
tion. A novel sensor for measuring soil strength was presented in an earlier work, and
patented as the OSSS (On-The-Fly Soil Strength Sensor). The focus of this thesis
is the design and evaluation of a control system for this new method of sensing soil
strength. The graphical user interface has been implemented with National Instru-
ment?s LabVIEW r? software, and drives an analog PID-type (proportional, integral,
and derivative) position controller for a hydraulic cylinder. This controller causes
the strength sensor to vertically oscillate in the soil at consistently spaced distance
intervals, regardless of tractor speed variations. Successful development enables this
system to collect soil strength information over an entire field at all desired depths in
a short period of time.
v
Acknowledgments
First and foremost I would like to thank Our LORD and Savior Jesus Christ,
from whom all blessings flow. And I would like to express my gratitude to the
members of the graduate committee, who supervised the thesis project; Mr. Bob
Washington, Supervisory Agricultural Engineer, for assistance with frame design;
Mr. Eric Schwab, agronomist; Mr. Morris Welch, engineering technician, Mr. Dexter
LaGrand, electronic technician; and Mr. John Walden, model maker.
Thanks are also due to my beautiful wife Stacy and my children Dant?e, Shyrell,
Janoah, Christian, Journey, and the twins Willie and Clifford III, who supported and
encouraged me through my academic pursuits.
vi
Style manual or journal used Journal of Approximation Theory (together with
the style known as ?aums?). Bibliography follows IEEE Transactions Reference Style
Computer software used The document preparation package TEX (specifically
LATEX) together with the departmental style-file aums.sty, MATLAB and LabVIEW.
vii
Table of Contents
List of Figures x
1 Background and Literature Review 1
1.1 What Are Hardpans and How Are They Formed? . . . . . . . . . . . 1
1.2 How Hardpans Affect Crop Production . . . . . . . . . . . . . . . . . 1
1.3 Remediation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Vertical Soil Strength Measurement Devices . . . . . . . . . . . . . . 4
1.4.1 Cone Penetrometers . . . . . . . . . . . . . . . . . . . . . . . 4
1.4.2 Multiple-Probe Cone Penetrometers . . . . . . . . . . . . . . . 6
1.4.3 The Shortcomings of Vertical Soil Strength Measurement Tools 8
1.5 Horizontal Soil Strength Measurement Devices . . . . . . . . . . . . . 9
1.5.1 Single-Depth Horizontal Soil Strength Measurement Devices . 10
1.5.2 Three-Depth Soil Strength Measurement Devices . . . . . . . 13
1.5.3 Soil Strength Measurement Devices That Measure More Than
Three Depths . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.6 Introduction of the On-The-Go Soil Strength Sensor . . . . . . . . . . 19
2 Control System Modeling 21
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.1.1 Typical Motion Profiles . . . . . . . . . . . . . . . . . . . . . . 21
2.2 Description of proposed hardware . . . . . . . . . . . . . . . . . . . . 23
2.3 Model description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3.1 Modeling the sensors and other signals . . . . . . . . . . . . . 24
2.3.2 Modeling the valve characteristic . . . . . . . . . . . . . . . . 27
2.3.3 Modeling the hydraulic system . . . . . . . . . . . . . . . . . . 27
2.4 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.4.1 Step Response . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.4.2 Deadband . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.4.3 Sinusoidal response . . . . . . . . . . . . . . . . . . . . . . . . 33
2.4.4 Sinusoidal and Chirp input . . . . . . . . . . . . . . . . . . . . 33
2.4.5 The plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
viii
3 Control System Design and Implementation 41
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.2 The Block Diagram Flow . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.3 Command Signal Output . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.3.1 DGPS to Calculate Oscillation Frequency . . . . . . . . . . . 42
3.3.2 Command Voltage . . . . . . . . . . . . . . . . . . . . . . . . 44
3.3.3 System Startup and Configuration . . . . . . . . . . . . . . . 47
3.3.4 Operation Mode . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.4 Deadband Compensation . . . . . . . . . . . . . . . . . . . . . . . . . 53
4 Experimental Evaluation 56
4.1 Experimental Approach . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.1.1 Step Response Tests . . . . . . . . . . . . . . . . . . . . . . . 58
4.1.2 Plant Frequency Response . . . . . . . . . . . . . . . . . . . . 60
4.1.3 PID Controller Response . . . . . . . . . . . . . . . . . . . . . 60
4.1.4 Closed Loop Response . . . . . . . . . . . . . . . . . . . . . . 60
4.1.5 OSSS In The Field . . . . . . . . . . . . . . . . . . . . . . . . 63
5 Conclusions and Future Work 65
5.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Appendices 67
A The Printed Circuit Board 68
A.1 37 Pin Connector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
A.2 9 Pin Connector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
B The Frame 72
Bibliography 75
ix
List of Figures
1.1 Diagram of hand-held soil cone penetrometer used to measure the
depth and degree of soil compaction. [1] . . . . . . . . . . . . . . . . 5
1.2 Side and front view of multiple-probe soil cone penetrometer created to
expedite the measurement of soil compaction across a row and through-
out a field [2] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3 Horizontal tip penetrometer probes. The items in the photo are (1)
cone tip, (2) cone rod, (3) cone blade, (4) prismatic tip, (5) prismatic
rod, (6) prismatic blade, (7) carrier bolt, (8) load cell, (9) I/O cable,
(10) load cell housing, (11) housing bolt, (12) nut, and (13) cable
protector [3] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.4 Acoustic compaction layer detection device. Tine design (A) with a
cone (B) and a microphone (C) [4] . . . . . . . . . . . . . . . . . . . 12
1.5 Schematic of the vertical smooth blade (edge is on the left side) [5] . 14
1.6 Instrumented Subsoiler [6] . . . . . . . . . . . . . . . . . . . . . . . . 15
1.7 The three-depth soil mechanical impedance sensor designed by Chukwu
and Bowers [7] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.8 Operational concept of the soil strength profile sensor (OTG-SSPS): [8] 17
1.9 Profile Sensor mounted on a tractor [9] . . . . . . . . . . . . . . . . . 18
2.1 The OSSS cycles once for every two segments. Vertical white lines
represent segments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.2 SIMULINK block diagram model of the OSSS . . . . . . . . . . . . . 25
2.3 Plots of the step response. Blue line plot is of the system with dead-
band. The red (+) plot is of system with deadband compensation.
Position as a result of two step functions. First step a time 0 from 0
volts to 10 volts and the second step function at time 20 seconds from
10 volts to 0 volts with valve delay. . . . . . . . . . . . . . . . . . . . 29
x
2.4 Deadband compensation . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.5 Simulation signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.6 Normalized output position and normalized flow rate plotted together.
