CA2066722A1

CA2066722A1 - Soil test apparatus

Info

Publication number: CA2066722A1
Application number: CA002066722A
Authority: CA
Inventors: Kenneth A. Sudduth; John W. Hummel; Robert C. Funk
Original assignee: Kenneth A. Sudduth; John W. Hummel; Robert C. Funk; Agmed Inc.
Current assignee: AGMED Inc
Priority date: 1989-09-22
Filing date: 1990-09-20
Publication date: 1991-03-23
Also published as: WO1991004484A1; EP0493524A4; US5038040A; EP0493524A1

Abstract

2066722 9104484 PCTABS00003 A soil test apparatus for field use comprises structure adapting the apparatus for transport over a field for testing the soil thereof; an infrared radiation generator (20) for producing infrared radiation at a plurality of predetermined wavelengths, an elongate light carrying member (24) coupled to the infrared radiation generator and extending therefrom for directing infrared radiation onto the soil; and a light detector (26) for detecting infrared radiation reflected from the soil and for producing corresponding electrical signals.

Description

~U~b l~
WO91/0448~ PCT/US9~/05374 . .;' .

. . .

~ . .

SOIL TEST APPARA~US `.
Back~round of the Invention ~.:
This invention is directed generally to the `~
field of near infrared reflectance analysis and more particularly to a novel and improved near infrared 05 reflectance sensing ~ystem for determining soil :constituents, for example, for use in agriculture or the ::l : . like.
Analysis of soil constituents is of particular :
in~erest to agriculture for optimizing conditions for : :
the raising o~ various crops. ~exetofore such analysis ~was done by taking numerous soil samples from~an area to ~ :;
e tested and subjecting the same to painstaking and time-coDsuming laboratory analy is.
We have proposed~ to greatly simplify this : 15~ process by the use of a near infrared (NIR) reflectance :.:
: : ;sensing system:sui~able for use in the fieId. :It has ~ : :
previously~been:proposed:to~use such sensing systems for other types~of:~analysis; ~or;example, for:analysis of. .--graln~constituene~ or the constituent contents of other W 0 91/04484 2 0 6 ~ 7 2 2 ~c-r/US~0/05374 bulk materials. However, in developing a system for determining soil constituents ~or in-the-field use, a number of other problems and ~actors arise which need to be addressed.
05 Among soil properties of interest are soil moisture content and cation exchange capacity (CEC).
However, perhaps of primary interest is the analysis of the organic carbon content o~ the soil. Accordingly, our sensing system is designed particularly with the analysis of organic carbon content in mind, although it might readily be adapted to analysis of such other properties as moisture content and CEC without departing from the invention. Among problems to be addressed in the design of the system were such matters as selection of design alternatives of the sample presentation mechanism, the design of the sensor and data acquisition systems and the processing and ana:lysis of the data acquired.
The primary considerations in selection of a sample presentation mechanism were control of the moisture content and surface roughness characteristics o~ the sample. Control of the samE~le moisture content was found to be possible by senslng below the soil sur~ace, where less variability in soil moisture would be encountered than at the surface. Control of the surface toughness characteristics of the sample was necessary, and we found this could be accomplished by a `-pressing, rolling, slicing, or other mechanical action. -These mechanical actions would be more easily accomplished below the soil surface, whPx~ we noted a more consistent}y ~riable soil would be found.
Subsurface sensing would also avoid any irregularities in sample characteristics due to the puddling or crusting which might occur on the soil surface.
once the need to sense a subsurface soil ~-sample was identified, three alternative means of ln ~;
situ and remote sensing were investigated: ~
... :

::.. ::. .

.. ; : , .. .: .. . ... . ,.. . . ... .... ,.. . . " . ... . ., .. . , , " . . . .. ........ .. .. . .