The blue (+) is the response while the black line is the flow rate. . . . 35
2.7 Position as a result of the chirp signal. The input starts a 0.0 Hz and
increases to 1.0 Hz. And full stroke. . . . . . . . . . . . . . . . . . . 36
2.8 Chirp signal from 0-3 Hz. And 20 cm stroke. . . . . . . . . . . . . . 37
2.9 Measurement range given tractor speed and cycle rate. . . . . . . . . 39
2.10 Three dimensional plot of system limits . . . . . . . . . . . . . . . . . 40
3.1 Block diagram of the control system . . . . . . . . . . . . . . . . . . 42
3.2 Flow Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.3 Algorithm to collect RMC sentence . . . . . . . . . . . . . . . . . . . 45
3.4 Algorithm to parse RMC in LabVIEW . . . . . . . . . . . . . . . . . 46
3.5 Command signal calculator in LabVIEW . . . . . . . . . . . . . . . 48
3.6 OSSS?s user interface . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.7 System with factory set deadband compensation. The solid line shows
the input to the system. The dotted line shows the output movement
of the shank with no deadband compensation in the amplifier. . . . . 54
3.8 System with deadband compensation adjustment in place. The solid
line shows the input to the system. The dotted line shows the output
movement of the shank with deadband compensation in the amplifier. 55
4.1 OSSS Mechanics Attached to Tractor . . . . . . . . . . . . . . . . . . 57
4.2 G(s) (Open Loop Magnitude Response) . . . . . . . . . . . . . . . . . 58
4.3 G(s) Open Loop response. . . . . . . . . . . . . . . . . . . . . . . . 59
xi
4.4 Closed loop frequency response (Simulated) . . . . . . . . . . . . . . 61
4.5 Closed loop frequency response (Measured) . . . . . . . . . . . . . . 62
4.6 Closed Loop Response (collected in the field) . . . . . . . . . . . . . . 64
A.1 Printed Circuit Board . . . . . . . . . . . . . . . . . . . . . . . . . . 69
B.1 Frame1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
B.2 Frame2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
xii
Chapter 1
Background and Literature Review
1.1 What Are Hardpans and How Are They Formed?
A hardpan is a layer of soil that is more compact than the layers above and
below it. Hardpans are subsets of soil compaction and make up a small portion of the
soil profile. In structural engineering, soil compaction is essential to forming stable
construction foundations [10]. However, soil compaction is a problem in agriculture
because it reduces crop yield. Soil compaction is present in all soil and it reduces crop
yield by inhibiting seedling emergence and root penetration. These areas of increased
soil strength beneath the topsoil which are formed by either natural processes or vehi-
cle traffic, and hinder proper moisture levels necessary for successful crop production
[11].
1.2 How Hardpans Affect Crop Production
In [12], Wells et al reported that crop yields were reduced by 25%, 30%, and even
40% when soil compaction problems existed. Some of this reduction is due to root
restriction. Plants need an adequate rooting environment for optimal crop production
[13]. Hardpans have been known to prevent roots from penetrating to depths of soil
that could sustain plants during periods of short-term drought [11].
1
Another way that soil compaction reduces crop yield is by limiting the essential
elements. When soil is severely compacted, tightly packed soil particles affect soil
moisture and soil air, which are essential for plant growth [13]. Pore-size distribution
and increased soil strength are responsible for limitations of soil moisture and air
content [4]. Lack of moisture and air causes a reduction in photosynthesis, thus
reducing crop yields [13].
1.3 Remediation
To allow plant roots to penetrate to less compact, moister horizons, many pro-
ducers rely on some form of deep tillage to break through the hardpan layer [11].
Because tilling beneath the hardpan will break up the compacted area and allow
deeper rooting and improve plants? ability to withstand short-term drought, crops
grown in tilled fields, where soil compaction was previously a problem, have a better
chance to grow successfully [11].
Traditional tillage treatments have focused on preventive maintenance and have
not had the benefit of diagnostic evidence [14]. Though hardpan properties vary in
strength and depth due to management practices and soil and crop factors [4], most
farmers uniformly till at the maximum depth of their tillage equipment [15]. However,
studies have shown that the hardpan varies greatly in depth and strength from field
to field, or from one area to another within a field [11], [4], [14].
Excessively deep tillage can waste energy, increase surface erosion and decrease
crop yields due to excessive soil disturbance [11]. Changing the tillage depth according
2
to local soil conditions preserves soil ecology and saves energy [16]. In [15], Raper
et al found that the necessary tillage power could be reduced by 27% with variable-
depth tillage compared to uniform-depth tillage. Variable-depth tillage also caused
less damage to cover crops, thereby increasing crop yields [15]. Therefore, it is clear
that variable-depth tillage is necessary to optimize the resources used in the tillage
process and maximize plant yields.
The inherent inefficiency of uniform tillage treatments has been recognized by
researchers who proposed tillage systems where the hardpan depth itself prescribes
the tillage treatment [14]. In other words, the tillage depth will be based on the
hardpan depth [14]. This variable-depth subsoiling requires accurate determination
of the hardpan layer depth to be conveyed to an actuation mechanism that controls
the tillage depth [4]. Another option would be map-based variable-depth subsoiling,
which can be realized with geo-referenced hardpan depth information [4].
Most tools currently used for variable hardpan detection can be classified as
either vertical or horizontal measurement devices. The vertical and horizontal devices
both have a force transducer to measure the soil penetration resistance as they move
through the soil profile. The depth at which the necessary penetration force peaks is
the hardpan depth (if a hardpan exists). The difference in the two types of tools can
be explained by the ways in which they move through the soil profile.
3
1.4 Vertical Soil Strength Measurement Devices
Vertical devices sense the penetration resistance of soils and measure the force
needed to push a cone tip into the soil vertically [14], [17]. As the probe is being
forced into the ground, it encounters layers of various compaction intensities. The
force necessary to penetrate each layer increases as the compactness or density of that
soil layer increases.
Vertical devices have the benefit of continuous vertical measurements. With data
collected with these devices, a continuous soil strength profile can be achieved in the
vertical direction. The drawback of vertical devices is their stationary (in the lateral
sense) data collection method. This stop-and-go method limits the measurement
capabilities in the horizontal (lateral) directions and increases the time needed to
survey a field.
1.4.1 Cone Penetrometers
One such vertical device is the cone penetrometer [1], which has traditionally
been used to assess soil penetration resistance [14], [17]. The first penetrometers were
hand-held tools like the one shown in Figure 1.1.
This tool has a load-measuring device in series with the shaft and handle. The
load-measuring device, which is aligned with the shaft and handle, measures the force
applied to the handle as the cone is pushed into the soil.
4
Figure 1.1: Diagram of hand-held soil cone penetrometer used to measure the depth
and degree of soil compaction. [1]
5
Variations of cone penetrometers may have been used as early as the 1800?s.
Although it is not well documented, a needle-type penetrometer was used to esti-
mate various types and consistencies of soil [17]. The Danish railroad companies used
pocket penetrometers to determine maximum allowable bearing pressure in the 1930?s
[17]. In 1948 the U.S. Corps of Engineers developed hand-operated cone penetrome-
ters for predicting trafficability of vehicles at the Waterways Experiment Station in
Mississippi [17]. The first tractor-mounted penetrometer was designed and built by
J. R. Williford, O. B. Wooten, and F. E. Fulgham in 1972. This penetrometer was
mounted on rails that allowed its one probe to be moved from side to side [18]. An
X-Y plot was constructed with the data collected.
1.4.2 Multiple-Probe Cone Penetrometers
Although soil cone penetrometers can determine soil strength and various levels
of soil compaction, they are still not fast enough when surveying large fields, where
several samples are needed. In one experiment it took three days to collect 800 sets
of force-depth data using a single-probe device [2]. Some of the recently developed
cone penetrometers have multiple probes (see Figure 1.2) that are forced into the soil
simultaneously.
Adhering to the standards set forth by the American Society of Agricultural
Engineers in 1993 (ASAE Standard S313.2), Raper et al developed a multi-probe soil
cone penetrometer (MPSCP) containing five probes instead of one [2]. With this
6
Figure 1.2: Side and front view of multiple-probe soil cone penetrometer created to
expedite the measurement of soil compaction across a row and throughout a field [2]
7
MPSCP, the data from five positions across the row could be collected at one time,
reducing the time necessary to collect data from a plot.