r~
WO()l/0448~ PCT/~S90/0537~
.~ ,j, Option 1 Transport of the soil sample to a remote sensing location while maintaininq the sample structure (for example, as in a soil core).
Option 2 Transport of a fractured soil sample to a 05 remote sensing location by an auger or similar device, followed by reconsolidation of the sample for measurement. ~
Option 3 In situ sensing of a surface prepared by some type of furrow opener. Option 2, transport of a 10 fractured soil sample, was eliminated from consideration due to several ~isadvantages. This system would have an inherent lag time, severely limiting operatins speed.
The process of soil detachment, transport, and repacking could introduce bia~ due to size and/or density sorting 15 of the soil particles. However, option 2 did have several advantages: intermittent sampling with a sample device would be possible; the sen~or optical path could be made compact; and a reflectance standard could be incorporated into the mechanism.
Option 1, transport of a aonsolidated soil sample, was considered in more detail. This concept u~ed an automated davice to extract a soil core and to position the coxe for scanning through a window in the side of the soil coring tube. A pneumatically driven `
core sampler was fabricated to test the soil coring concept in the laboratoryO The sampler used a 150 mm ætroke double-ac~ing cylinder controlled by a four-way solenoid actuated valve co~nected to a 1 MPa building air supply. A 12 V time deIay relay provided the 30 control input to the solenoid, porting air to the head ~ -of the cylinder ~or an adjustable time interval when an input signal was applied. The relay was set such that the sampler experienced a minimum dwell time at the fully extended position and then began its return stroke, with a total cycle time of 0.4 sec. A
spring-loaded pivoting break-away action was provided between the corin~ unit and a carrier subplate so that the corer could maintain posi~ion during the coring .

. .

. ; . . , , : . ~ . . , . . . . , . . . : - , . ; . , .: ., , . . .. ` :

w~l/n~X~ PCT/US90/05374 operation, while the carrier was moving with a horizontal velocity.
Three interchangeable soil coring tubes could be attached to the cylinder rod. These tubes provided a 05 range of cutting and core compaction alternatives for use in varying soil conditions. Tws of the tubes were standard equipment for a JMC soil sampler (Clements Associates, Inc., Newton, Iowa). The JMC "wet" sampling tube, intended for use in wetter or more cohesive soil 10 conditions, had a long tapered cutting bit and considerable relief from the bit diameter (17 mm) to the tube diameter. The JMC "dry" tube bit was shorter and larger in diameter (19 mm~ with less relief. These two tubes were fitted with an external sleeve which 15 contained the soil core while providing a window through which the sensor could operate. The third coring tube was fabricated from 25 mm diameter steel tubing by chamfering the lower edge to create a cutting bit. No relief was provided between the bit area and the 20 remainder of the tube.
Initial stationary tests of the coring unit were accomplished with recompacted, moist samples of Drummer Silty Clay Loam obtained at the University of Illinois Agricultural Engineering farm. No difference 25 in core quality was observed betweerl the two JMC bits, with both collecting acceptable samples. The straight coring tube did not obtain a satisfactory core in these ;;
conditions, due to excessive adhesion of the soil to the inner diameter of the tube.
Additional soil coring unit tests were carried out in the soil bin at the Deere and Company Technical Center, Moline, Illinois. The soil used was a mixture of 40~ fine river sand and 60-~ clay, with a moisture -content of 8.5 percent. Stationary and moving tests ~-35 were completed at three cone index levels, 0.5 MPa, 0.75 MPa, and 1.0 M~a. The speed limit for forming an acceptable soil core with the coring unit as tested was approximately 0.25 m/s. However, it appeared that a ':
., .- ~, .
. . .

. .: ~. '' ',' "' " ' ', ' ' ' , ,' "' '', ' ,' ', ':': , ' ' .' ' . '''' ', ,' ,', ' .' ', "'. ', ' ."' ' . ' wo ~ 4qx~ 7 ~ ~ PCT/US9~/05374 .,,, more refined method of holding the coring unit stationary relative to the soil surface while sampling could increase the speed operating range. Only the straight bore tube produced acceptable cores in the soil 05 bin tests, and then only with marginal reliability.
Small differences in soil moisture or cone index level resulted, on occasion, in incomplete cores being o~tained. Cores collec~ed in this high sand content, low cohesion soil with the JNC tu~es fell apart easily.
Based upon the difficulties in obtaining a complete soil core reliably across a range of soil types and physical conditions, the core sampler method of sample presentation was eliminated from further consideration.
Because of the problems encountered with the remote sample presentation methods described above, it was decided to pursue in situ sensing. This method had disadvantages in difficulty of re~lectance calibration and inability to hold the sample stationary while data ~ i were being acquired, but it seemed to hold the best promise for development of a workable prototype ~ield sensor.

- Oble~cts and SummarY of the Invention Accordingly, it is a general object of this invention to provide a novel and improved soil analysis apparatus for field use.
Brie~ly, and in accordance with the foregoing object and other considerations, a soil test apparatus for field use in accordance with the invention comprises ~`
means for adapting said apparatus for transport over a field for testing the soil thereof; infrared radiation generating means for producing infrared radiation at a plurality o~ predetermined wavelengths; light carrying 35 means coupled to said infrared radiation generating -means and extending therefrom for directing infrared !`.
radiation onto the soil; and light detecting means for WO~ )44~ PCT/US90/0537~ ~
~'

2 0 ~ 6 ~ 2 2 - 6-detecting infrared radiation reflected from ~he soil and for producing corresponding electrical signals.

Brief Description o~ the_Drawinqs 05 The features of the pr~sent invention which are believed to be novel are set forth with particularity in the appended claims. The organization and manner of the operation of the invention, together with the further objects and advantages thereof may best be understood by reference to the following description, taken in ~onnection with the accompanying drawing in which like reference numerals identify like elements, and in which:
Fig. 1 is a side elevation~ partially broken away and somewha~ diagrammatic in form, of a soil analysis test apparatus in accordance with the invention;

Fig. 2 is an enlarged, broken away alnd somewhat diagrammatic view of a sensor assem~ly or portion of the apparatus of Fig. l; ~;
Fig. 3 is a top plan ~iew of a circ~lar variable filter segment with related timing components in accordance with a preferred embodiment of the apparatus of Fig. l;

Fig. 4 is a schematic diagram of a circuit useful with -the ~iming arrangement of Fig~ 3: ~
. .
Fig. 5 is a schematic circuit diagram of a reflectance signal input circuit in accordance with a preferred embodiment illustrated herein;
Fig. 6 is a flow chart illus~rating a data acquisition -and analysis process useful with the apparatus of the -invention; ;-: ,.. -Fig. 7 is a graphical representation of a normalized peak response curve obtained with the apparatus of the invention; ~-.:
..:' ,: .
.
::: -.... " -: -- . ; ~ . . ., . . ~
:~. . . : ; , . . ,, , , . , , ,. : , . :.

wo ~1 /n44~1 0 ~; ti r~ PCl /US90~0~37 Fig. 8 is a graphical representation of the relationship between sample distance and signal level obtained with apparatus of the invention;

05 Fig. g is a graphical representation of deviation in the raw reflectance curve indicating a wavelength calibration in the operation of the processor portion of the apparatus of the invention;

Fig. 10 is flow chart illustrating a preferred mode of operati~n of the apparatus o~ the invention; and Fig. 11 is a graphical representation indicating further the process of obtaining baseline corrected data from the raw data obtained from the apparatus of the invention.
Detailed Description o~ the ~ t~,ted ~mbodiment Referring now to the drawings, and initially ::.
to Fig. 1, a near infrared (NIR) re:Electance sensing syStem is designed, fabricated, and evaiuated for ~ield use in.testing soil. The overall design objective for this soil test apparatus (e.g., for predicting organic matter content of ~oil in the fleld~ is to implement a prediction method on a near real-time basis. Specific 25 :design ob~actives for the sensor system are: -: 1) a bandp~ss of 60 nm or less over a minimum sensing range ~rom 1700 to 2420 nm, to implement the previously ':
selected sensing method; :~
: 2) an essentially continuous (in wavelength) sensing ~:
method, to allow flexibility for additional optimizatlon of the wavelengths selected for the prediction algorithm; ` .

3) Potential, with additional refinements if~needed, to --acquire enough information to make a control decision 35 every ~4.5 seconds (This rate ccrresponds to a lo m (40 : :
ft~ spacing at a 21~ m/s (5 mi/h) travel speed); ~.`; .

4) ability to predict soil organic matter content with :-;
a standard error o~ prediction less th~n 0.5 percent (or ,: . - . . .: ,": . , . ~, . , .. . . , , . "

WO~1/0448~ PCT/VS90/~374 .
less than 0.29 percent organic carbon) for the 30 Illinois 50ils in an initial, preselected calibration dataset;