1.4.3 The Shortcomings of Vertical Soil Strength Measurement Tools
In large-scale field settings, the vertical cone penetrometer is an impractical
method of determining soil compaction [14]. Multiple-probe cone penetrometers are
faster than the single-probe penetrometers and are less labor-intensive, but they are
still relatively slow. Another problem with this system is that the multiple-probe cone
penetrometer, like the single-probe penetrometer, can only determine soil strength at
discrete sampling points. This stop-and-go sampling procedure is hard to incorporate
into a continuous variable-depth tillage practice [4] and maps interpolated from this
point data are limited due to small measurement density [5]. Hand-held penetrom-
eters like the one shown in Figure 1.1 have been found to be difficult to use if the
soil contains very compact layers [19]. When the soil is compacted, the operator has
trouble maintaining a constant speed when inserting the unit into the ground [19].
This is a problem because speed variations can influence the readings [19]. Machine-
driven penetrometers use hydraulic fluid power to force the probe into the ground
at a constant speed. However, this hydraulic power comes with bulky machinery
and a tractor to supply the fluid flow. Horizontal devices would have comparable
hydraulic machinery but would require less time to collect data compared to vertical
penetrometers.
8
1.5 Horizontal Soil Strength Measurement Devices
Several attempts have been made to develop horizontal cone index measurement
devices. Horizontal devices measure the force needed to push a wedge tip through
the soil horizontally. As the probe is being pulled or pushed through the soil, it will
encounter differences in soil mechanical impedance [20]. The mechanical impedance
increases as the compactness of the soil increases. Unlike the vertical devices, the
horizontal devices are designed to be inserted into the soil and pulled. This method
of soil strength measurement eliminates the stop-and-go drawbacks of the vertical
system, but unfortunately presents disadvantages of its own.
Horizontal devices have the benefit of continuous horizontal measurements. With
the data collected with horizontal devices, a continuous soil strength profile can be
achieved in the horizontal direction. Horizontal or on-the-go soil mechanical resistance
measurements allow substantial increase in measurement density and speed [16]. The
disadvantage of horizontal devices is their constant depth measurement limitations.
The depth of measurement remains constant while the hardpan depth is varying
throughout the field. Because the soil strength profile is measured for only certain
depths, it is possible to incorrectly measure the hardpan location. Many of these
devices are discussed subsequently.
9
1.5.1 Single-Depth Horizontal Soil Strength Measurement Devices
Horizontal Blade Penetrometers
Alihamsyah and Humphries built a tool capable of rapid soil strength determina-
tion at one depth [14]. This tool (seen in Figure 1.3) uses a prismatic tip penetrometer
attached to a force transducer [8].
Alihamsyah et al developed and tested this device in 1990 and found that this
horizontal penetrometer could serve as soil impedance measurement device [3]. Ali-
hamsyah proposed the idea of mounting a horizontally operating soil strength mea-
suring tip near the front of a tractor providing soil physical data to a microcomputer
[3]. This system could be an integral part of a control system that would adjust
tillage equipment automatically [3].
Acoustic Compaction Layer Detection
M. Tekeste, a research graduate student assistant at the University of Georgia,
developed an acoustic system that could be used as an on-the-go hardpan detection
device [4]. This acoustic compaction layer detection device (ACLDD) (seen in Figure
1.4) measures the sound level as it is drawn through the soil at different depths [4].
Figure 1.4 shows that this acoustic tool has a design similar to horizontal mea-
surement tools. The difference between the ACLDD and the other horizontal devices
is that a sound transducer is employed instead of a force transducer. Tekeste?s hy-
pothesis was that the sound level would be proportional to soil compactness. The
10
Figure 1.3: Horizontal tip penetrometer probes. The items in the photo are (1) cone
tip, (2) cone rod, (3) cone blade, (4) prismatic tip, (5) prismatic rod, (6) prismatic
blade, (7) carrier bolt, (8) load cell, (9) I/O cable, (10) load cell housing, (11) housing
bolt, (12) nut, and (13) cable protector [3]
11
Figure 1.4: Acoustic compaction layer detection device. Tine design (A) with a cone
(B) and a microphone (C) [4]
12
sound recorded by the system would be of higher amplitude when the microphone
was being pulled through more compact soil.[4]. The ACLDD is capable of detecting
the soil compaction layer effectively and inexpensively [4].
1.5.2 Three-Depth Soil Strength Measurement Devices
Strain gauge array
Adamchuk designed and tested a vertical smooth blade (VSB) device to dynam-
ically measure soil penetration resistance at three depths [5] (see Figure 1.5).
This device consists of a tapered cantilevered beam with an array of strain gauges
that cut through the soil [5]. The varying soil resistances cause deformations in the
beam that are detected by the strain gauges [5]. The available measurement depths
would be ?p1?, ?p2?, and ?p3?, as seen in Figure 1.5.
Instrumented Subsoiler
Manor et al developed an instrumented subsoiler to map hardpans [6]. This
device, shown in Figure 1.6 is also capable of soil strength measurements at three
depths.
This system works with three load cells and a potentiometric sensor that mea-
sures the resultant direction and magnitude of the forces acting on the shank [6]. The
resultant forces would be different depending on its depth relative to the hardpan [6].
This system has the hydraulic actuator so that the depth can be adjusted.
13
Figure 1.5: Schematic of the vertical smooth blade (edge is on the left side) [5]
14
Figure 1.6: Instrumented Subsoiler [6]
Instantaneous Multiple-Depth Sensor
ChukwuandBowersdevelopedathree-depthcontinuoussoilmechanicalimpedance
sensor (seen in Figure 1.7) [7]. This design uses three load cells mounted on a blade
carrier. These load cells measure the force applied to the penetrometer tips [7].
1.5.3 SoilStrengthMeasurementDevicesThatMeasureMoreThanThree
Depths
Load Cell Array
S. O. Chung et al developed an on-the-go soil strength profile sensor using the
load cell array (OTG-SSPS). This device, seen in Figure 1.8 is capable of measuring
soil strength at five depths. This device consisted of multiple sensing tips attached
15
Figure 1.7: The three-depth soil mechanical impedance sensor designed by Chukwu
and Bowers [7]
16
Figure 1.8: Operational concept of the soil strength profile sensor (OTG-SSPS): [8]
to load cells. Each sensing tip gives a soil strength measurement. The OTG-SSPS
obtained continuous ?cone index like? measurements at discrete depths as it was
pulled through the soil [8].
17
Figure 1.9: Profile Sensor mounted on a tractor [9]
Profile Sensor
P. Andrade et al developed a profile sensor (PS) which consisted of eight cutting
edges supported by load cells [21]. This device, seen in Figure 1.9, was designed to
provide information on soil resistance at eight depths [21].
The PS measures forces on load cells located inside the shank [9]. Based on the
instrumented tine developed by Glancey et al, [22] and the texture/compaction index
(TCI) , the profile sensor measures force on the eight cutting edges as the tine is
pulled through the soil [21].
18
All of these horizontal strength measurement devices are capable of hardpan
detection at multiple, different depths but they all have one inherent shortcoming.
Their common flaw is their discrete depth measurement [14]. These tools are designed
to be inserted in the soil and pulled at a constant depth. Therefore, the measure-
ment depths are constant while the hardpan depth is not. When the actual hardpan
depth is different from the sensor depth, erroneous or missing soil compaction depth
measurements become a problem [14]. If the placement of the sensors is not close
enough to the hardpan, it could go undetected as it would fall between two sen-
sors [14]. In other words, the sensor must make contact with hardpan to give correct
measurements. If the sensor does not make contact, the hardpan remains undetected.