5) Tolerance of dust, temperature fluctuations, shock 05 loads, and vibration, such as would be encountered if the sensor were operated in the field.
The initial choice in the design process is between purchase of a commercial NIR instrument and design and fabrication o~ a new sensing system tailored to meet the specific requirements outlined above. The commercially availa~le NIR instruments generally fall into two categories (Williams, 1987a). The grating monochromator and tilting filter instruments measure reflectance at more wavelengths than required and at a ~ `
narrower bandwidth, and this excess capability results in high costs. Additionally, these instruments are not designed for the environmental stresses imposed by use in a field environment, and would be limited to laboratory use. Fixed filter instr~ments are more rugged and less expensive, but obtain reflectance measurements at fewer wavelengths than required by the organic matter prediction algorithm. Since no oommercially available NIR instrument met all the design objectives, it was decided to develop a prototype sensor specifically targeted at NIR reflectance sensing of soil organic matter content.
Once the decision was ~ade to design a dedicated instrument, the wavelength selection mechanism and detector were chosen. These are the major components o~ the optical sys~em, and as such dictate much of the configuration of the remainder of the .. . .
system.
A review of available NIR photoelectric - - -detectors led to the choice of a lead sulfide (PbS) unit for this design. The PbS detector has advan~ages over the other available types (notably PbSe~and In~s) in the ; r, -;' areas of cost, responsivity, and abiIity to operate ~;~;
without cooling. The next deci~ion in detector ., -~ ' .' .' WO 91/044~-~ 2 ~ ~ ~ 7 ~ ~ Pcr~us9n/0s374 . . .
_9_ selection is the choice of a single element detector or an array detector. A single element detector would be used with a wavelength selection mechanism (in this case, a monochromator) which scans the wavelengths of 05 inter~st sequentially onto the detector. An array detector could be used if the wavelength selection mechanism (in this case properly termed a spectrograph, but usually called a monochromator) fscuses the wavelengths of interest into a line image on a flat focal plane, thus providing simultaneous sensing at all wav~lengths.
Three design alternatives were considered for the wavelength selection mechanism; a grating monochromator, a prism, and a circular variable interference filter. The grating monochromator is the usual device used in NIR laboratory spectrophotometers (McClure, 1987), but environmental considerations such as dust and vibration sensitivity make its use more difficult for a field instrument. Worner (1989) constructs a visible spectrophotometler using a prism and a linear array detector in an attempt to overcome the environmental problems seen with gratings. Circular variable filters have been used in rugged field instruments for portable color measurement (Jauch, 1979) and for airborne infrared spectral measurements (Hovis et al, 1967).
Using the above detector and monochromator alternatives, five possible combinations are ! .~ , identified:
1. Grating and single detector 2. Grating and array detector 3~ Prism and sin~le detector 4. Prism and array detector 5. Circular variable filter and single detector Option 1, a grating monochromator with a single detector, requires oscillation of the grating to scan a~l wavelengths of interest onto the detector. The mechanical free~om requ1red for this movement would also - ' .

W O 91/044X4 PC~r/US90/05374 2~722 --10-- .

make the device prone to vibration-induced inaccuracies.
Another possible problem with a grating instrument was dust contamination; even a small amount of dust on the surface of a grating would render it useless, thus 05 requiring its replacement. HoweverO it was felt that the monochromator could be sealed well enough that this sensitivity to dust would not be a major problem in field use.
Options 3 and 4 use a prism monochromator, however, none was available as a stock item in the wavelength range required. The costs and lead time associated with custom prism design and fabrication were not desirable within the scope of this project. Use of the prism with a single detector requires a scanning ~-mechanism and would entail the same type of vibration problems seen in the grating system. The nonlinear dispersion characteristics of a prism make it difficult to provide data at equal wavelength spacings with an array detector. Due to these problems and the fact that the prism monochromator does not have any distinct advantages over the grating monochromator if the latter - -`
could be sealed against dust, the prism monochromator options were dropped from consideration.
Primary considera~ion, then/ was between option 2, grating monochromator and array detector, and option 5, circular variable filter monochromator and single element detector. A detailed vendor sur~ey was ~-completed to idsntify the optimal stock components usable in these two alternatives. After considerati~n `~
30 of the components thus identified, the combination of ~ -circular variable filter monochromator and singla element detector was selected. ~The primary reason for this selection was the greatar flexibility offered by the circular variable filter tCVF) approach.
A CVF was available with a wavelength range of 1600 to 2900 nm and a bandwidth of approximately 55 nm (Optical Coating Laboratory, Inc., Santa Rosa, CA). ~;
This provides extra capability on either-end of the - ;
' ~';. ' , . . . . ... .~ ... , ,, . , . :, . . ., ., ,, . .,, , ,: , . . . .: , WO9l/044~ 7 ~ !~ PCT/UIS90/0~37~
' .`.1 required sensing range (1700 to 2420 nm) while meeting the 60 nm bandwidth requireme~t. Additional flexibility is realized with the CVF since reflectance readings can be taken at any desired point in the wavelength range, 05 ~ubject only to the limita~ions of the data acquisition system. In contrast, the best combination of grating monochromator and linear array detector did not completely cover the required range and allowed sensing only from 1720 to 2380 nm. This combination had a theoretical bandwidth of 36 nm, but could output only 16 reflectance readings, corresponding to the 16 elements of the linear array. Other factors favoring the choice of the CVF system are its greater tolerance of dust and vibration, and a reduced degree of complexity in the interface electronics (due to the use o~ one data channel versus 16). A possible drawhacX of the CVF
system is that it would not acquire data at all wavelengths simultaneously, but rather in sequence. It is possible that sequential wavelength scanning might require holding the soil sample stationary while data are being acquired, so that all wavelengths are scanned on an identical area of the sample.
It should be understood that all of the design alternatives mentionad herein are to be considered as falling within the scope of the invention. The particular design choices and the embodiments more particularly described hereina~ter are Por purposes of description, and should not be construed as limiting the invention in any way.
An overall schematic of the NIR soil organic matter sensor is shown in Figs. l, 2 and 3. The optical `
path of the sensor includes a broadband NIR source 20, a quarter-segment circular variable filter (CVF~
monochromator 22, a ~iber optic bundle 24 bePore the sample, and a lead sulfide de~ec~or 26 to mPasure the re~lected energy. Output of the detector 26 is conditioned by a pre-amplifier 28 and input to processing means 80 comprising a personal computer or ' .

.. . .

wo ~ I /n~4~1 PCr/US90/05374 2~6722 other suitable processor and an analog-to-digital (A/D) converter through which the data enters the processor.
A housing 25 forms the main mounting structure for this part of the instrument. Instead of using a 05 separate chopper disk, modulation of the lamp output radiation for low frequency noise and drift rejection is accomplished with the filter disk 22 itself. By spinning ~he filter disk at a sufficient rate and using the three-quarters of the filter disk which blocks the light path to perform the modulation function, the need for a separate motor, chopping disk, and sensing electronics is avoided.
To allow adjustment of the filter disX
rotation speed, a servo-controlled motor-generator 30 is used (Motomatic E-350, Robbins & Meyers, ~opkins, MN).
The permanent magnet DC motor-generator set 30 is mou~ted under the filter 22 and the filter disk 22 is attached directly to the motor-generator shaft 32. A
solid-state electronic controller and speed setpoint potentiometer 36 are attach2d to the rear of the filter housin~ 25. The system is powered through a transformer by 115 VAC. ~lthough the maximum speed of the motor-generator set is much higher, the design operation speed in this application is lO rpm or less, due to balancing considerations in the filter disk assembly.
A 50 W, 12 V quartz halogen automotive-type lamp driven by a laboratory power supply is used as the illumination source 20 for the sensor. The lamp mounting allowed three-axis adjustment for focusing and `
positioning the lamp image. A spherical biconvex lens 50 is mounted in tha upper surface of the filter housing to focus the lamp image through the input slit 48 and onto the surface of the CVF 22. The wavelength of the ;~
light which passes through the CVF at any point on the filter i5 a linear function of the angular position of that point rela~ive to the leading edge of the filt r.
Therefore to ob~ain monochromatic (or nearly "
monochromatic) light from the system, a plate or shield ., ; ',,'.'-' '.

W091/0~8~ 2 ~ s~2 ~CT/US90/05374 ;.,~,~

46 with a narrow radial slit 46 (2 mm wide by 10 mm radial length) is mounted about 5 mm above the surface of the filter. Since the projected image of the lamp filament is of a similar size and shape, only a small 05 portion of the lamp energy is blocked at the slit.
An opto-interrupter 40 is mounted within the filter housi~g such that its optical path is broken once per filter disX revolution by a small tab 42 attached to the circumference of the disk 22. The timing pulse generated by the opto-interrupter is conditioned to TTL
levels by a Schmitt trigger circuit (~ig 4). This TTL
signal is then used to compen~ate for any variations in the speed of revolution of the filter disk and to provide a positive angular position reference for wavelength determination.
Two positions for the timing tab were used for the laboratory tests. The tab position used initially was opposite the CVF segment 44 on the filter disk 22.
Later, the tab was placed adjacent to the CVF segment on the filter disk. The tab was moved so that its timing pulse would coincide more closely with the analog re~lectance signal which was generated by the photodetector 26 when the CVF was a:ligned with the slit 48.
The monochromatic light from the CVF is directed to the soil surface by means of a 610 mm Iong silica fiber optic bundle 24 with a use~ul transmission range of 350 nm to over 2400 nm, obtained from Volpi --Fiber Optics, Auburn, NY. The bundle is termed a section converter, as the fiber area changes shape from a 1 mm by 10 mm rectangular section at one end 52 to a 3.6 mm circular cross-sec~ion at the o~her end 54. The rectangular section end 52 o~ the fiber bundle 24 is mounted approximately 5 mm b~low the surface of the CVR
35~ 22 and in Iine with the input slit 48, thus collecting -~
the majority of the filter output energy. The fiber 24 then exits the bottom of the filter housing and is routed to a-light-tiqht sample chambar 60. Within the :: - ' :;