1.6 Introduction of the On-The-Go Soil Strength Sensor
Vertical penetrometers are capable of continuous vertical cone index measure-
ment but they are limited to discrete horizontal (across the field) samples. The
horizontal penetrometers studied to date are capable of continuous horizontal cone
index measurements but are limited to discrete vertical depth measurements. E. Hall
and R. Raper proposed the idea of an on-the-go horizontal soil strength measure-
ment system that oscillates vertically as it is pulled through the soil[14]. With this
device, continuous depth soil cone index measurements can be achieved by using a
single vertically oscillating sensor [14]. Like the other horizontal soil strength mea-
surement devices, the On-The-Go Soil Strength Sensor (OSSS) includes a downwardly
extending shank with a sensing tip mounted on the leading edge [23]. The difference
19
between this device and the other horizontal soil strength measurement devices is its
reciprocating motion. This horizontal motion would allow continuous variable-depth
soil strength profile measurements over a large area [23]. The reciprocation motion of
the OSSS works similarly to a sewing machine. The hydraulic valve (motor) controls
the shank?s (needle?s) vertical movement. The ground (fabric) moves horizontally
relative to the shank. This motion causes the sensor (eye) to move through the soil
in a sinusoidal motion. The horizontal motion of the device being pulled through the
soil combined with a vertical oscillation enables a continuous sinusoidal (vertical and
horizontal) measurement of the mechanical impedance throughout the profile [20].
The OSSS concept was tested in the soil bins at the USDA National Soil Dynam-
ics Laboratory at Auburn University in Auburn, Alabama. The current need of the
OSSS project is to design and develop the automatic control system and hardware.
The control system serves as the brains of the system and controls the shank oscilla-
tion. The hardware of the system consists of the tool bar on which the OSSS rests,
the hydraulic system that powers the OSSS, and the electrical components for signal
conditioning. With the development of the control system and hardware components,
the work of E. Hall and R. Raper can come to fruition.
20
Chapter 2
Control System Modeling
2.1 Introduction
The speed and range of the On-the-fly Soil Strength Sensor (OSSS) data collect-
ing capabilities are limited by the hydraulic system (particularly the slew rate of the
rod extension) that actuates its motion. However, the OSSS operator would need
to transverse the path at a reasonable speed and collect sufficient data to determine
hardpan depths thoroughly throughout the field. To assist with determining the lim-
its of speed at which the operator can travel and collect data, a system model and
simulation are studied using MATLAB & SIMULINK.
2.1.1 Typical Motion Profiles
Figure 2.1 shows a diagram of how the OSSS should oscillate. The space between
the white vertical lines is the distance for each soil strength profile. If the distance
between the white lines is 10 meters then the cycle rate is 10 m per cycle. The
large amplitude sinusoidal line is the vertical position of the soil strength sensor when
the OSSS is pulled through the soil. The semi-straight horizontal line in the middle
represents the hardpan.
A LabVIEW software program controls a power amplifier which, in turn, drives
the proportional directional valve to produce the vertical oscillation of the tine. The
21
Figure 2.1: The OSSS cycles once for every two segments. Vertical white lines repre-
sent segments.
tine is set to cycle at specified horizontal distances (e.g. 15 horizontal feet per cycle)
regardless of the tractor speed. If the tractor travels too fast, a sinusoidal motion
with the desired amplitude is unachievable. One method to achieve increased tractor
speed would be to restrict the depth range to those depths that might actually contain
the hardpan instead of allowing the full range of the hydraulic cylinder to be used.
For example, peaks in soil strength typically occur at depths approximately 20-40
cm beneath the soil surface [11]. Therefore, the measurements could be restricted
to this range. This smaller oscillation range would allow the system to cycle faster.
However, restricting the measurements to the hardpan would ignore deeper rooting
restrictions that could be 50 cm or deeper.
22
Another option could be to extend the cycle distance. The problem with this
approach is the tremendous variation in site-specific hardpan depth. Using descriptive
semivariogram statistics, Raper et al [11] found variations in the depth to hardpan as
close as 12.4 m in some Southern U.S. fields. Therefore, if the cycle rate is too low,
a good map of the hardpan can not be created from the data. What is needed is a
three-dimensional plot of speed versus cycle rate versus cycle range that would allow
one to determine the limits of the system.
2.2 Description of proposed hardware
The OSSS unit consists of three components: a sensing tip, a shank, and a force
transducer. The OSSS shank was designed to provide a method of inserting the force
transducer and the soil strength sensing tips into the soil. The shank was designed
to be pulled at a perpendicular rake angle to the soil surface at a maximum effective
measuring depth of 600 mm.
The control program was written in LabVIEW 1 to control the cycle rate and
range with a command signal realized by National Instrument data acquisition card
(model DAQ 6025E). This data acquisition card is capable of producing a 10 volt
analog signal which is amplified by the PID module (Vickers model EEA-PAM-561-
D-32).
GPS navigation is used to calculate the tractor speed as well as the geographical
position of the tine. A depth sensor measures the relative distance of the OSSS to the
1LabVIEW is a product of National Instruments
23
soil surface and a draw-wire displacement sensor (Micro-Epsilon model WDS-Z100)
measures the position of the oscillating tine relative to the OSSS frame. The force
transducer that measures the tillage force necessary to push the OSSS through the
soil is the SENSOTECr?2 GR3 load beam (SENSOTECr?, Columbus, Ohio 43228),
with a 4.45-kN measurement capacity. The GR3 load beam is a cantilever beam
design, capable of measuring tensile and compressive loads. The information from
the force transducer is stored with the DGPS (Differential Global Positioning System)
data in a form that can be used to draw 3D maps of the hardpan for the entire field
post data collection.
2.3 Model description
To ensure that the system could perform as needed, a model was constructed in
SIMULINK (see Figure 2.2).
2.3.1 Modeling the sensors and other signals
The ?Draw-Wire Sensor? block in Figure 2.2 was modeled as a gain block with
the minimum displacement of the stroke being 0 volts and the fully extended stroke
represented as 10 volts. This sensor has a gain of 9.4 v/m and is used to measure
the actual extension of the rod. Moving clockwise around the diagram, there is
2Use of company names or trade names does not imply endorsement by USDA-ARS or Auburn
University.
24
Figure 2.2: SIMULINK block diagram model of the OSSS
25
the summing junction, which represents subtracting the voltage of the displacement
sensor from the command signal; the difference is the error signal.
The command signal may be modeled as a sine wave, a chirp (frequency increas-
ing over time) signal, or a step function. The sine wave has adjustable amplitude and
frequency that are calculated from the parameters in a separate MATLAB m-file.
The parameters used to determine the amplitude and frequency of the sine wave are
the measurement range and the tractor speed cycle rate ratio, respectively. The chirp
signal is a sinusoidal wave with amplitude equal to 1 and a frequency that increases
over time. The amplitude of the chirp signal is scaled up and biased so that the
output is 0-10 volts. The step function consists of two steps. The first step at time
0 s is from 0 volts to 10 volts and the second step at time 10 s is from 10 volts to 0
volts. This choice of input signal is made with a ?Multi-port Switch? block.
The reference manual used ?percent of voltage? instead of voltage in the com-
mand signal to flow rate chart [24]. Therefore, the error signal is converted to a
percent of command signal using a gain block labeled ?voltage to %? such that 100%
represents 10 volts. A choice is made whether the error signal is deadband compen-
sated or not with a manual ?switch? in the simulation diagram. This conditioned error
signal flows to the block labeled ?Valve Delay?. Valve characteristics are discussed
below.