W091/044X4 PCT/US90tO537~ ~ ~
2~7~ -14-sample chamber, the circular cross-section end of the fiber optic bundle is mounted to a sensor head assembly 65. This mounting is adjustable, allowing optimization of the location of the fiber exit cone with respect to 05 the detector and the soil surface.
The sensor head assembly 65 consists of an aluminum housing 66 with a quartz aperture on window 68, -:
and the PbS detector attached thereto. Input monochromatic light from the fiber optic bundle 24 passes through the quartz aperture window 68 and illuminates a circular area on the s~mple surface 70. A --portion of the energy is di~fusely reflected from the sample and passed back through the quartz aperture and collected by the OTC-22-53 PbS photodetector 26 ~OptoElectronics, Petaluma, CA). This detector has a useful sensing range from lOOO to 3500 nm, a 3 mm by 3 mm square sensing area, and could be~ thermoelectrically cooled for increased sensitivity.
Excitation and preampli~ie!r circuitry (Fig. 5) for the PbS includes a high-gain single stag~e amplifier AC
coupled to the output of the detector, which is insensitive to low frequency drift in the detector output. A DC powex supply (not shown) provides +15 V DC
to this circuit.
.
The processing means 80, indicated diagrammatically, is a Metra~yte DAS-16 analog and digital input-output (I/O) expansion board installed in an AT-eompatible computer (such as a Texas Instruments 30 Business Pro~ running at 12 MHz. Features available on ~i the DAS-16 board include l6 single end or 8 differential an~log input channels scanned by a 12 fit successive approximation analog to-digital ~A~D3 converter, a three-channel programmakle interval timer, two channels 35~ of 12 bi~ digital-to-analog output, one 4 bit digital input port, and one 4 bit digital output port. The so~tware is provided with the DAS-16 includes a machine language driver which con~rols I/O operations by calls .~
','' .' WO ) I / 04~1X 1 2 ~ PCr/ 1 )S90/0537 . ,;

from interpreted or compiled BASIC. Drivers are available in other programming languages and the user could develop customized I/O control routines in assembly language if desired. Th~ manufacturer's stated 05 maximum sampling rate for the DAS-16 is 60 Khz, which can be achieved by direct memory access (DMA) transfer of the digitized input data to computer memory.
Interrupt driven transfers or direct data transfers to a BASIC array allows the A/D system to attain sampling rates of up to 3 KHz, according to the manufacturer.
Data collection from the soil organic matter sensor requires use of both analog and digi~al inputs on the DAS-16. One differential analog channel, configured with a + 5 V range, is used to collect data from the photodetector preampliier (Fig. 5). The TTL output ~rom the filter disk timing circuit (Fig. 4)is input as a digital signal so that the time during which its level is high dLring each disk revolution can be determined by counting gated timer pulses. ~he t:iming tab signal is also used to gate the AC c~nverter and synchronize date collection.
The required A/D sampling rate is appraximately 10 KHz, based upon the geometry of the CVF
22, a 10 rpm maximum frequency of revolution for the 25 filter disk, and a desire to obtain the reflectance data ~-on a 5 nm maximum spacin~. With this high sampling rate it was decided to transfer the A/D data directly to an array, using a program written in IBM compiled BASIC
(see Fi~. 6). This program allows analog data ~;
acquisition simultaneously with counter operation to time the width of the gate created by the rotation of the filter disk~ The data from up to ten consecutive revolutions of the filter disk can be acquired, stored on disk for later analysis, and displayed for visual verification.
To test the response of the detector, a window 92 (see Fig. 3) was placed in the filter disk directly ~ opposite the CVF segment~ This was for test purposes : '-:
, : . . .: ~. .: .. .,. . , .. - , , , ~ : : . .

W~ 44~4 ~ ~ ~ PCT/US90/OS374 .
only and forms no part of the invention. During normal operation this window is completely occluded, but for ~
detector response tests a special shutter with two -openings was installed. The first opening allows -~
05 measurement of the pulse response of the detector while simulating a pulse caused by the interaction of the slit ~ -and a point of interest on the CVF, while the second, wider opening allows the final step response of the detector to be quantified. The data acquisition system is used in this response test to record the detector output from a cPramic standard reflecting surface at filter disk periods from lOo ms to 300 ms. To avoid saturating the detector, the lamp voltage is set at 5.0 V and t~ree layers of lens tissue placed between the lamp and the focusing lens. The relationship between the 1 mm pulse response and filter disk period is shown in Fig. ~. As a compromise between the increasing signal level with longer disk period and the desire to ;
collect data as quickly as possible for field operation, a filter disk period of 200 ms (speed of 5 rpm) was selected for subsequent tests. -Another functional test investigates the relationship between sample distancel and signal le~el to ~`
determine the optimum operating distance from the soil ;~
surface. Using a ceramic standard as the reflecting surface, the loca~ion of the sensing head is varied from 614 mm to 25.4 mm above the surface and the output signal recorded.
A curvilinear relationship was found between sample distance and signal leveI (Fig. 8). The decrease in the signal with increasing sample distance indicates ~; that operation at the minimum di~tance is desirable.
Decreasing sample distance, however, will have two detrimental effects. First, in a field unit some allowance must be made to compensate for the inability to hold distance per~ectly constant. Also, decreasing the distance to the sample also decreases the area of tfie sample being sensed. For a nonuniform material such ... .
: -~i3 Jh~
WO~l/0448~ PCT/US90/0537~
. ,;
' ~!;S
~17-as soil, it is desira~le to sense a suffioiently large area to average out signal differences due to any heterogeneity found within a given sample. As a compromise between signal strength and these two 05 effects, a nominal sensor-to-sample operating distance of 15 mm was selected.
During a test designed to co~pensate for changes in illumination, detector response, and other optical system variations, each sensor reading is referenced to the reading from a ceramic disk, a substance widely accepted for standardization of NIR
instruments. Besides providing a means to calibrate for system variations, the ceramic reference also ena~les conversion o~ the response to a percent reflectance (or decimal reflectance) basis.
Two identical ceramic disks are used in this calibration procedure. One 50 mm diameter disk, mounted in a flat black aluminum block, is used as the reflectance standard for the sensor. The other disk is sent to the USDA Instrumentation and Sensing Laboratory, where its reflectance characteristics are obtained by comparison with a standard sample of slightly compressed sulfur. A series of ten paired readings of both ceramic disks is then completed with the sensor in the laboratory. The mean of these ten readings is used to compute the decimal reflectance characteristics of the ceramic disk used ~or sensor calibration.
Several data proce~sing steps are necessary to convart the raw digitized data obtained from the sensor and stored by the data collection program to the percent reflectance data needed to calibrate the sensor and predict soil organic matter content (Fig. lO). The algorithms are implemented in three BASIC programs.
CORRl.BAS reads th~ data files created by the data acquisition program, performs scan averaging, baseline corrections, and wavelength calibration, and writes an output file containing the corrected raw data.
IN~ERPl.BAS reads in raw data ftles for a soil sample ~

.~ .

W09l/0448~ PCT~US90/~374 2~722 and the corresponding ceramic reference sample, interpolates these data to the same evenly spaced (in wavelength) points, ratios the two datasPts to obtain a percent reflectance reading, and stores the reflectancP
05 data. ~AKl.BAS reads the reflectance data files for a set of samples, smooths the data to the required point spacing, and outputs a file formatted correctly for the calibration and prediction programs.
Multiple scans of raw reflectance data taken on each sample are average~ point by point to improve the signal to noise ratio. The main analyses were completed with 10-scan aver~ging, but some supplementary analyses averaged other numbers of scans. Further analyses were similar in most respects; however, different reference materi~ls and numbers of data points were used in some instances.
A dynamic baseline correc~lon algorithm is used to convert the AC coupled raw re~lectance signal to a DC signal with a baseline level of zaro A/D counts (see Fig. 11). Referring to the schematic of the detector and preamplifier (Fig. 5), the parameter which varies directly with the level of incident radiation is ~he current through the photode~ector, Iin. The parameter digitized by the data colIection system is the output voltage of the preamplifier V ou~ . The general relationship between these two parametPrs at any point wa~ given by:
i~n ~ kl + k2 * VOu~ + k3 * J' Vouc dt (12) .:, . . .
30 l'he ouepue voleage, Vout could be con~erted eo A/D counes by applicaeion o~ a gain and o~fse t:

Dr~ ~ aO ~ al * V~ut ( 13 ) 3 5 VOut - ~3 + a4 * Dr~ ( 14 ) ;