26
2.3.2 Modeling the valve characteristic
The ?Valve Delay? block represents the inherent delay characteristics of the
solenoid within the valve. The controller output (control signal) goes through a
look up table, where it is converted to a fluid flow rate (q). A lookup table for the
command signal versus flow rate was tabulated from the information given in the
valve manual (Vickers model GB-C-2007C). A graph digitizer was used to sample
the valve characteristic curve - these samples were not uniformly spaced, so linear
interpolation is not straightforward. Then, the non-uniformly sampled table was
converted to an equation using the Microsoft Excel trend line function. The equation
was finally converted to a lookup table of uniformly spaced values in the MATLAB
m-file. The lookup table labeled ?v-q? converts the control signal percentage to flow
rate.
2.3.3 Modeling the hydraulic system
The tractor hydraulic flow rate of the John Deere 6410 is limited to 113 l/m (30
g/m) and this limit is seen next in the block labeled ?Flow Rate Limit.? The units of
flow rate are converted to m3/s. Flow rate divided by the area of the cylinder (m2)
yields the rod velocity (m/s).
A hydraulic cylinder with a 6.35 cm (2.5 in) bore, a 3.493 cm (1.375 in) piston
and a 106.68 cm (42 in) stroke was used for model. These cylinder dimensions are
of the Chief model WP #286-242 which is used on the OSSS. The rod and piston
areas are modeled in two separate gain blocks labeled ?Rod? and ?Piston.? The sign
27
of the fluid flow triggers a switch that directs flow through the piston gain block or
the rod gain block. The output of the ?Flow Direction? block is the velocity of the
rod actuation. This velocity is integrated over time in the integrator block labeled
?Velocity to Position? to give rod position. The rod position is graphed in the scope
labeled ?Position?.
2.4 Simulation Results
2.4.1 Step Response
The response of the rod as a result of a step function is shown in Figure 2.3.
Ideally, the 10 volt signal would produce maximum extension. The time to extend
the rod fully could be calculated with the basic hydraulic equation 2.1.
t = lv = lq
a
(2.1)
where v is velocity of the extending rod, q is fluid flow rate, a is cylinder area, t is
time, and l is stroke length. This equation does not, however, take into account the
deadband or the valve delay. In the SIMULINK model, the deadband and the valve
delay (see Equation 2.2) are accounted for, thus allowing a better determination of
the actual time the system takes to complete one cycle.
ValveDelay = 1[tau,1] (2.2)
28
Figure 2.3: Plots of the step response. Blue line plot is of the system with deadband.
The red (+) plot is of system with deadband compensation. Position as a result of
two step functions. First step a time 0 from 0 volts to 10 volts and the second step
function at time 20 seconds from 10 volts to 0 volts with valve delay.
Figure 2.3 shows the response of the system with unity gain and the step function
input as described above. This figure shows the system response with no deadband
affect in the red ?+? plot. The effects of the deadband are shown in the blue line
plot. With a step from 0 to 10 volts we would like to see the rod position extend from
0 to 100%. This is not the case with the blue line plot because of the deadband.
29
2.4.2 Deadband
A plot of the command signal to flow rate chart is shown in Figure 2.4 represented
by the solid blue line. The flat region of this plot is the deadband. The deadband is the
range of control signal that will produce a flow rate of zero. This deadband is useful
in systems with appreciable noise because it will prevent noise induced movement
of the actuator. However, because valve deadband is evident with small signals, the
dead zone affects the system negatively when short rod movements are desired. In
fact, rod motion halts whenever the control signal is too small, since the commanded
control effort falls inside the valve deadband. In the simulation plot in Figure 2.5 it
can be seen that a deadband compensator would improve the response.
There is a compensator onboard the amplifier that can be adjusted with a po-
tentiometer. However, since the onboard compensator should not reduce the valve
deadband to zero (because of noise), there will still be a dead zone. This dead zone
is considerably smaller than the blue line plot in Figure 2.4. To further improve the
system?s small signal response a compensator was programmed in the control soft-
ware. This compensator will adjust the output signal of the DAQ card causing the
command signal to ?hop? over the deadband region. The only time the command
signal will be found in the dead zone is when the desired movement of the cylinder
is zero. Then the command signal and the flow rate will be zero. A plot of the
compensated signal is shown in Figure 2.4 represented by the red (+) line.
30
Figure 2.4: Deadband compensation
31
Figure 2.5: Simulation signals
32
2.4.3 Sinusoidal response
Figure 2.5 consists of four graphs: command (input) signal; conditioned error
signal; flow rate; and rod position. Each graph consists of two plots. The blue (solid)
line shows the effects of the deadband while the red/dashed line shows a deadband
compensated system. The input signal is the same for both the compensated and
the uncompensated system. For the sake of simplification the input signal is scaled
up to represent the desired range of motion (20-40cm). With the uncompensated
system (graph B blue/solid line) it is seen that the signal controlling the flow rate is
in the deadband region most of the time. Graph ?C? (blue/solid) shows that while
our control signal is in the deadband range there is no flow to the rod. And the
effects of no flow can be seen in graph ?D? (blue/solid) as the position of the rod is
unchanged. As a result, the rod will not oscillate. The blue lines in the figure show
the system with deadband compensation. Graph ?B? shows that the control signal
skips over the deadband region. As a result there is flow to the cylinder when small
movement is needed. Therefore the cylinder oscillates in small measurement ranges
as seen in graph ?D?. It is concluded that deadband compensation is not only feasible
but necessary since the 20-40 cm region is the primary 3 region.
2.4.4 Sinusoidal and Chirp input
The cylinder diameter and tractor flow rate limit are issues at high frequency
oscillations. Simulation plots reveal when the upper limits are reached: the system
3The primary region is where the hardpan is found.
33
is unable to keep up. And the response plot takes on a triangular wave form. Figure
2.6 shows this ?clipping.? In this figure the flow rate and the response are normalized
and plotted together. The blue/plus line is the response and the black line is the flow
rate. It can be seen that the flow rate is maxed out while the rod is actuated and
then crosses zero when the rod actuation direction changes. Because the velocity of
the collapsing rod is greater than the extension velocity the sinusoidal motion is seen
at the lower part of the plot. When the rod reaches 100% and changes direction a
triangle shape is seen because the cylinder never reaches 100% extension. In other
words the higher the oscillation frequency the lower the ability of the system to keep
up.
With a chirp signal input the limits of the oscillation frequency at a certain mea-
surement range can be explored. Another way of stating this is how fast the tractor
can travel since the oscillation frequency is directly proportional to the tractor speed.
With a frequency sweep from 0-1 Hz Figure 2.7 shows that the desired amplitude
can be maintained only for lower frequencies before it declines. It is seen that at 0.3
Hz the system is no longer able to produce a 100% (106 cm) response. This model
can be used to determine the maximum measurement range with a certain tractor
speed and cycle rate or the maximum tractor speed for a certain measurement range
and cycle rate.
For example, the command signal for the simulation that produced Figure 2.8
was set to produce an oscillation from 20-40 cm which is the region the hardpan is
typically found [11]. Figure 2.8 shows the frequency being swept from 0-3 Hz. It
34
Figure 2.6: Normalized output position and normalized flow rate plotted together.
The blue (+) is the response while the black line is the flow rate.
35
Figure 2.7: Position as a result of the chirp signal. The input starts a 0.0 Hz and
increases to 1.0 Hz. And full stroke.
36
Figure 2.8: Chirp signal from 0-3 Hz. And 20 cm stroke.
can be seen that the system has a favorable response until about 1.5 Hz where the
system seems to go unstable with unpredictable results. The frequency is directly
proportional to the tractor speed and indirectly proportional to the cycle rate. A
frequency of 1.5 Hz will be seen at two combinations of the speed rate ratio. A three
dimensional surface plot was created to show the limits of the system.
37
2.4.5 The plot
Using information from the simulation and Excel?s trend line function a piecewise
function (see Equation 2.3) was created. With this function we were able to calculate
the maximum range given at a certain rate and speed. A matrix similar to Figure
2.9 was created with this piecewise function and used to create the surface plots in
Figure 2.10.
y =
?
???
???
???
???
104.4 x < 0.24
2301.3?x3 ?2674.4?x2 + 853.95?x + 20.56 0.25 < x < 0.44
32.603?x?1.008 0.45 < x
(2.3)
With similar surface plots we can determine the tractor speed and cycle rate
allowable for a desired measurement range. More importantly, the piecewise function
used to create the surface plots can be programmed into the software to control the
command signal and to provide information to the operator. For example, a scientist
or farmer could ?tell? the control program that the system should cycle every 3 m
at a depth of 25-50 cm. As the operator drives the tractor at a comfortable speed
the program will control the amplitude and frequency of the tine to achieve this goal.
The software program will also warn the driver when the tractor speed exceeds the
limits and makes that goal unachievable.
38
Figure 2.9: Measurement range given tractor speed and cycle rate.
2.5 Conclusions
The OSSS system was modeled using Mathworks MATLAB and SIMULINK. The
dimensions of the hydraulic cylinder and the values of the amplifier parameters were
used to simulate the response of the OSSS system. Through a simulation study, we
have come to the conclusion that two issues must be considered in the control system:
? limited flow capacity from the tractor - limits the slew rate and bandwidth of
the OSSS.
? valve deadband - affects positioning accuracy at low signal levels.
39
Figure 2.10: Three dimensional plot of system limits
40
Chapter 3
Control System Design and Implementation
3.1 Introduction
In order for the OSSS to work it needs a control system which will drive the OSSS
and make it perform its oscillation task. This closed-loop control system consists of
a computer program, an amplifier, an A/D converter, and a displacement sensor.
One of the goals of the design is to minimize human interaction, thus freeing
the operator to drive the tractor. Therefore, a control program was designed to
handle the input/output signal processing. The control program controls shank/tine
oscillation and data collection according to predetermined inputs from the operator
and the tractor speed. The controller is designed and constructed to run on a laptop
computer using LabVIEW version 7.1. The simple block diagram of the control
scheme is shown in Figure 3.1.
3.2 The Block Diagram Flow
The ?command? signal is generated in the control software and sent to the Pro-
portional, Integral, and Derivative (PID) Module via an electronic data acquisition
card and the printed circuit field connector board. A signal from the PID module
controls the solenoid valve (a part of the ?plant? block). The system receives shank
41
Figure 3.1: Block diagram of the control system
position feedback from the draw wire sensor. The PID controller compares the mea-
sured shank position ?mp4? against the ?command,? and automatically produces the
control signal ?d4?. The PID module produces a control signal based upon three com-
ponents: instantaneous error, rate of change of feedback, and error accumulated over
time. The relative weight to each component is to be determined in future work, and
depends upon the dynamic capabilities and performance requirements of the OSSS
system. The Flow chart for the control system is shown in Figure 3.2.
3.3 Command Signal Output
3.3.1 DGPS to Calculate Oscillation Frequency
When the system is in run mode, DGPS (Differential Global Positioning System)
information is read through the serial port on the computer. This information is used
42
Figure 3.2: Flow Chart
43
to calculate the shank oscillation frequency, using the tractor speed. The DGPS
transmits signals according to the National Marine Electronics Association standard
(NMEA 0183) in the form of hexadecimal strings via serial RS232 cable. A portion
of the control system is designed to read the NMEA information from the serial port
and rebuild the NMEA sentences received from the DGPS receiver. The ?NMEA
read? algorithm reads one hexadecimal byte from the serial input and compares the
values to standard NMEA tokens. Each NMEA string begins with a ?$GP? (24
hex) and every time this symbol is found, a new line is created. The next bytes
(separated by commas) are read and stored on this line until another ?$GP? is found.
The next set of information indicates which NMEA string is being read. Although
there are several NMEA strings to choose from, the RMC (recommended minimum)
string has latitude, longitude, and speed information and is the only one used by the
software program. The latitude and longitude information will be stored with the
soil strength data and the speed is used for the oscillation frequency calculation. The
DGPS algorithm looks for the RMC string alone, and the rest are ignored. The RMC
information is parsed out of the rest of the NMEA information. The speed is sent to
the frequency calculation algorithm and the coordinate information is stored to a file
with the data.
3.3.2 Command Voltage
This command signal is an incremented sinusoidal voltage calculated from the
horizontal distance traveled by the tractor and the cycle rate.
44
Figure 3.3: Algorithm to collect RMC sentence
45
Figure 3.4: Algorithm to parse RMC in LabVIEW
46
Figure 2.1 shows how OSSS will oscillate. Using the speed information from
the DGPS, the command signal voltage is recalculated for the shank movement.
The algorithm that collects the speed information runs at about 0.5 Hz so there
is a 2 second delay between each speed adjustment. The speed is divided by the
time (recorded by the computer) since the last speed adjustment to get the distance
traveled. The equation used to calculate the voltage output is:
VS = R?sin(pi ?(DT ?CR?floorleftDT/CRfloorright)) + Z (3.1)
Here VS is the command voltage, DT is the distance traveled by the tractor, and
CR is the cycle rate which is the distance traveled by the tractor per shank oscillation
set by the operator. R is the measurement range set by the operator, and Z is the
depth of measurement set by the operator. Once the command signal is calculated it
is routed to the amplifier through the data acquisition card.
3.3.3 System Startup and Configuration
Under normal circumstances the operator will control only power up, start, stop
and shutdown while the control program handles all major tasks associated with each
command. For the operator these are one button commands requiring minimal con-
centration. The software program controls toggling of individual digital channels that
control the PID module, the analog signal that positions the shank and everything
47
Figure 3.5: Command signal calculator in LabVIEW
48
else including system monitoring, oscillation frequency, and data collection. The user
interface for the OSSS system is shown in Figure 3.6.
This screen shot shows what the operator sees while using the OSSS system. On
the right side of the user interface is the configuration section, where the measurement
configurations are set by the operator. In the configuration section, operating test
parameters can be employed to determine if the system is functioning properly. The
section to the left is designed to be similar to the front panel of the Vickers amplifier.
Most status LED?s on the amplifier can be read on the user interface. Most of
the information from the monitoring points on the front panel is displayed for the
operator. The PID, ramps, and deadband adjustment points can not be controlled
from the control program and are not available on the user interface. The command
signal gauge in the middle of the simulated amplifier is not available on the actual
amplifier. This two marker gauge shows the command signal output with one marker
and the displacement sensor input with the other. The OSSS system is enabled from
this section of the user interface.
At the start of the system, the operator starts the OSSS with the ?Power Up?
button on the user interface. After the operator powers up the system he/she waits
for the ?Initialization Progress? to complete. In the mean time, the system runs
the initiation sequence and checks necessary parameters for the system. The PID
module is turned on, followed by a delay before the drive and the PID are enabled.
When the system becomes stable the integrator is enabled1. Immediately after the
1This delay before the integrator is enabled prevents pressure spikes in the hydraulic system.
49
Figure 3.6: OSSS?s user interface
50
initialization progress completion, the OSSS system goes into a ready/standby mode.
In this ?ready? mode the analog and digital channels as well as the DGPS signals
are monitored. This information is available to the operator on the front panel of the
control program.
While in the ?ready? mode the measurement configuration is to be set. The
measurement configuration section located close to the middle of the user interface was
designed to be as simple as possible. The values set in the measurement configuration
determine the depth, range and frequency of the shank oscillation. The parameters
to be set are:
? Height-the distance from the OSSS sensor to the soil surface while the OSSS
cylinder is fully retracted. Units are in cm.
? Cycle distance-the distance traveled by the tractor for OSSS oscillation
? Depth-the average distance of the hardpan beneath the soil surface
? Range-the amplitude of the OSSS oscillation
The vertical fill slide located to the right of the configuration section is a display
of the measurement settings. This visual aid allows the operator to see how the
measurement parameters will affect the shanks motion. In the configuration mode
the operator can see if the shank will travel out of the soil or if the measurement
settings make sense. When the OSSS is oscillating the fill slide will show the OSSS
shank position.
51
3.3.4 Operation Mode
After the system has been powered-up and the measurement configuration have
been set, the operator is ready to start collecting data. The data collection sequence is
started when the ?Oscillation? and ?Write? switches are turned on. The ?Oscillation?
toggle switch in the up position allows the command voltage to be updated and sent
to the amplifier. At the end of a pass the operator will stop the system using the
?Oscillate? switch. When the ?Oscillate? switch is down the system will stop. In the
stop sequence, the shank is raised out of the soil in preparation for any turning of the
tractor and the data storage is halted. The system will return to the ready mode and
await the next command. The ?Shutdown? button is employed when the OSSS is to
be turned off. The shutdown button also works as the emergency stop if it is activated
before the ?Oscillate? switch is toggled down. When the shutdown command is given,
the program will send a signal to raise the shank and the PID module will be turned
off. After the shutdown/emergency stop button is employed, the system will have
to go through the power up sequence before the start/data collection button can be
read.
The operation mode consists of two sequences: data collection and command
signal output. In the data collection sequence information from the digital inputs are
monitored for alerts and warnings from the PID module. Analog data is collected from
the circuit board, which includes PID module information and sensor information.
The DGPS data is collected in this sequence as well. If the system is on standby mode,
this information is only displayed on the front panel. When the start command is
52
activated, the information from the data collection sequence is used to calculate the
command signal in the command signal output sequence. When the tractor stops the
shank will stop and hold its position until either the tractor moves again or the stop
sequence is initiated.
3.4 Deadband Compensation
Once the OSSS system was constructed and operational the issue of the deadband
was addressed. Figure 3.7 shows the response of the system with factory set dead
zone. It can be seen in Figure 3.7 that there is a delay in the response before the
shank motion changes direction. This delay is the result of an excessive dead zone.
On the front panel of the amplifier there are two potentiometers used to adjust the
deadband. Figure 3.8 shows the response of the system after the dead zone was
reduced. With the control system and the frame built the system can be tested.
53
Figure 3.7: System with factory set deadband compensation. The solid line shows
the input to the system. The dotted line shows the output movement of the shank
with no deadband compensation in the amplifier.
54
Figure 3.8: System with deadband compensation adjustment in place. The solid line
shows the input to the system. The dotted line shows the output movement of the
shank with deadband compensation in the amplifier.
55
Chapter 4
Experimental Evaluation
The final step in the project is the system performance evaluation. To use the
OSSS system safely and efficiently, the operator must be aware that the system has
limits, as stated in chapter 2. To operate in a range beyond these limits would be
dangerous. And to operate the system too far below these limits would waste time.
The purpose of this chapter is to explore the operational limits of the physical system.
This was done through the evaluation of the frequency response for the project results.
4.1 Experimental Approach
The OSSS system was first tested in a controlled environment (see Figure 4.1)
with a stationary tractor. The OSSS shank was not attached to the OSSS frame,
allowing the hydraulic cylinder to travel freely. In order to test the system response,
a set of frequencies ranging from .01 Hz to .5 Hz was used. To realize these frequencies
the simulated speed was incrementally increased over time. The shank position was
recorded with the command signal output, the oscillation frequency, and the PID
controller output in an Excel spreadsheet. This information was used to evaluate the
system.
56
Figure 4.1: OSSS Mechanics Attached to Tractor
57
Figure 4.2: G(s) (Open Loop Magnitude Response)
4.1.1 Step Response Tests
A step response is a typical experiment for this type of system. A step response
experiment was attempted but the hydraulic pressure spike was too much for the
hydraulic system to handle. The O-rings on the hydraulic connection could not
withstand the pressure spike and ruptured during the test process. Therefore, no
step response information is presented.
58
Figure 4.3: G(s) Open Loop response.
59
4.1.2 Plant Frequency Response
In this work, the ?plant? is considered to be that part of the feedback loop ly-
ing between the PID controller output and sensor output. The output of the PID
controller is the input to the plant, which is denoted G(s). A plot of the experimen-
tally measured frequency response G(?) can be seen in Figure 4.2. An approximate
transfer function model fitted to the model would be first order, with a bandwidth
approximately 0.03 Hz.
G(s) = Y(s)U(s) (4.1)
4.1.3 PID Controller Response
A plot of the PID output signal and the sensor input can be seen in Figure 4.3.
The noisy line in the sensor and the other line is the PID output. Much of the noise
in the sensor signal is due to the jerkiness of the rapid opening and closing of the
valve. At higher frequencies the cylinder is trying to keep up with the command
signal, therefore, the fluid through the cylinder is allowed to flow at rated flow rate
and the vibrations are greatly reduced.
4.1.4 Closed Loop Response
The closed loop transfer function is given by (4.2)
H(s) = Y(s)R(s) = C(s)G(s)1 + C(s)G(s)
60
Figure 4.4: Closed loop frequency response (Simulated)
The simulated closed loop frequency response is shown in Figure 4.4. The experimen-
tally measured response is shown in Figure 4.5, and it matches the simulated response
quite well. The closed loop system bandwidth is approximately 10 times higher than
the open loop bandwidth. For reference purposes, the measured phase response of
the system is shown in Figure 4.5. The phase response can be used to predict the
delay between the commanded position and actual position of the OSSS.
61
Figure 4.5: Closed loop frequency response (Measured)
62
4.1.5 OSSS In The Field
An experiment was conducted to test the OSSS in a field environment. The
oscillating frequency was set at 0.272727 Hz (period of 3.667 seconds). The OSSS was
pulled for about about 30s under closed loop control with the tine oscillating beneath
the soil surface. Figure 4.6 shows some the information from that was acquired from
that field test. The draw wire sensor output has a voltage bias that can be subtracted
out. The phase difference between the reference and output is consistent with the
experimentally measured frequency response (see Fig. 4.5).
63
Figure 4.6: Closed Loop Response (collected in the field)
64
Chapter 5
Conclusions and Future Work
5.1 Summary
The focus of this thesis was the design implementation and evaluation of the
automatic depth control system for the on-the-go method of sensing soil strength
(OSSS). The software was designed and implemented with LabVIEW 7.1 to drive a
PID controller for a hydraulic cylinder. This system oscillates a shank in the soil
at consistently spaced intervals regardless of tractor speed variations. The control
system and the hardware have been designed for the OSSS.
5.2 Future Work
A controller has been designed and built to operate the OSSS. This controller will
maintain the amplitude and frequency necessary to measure the soil strength profile.
The control system is designed and the integration of the software and hardware
are complete and verified. Future work could include determination of optimal PID
controller tunings, and completion of the data collection algorithm.
Testing reveals that the closed loop bandwidth of the system is limited by the
flow capacity of the hydraulic system. Higher bandwidth (the ability to oscillate more
quickly) requires that the hydraulic cylinder move more quickly. A smaller diameter
cylinder would yield faster response, at the expense of decreased force available to
65
lower or raise the OSSS. An alternative would be to increase the hydraulic pump
capacity, but that may likely require significant modification to the tractor.
66
Appendices
67
Appendix A
The Printed Circuit Board
A.1 37 Pin Connector
A.2 9 Pin Connector
68
Figure A.1: Printed Circuit Board
69
Table A.1: 37 Pin Connector
BD PIN# CONN PIN# SIGNAL TERMINAL
1 19 AIGND DAQ.69
2 18 COMMAND SIGNAL AMP.B6
3 17 ACH9 DAQ.66
4 16 ACH10 DAQ.31
5 15 ACH11 DAQ.63
6 14 ACH12 DAQ.61
7 13 ACH13 DAQ.26
8 12 ACH14 DAQ.58
9 11 ACH15 DAQ.23
10 10 SOLENOID AMP.Z28
11 9 SOLENOID AMP.Z26
12 8 -10 REF AMP.B2
13 7 +10 REF AMP.Z2
14 6 5B MODULE CHB DAQ.33
15 5 5B MODULE CHA dup
16 4 AIGND DAQ.68
17 3 DGND DAQ.13
18 2 +5 0 DAQ.14
19 1 AOGRND DAQ.54
20 37 AIGND DAQ.69
21 36 +5 1 DAQ.8
22 35 AIGND NONE
23 34 SNESOR DAQ.57, AMP.D2
24 33 GROUND GROUND
25 32 24v Vcc
26 31 AISENCE DAQ.62
27 30 CURRENT COMMAND AMP.Z6
29 28 SENSOR mA AMP.D6
31 26 INVERTED COMMAND INPUT AMP.Z10
28,30,32,36 29,27,25,21 GROUND UNCONNECTED
33 24 DV1 AMP.B8
34 23 DV2 AMP.Z8
35 22 DV3 AMP.B10
37 19 UNCONNECTED UNCONNECTED
70
Table A.2: 9 Pin Connector
PIN# SIGNAL TERMINAL
1 SOLENOID AMP.Z28
2 SOLENOID AMP.Z26
3 LVDT PIN1 AMP.B14
4 LVDT PIN2 AMP.Z22
5 LVDT PIN3 AMP.B16
6 GROUND GROUND
7 GROUND GROUND
8 LVDT PIN2 TEST.2 1
9 LVDT PIN3 AMP.Z22
71
Appendix B
The Frame
72
Figure B.1: Frame1
73
Figure B.2: Frame2
74
Bibliography
[1] ?ASAE standard,? S313.3 Cone Penetrometer. St. Joseph, MI, 2004.
[2] R. L. Raper, B. H. Washington, and J. D. Jarrell, ?A tractor-mounted multiple-
probe soil cone penetrometer,? Applied Engineering in Agriculture, vol. 15, no. 4,
pp. 287?290, April 1999.
[3] T. Alihamsyah and J. E. G. Humphries, C. G. Bowers, ?A technique for
horizontal measurement of soil mechanical impedance,? American Society of
Agricultural Engineers, vol. 33, no. 1, pp. 73?77, January 1990.
[4] Acoustic Compaction Layer Detection, no. 021089, Chicago, Illinois, USA, July
2002.
[5] V. I. Adamchuk, M. T. Morgan, and H. Sumali, ?Applications of a strain gauge
array to estimate soil mechanical impedance on-the-go,? Transactions of the
ASAE, vol. 44, no. 6, pp. 1377?1383, 2001.
[6] G. Manor and R. L. Clark, ?Development of an instrumented subsoiler to
map soil hard-pans and real-time control of subsoiler depth,? in ASAE Annual
International Meeting. Sacramento, CA: ASAE, July 2001.
[7] E. Chukwu and C. G. Bowers, ?Instantaneous multiple depth soil mechanical
impedance sensing from a moving vehicle,? Transactions of the ASAE, vol. 48,
no. 3, pp. 885?894, 2005.
[8] On-the-go Soil Strength Profile Sensor Using a Load Cell Array, no. 031071. Las
Vegas, NV: ASAE, July 2003.
[9] P. Andrade, S. Upadhyaya, B. Jenkins, and A. G. S. Filho, ?Evaluation of the
uc davis compaction profile sensor,? in ASAE Annual International Meeting /
CIGR XVth World Congress. Chicago, IL: ASAE, July 2002.
[10] A. Abou-Zeid, R. L. Kuchwaha, and D. S. D. Stilling, ?Distributed soil displace-
ment associated with surface loading,? in ASAE Annual International Meeting.
Las Vegas, NV: ASAE, July 2003.
75
[11] R. L. Raper, E. B. Schwab, and S. M. Dabney, ?Measurement and variation of
site-specific hardpans,? American Society of Agricultural Engineers, Tech. Rep.
01-1008, July 2001.
[12] L. G. Wells, T. S. Stombaugh, and S. A. Shearer, ?Crop yield repsonce to pre-
cision deep tillage,? in 2003 ASAE Annual International Meeting, no. 031083.
Las Vegas, Nevada, USA: ASAE, July 2003.
[13] R. L. Raper, D. W. Reeves, J. Shaw, E. van Santen, and P. Mask, ?Effect of
site-specific tillage on draft requirements and cotton yield,? in Beltwide Cotton
Conferences, January 2003, pp. 2025?2028.
[14] H. E. Hall and R. L. Raper, ?Development and concept evaluation of an on-the-
go soil strength measurement system,? Transactions of the ASAE, vol. 48, no. 2,
pp. 469?477, February 2005.
[15] R. L. Raper, ?Soil compaction management,? Encyclopedia of Agricultural,
Food, and Biological Engineering, pp. 902?905, August 2003.
[16] V. I. Adanchuk, A. V. Skotnikov, J. D. Speichinger, and M. F. Kocher, ?Devel-
opment of an instrumented deep-tillage implement for sensing of soil mechanical
resistance,? Transactions of the ASAE, vol. 47, no. 6, pp. 1913?1919, 2004.
[17] J. V. Perumpral, ?Cone penetrometer applications ? a review,? Transactions of
the ASAE, vol. 30, no. 4, pp. 839?944, July 1987.
[18] J. R. Williford, O. B. Wooten, and F. E. Fulgham, ?Tractor mounted field pen-
etrometer,? Transactions of the ASAE, vol. 15, pp. 226?227, 1972.
[19] R. R. Price and J. Theriot, ?Development of a frame to automatically insert a
hand-held penetrometer,? in ASAE Annual International Meeting, no. 031077.
Las Vegas, NV: ASAE, July 2003.
[20] H. E. Hall, ?Development of an on-the-fly mechanical impedance sensor and
evaluation in a coastal plains soil,? Master?s thesis, Auburn University, Auburn,
Alabama, December 2001.
[21] P. Andrade, U. Rosa, S. Upadhyaya, B. Jenkins, J. Aguera, , and M. Josiah,
?Soil profile force measurements using an instrumented tine,? in ASAE Annual
International Meeting. Sacramento, CA: ASAE, July 2001.
[22] J. L. Glancey, S. K. Upadhyaya, W. J. Chancellor, and J. W. Rumsey, ?An
instrumented chisel for the study of soil-tillage dynamics,? Soil Till. Res., vol. 14,
no. 1, pp. 1?24, 1989.
76
[23] R. L. Raper and H. E. Hall, ?Soil strength measurement for site-specific agricul-
ture,? U.S. Patent No. 6647799, 2003.
[24] Vickers, Proportional Directional and Throttle Valves GB-C-2007C, Vickers Inc.,
April 1995.
77