W O ~1/044X~ ~ ~ 7 ~ ~ PCT/US90/0537~
, i,~ ,.
--19-- ' Equacion (12) could be digieized and rPwritten with ehe raw A/D daca counts, D~w, replacing Vo~t and the correceed A/D counes, DCor~ repLacing iLn:

n Dco~r ~ Cl ~ C~ * D~ + c3 * ~ (Dr~ + c~,) (15) 05 1 .
Equation (15) could be simplified by setting c2 to one, since the ~nits Of Dr~W were the same as the units of Dco~ ~ Also, c~ could be removed rom under ehe su~mation.

n Dcorr ~ Dr~ + cl ~ c~ * n + C3 * ~ Dr4" (16) Deter~ination of Dco~r for any gi~en Dr~ then required the valuPs of ehree paramecers to be deeer~ined. These values were unique to each parcicular daea curve, due eo the differences in total reflected energy becween samples. To determine c~, c~, snd c~ lt was necessary to make use of the ~act that DCosr should, on the average, be ~ero in the baseli~e portions of ;
the curve Then, in ehese two poreions of the data curve, Equation 16 cou.ld be rewritten as:
n D~ cl - c~ * n - c3 * ~ Dr~ (17) ,, .
Equstion (17) was ehen fit to ~the data in the baseline areas of the 25 curve usfng a least-squares ~ultiple linear regression, whe~e the independent ~ -, variables at each daea pOi;lt were the index number of the daea poine and the summacion of Che raw A/D data up to ehat poinc. The dependent variable was ~-the raw A/D data value. The equation was fie to a 167-point seceion o the 30 ~ ba~eline OD Qith~ side >f th- tara portion of the curve, corresponding to ..
.:

: . . . - .
:

.: :

W O')1/04~X~ PCT/~S90/0537~ , ~ 2 -20-16.67 ms at a 10 kHz sampling rate. In ehis way, the baseline correction was :.
fic co the mean of any 60 Hz noise present in ehe data. Th2 algoriehm to determine the baseline correccion coefficients was ~
05 :: :Defining: ~ - .
X~n-n~ ~ 1 (18) Xtn-n~ n ~19) n ~n-nl~,2 ~ ~ Drd~ (20) Y~n-nl) ~ Dr~ (21) ~here the range of n was from 334-co 500 (with ni-333) and 901 to 1067 (with ni-733) .:;
1 5 ' - . : .
Then:
: b ~ y (22) .
.~ . .
",.' .

cl - - bo (23) ' bl (24) r, : : ' C3 - - b2 (25 25. :

::

- :. .
~ 35 - : . .~ . :

~J~
W09l/n448~ PCT/US90/05374 .. . .
~.J,,i~' :
-21~

With the necessary constants determined in this manner, Equation (16) is then applied pointwise to the raw A/D data to generate a baseline corrected raw data curve (Fig. 11).
05 The calibration developed by use of fixed filters is applied to the data to convert the A/D point number to its corresponding wavelength. An additional calibration is used to compensate for the remaining wavelength variation at the initial step response portion of the curve (Fig. 9). The wavelength at ~he half-peak point on this portion of the curve is calculated and compared to the mean value of the half-peak point obtained from a set of 225 soil and cieramic reflec~ance scans. This wavelength offset is then applied to each point on the curve so that the half-peak point of each individual curve is coincident with the mean half-peak point. Therefore, the initial portion of the response curve is normalized with respect to wavelength, but any sub-period speed variations could cause shi~ts elsewhere in the curve.
Due to wavelength calibration differences and variations in filter disk pariod, a given A/D point will not correspond to the sam~ wavelength for all reflectance readings. Therefore, it is necessary to interpolate the baseline corrected sensor response data to a standard wavelength spacing for pointwise di~ferencing of soil and ceramic readings and additional analysis. Points are generated every 5 nm from 1600 nm to 2700 nm using piecewise cubic spline interpolation 30 ~ algorithms presented by Spath (1974).
The interpolated, baseline corrected raw data obtained from the sensor axe converted to decimal reflectance (percent re~lectance/100) by comparison against data obtainèd from a ceramic disk with a known reflectance in the range of interest. Repeated rPadings o~ the ceramic disk are done on a frequent basis to that compansation can~be made for lamp output fluctuations or ., , - .~ '.. ...

- , . " ... ....... ..... . ..

wo 91/O~g~ 2 2 PCT/US90/OS374 o~her changes in the optical path of the sensor.
Decimal refleGtance at each point is calculated byO

RE:FL ~ REFL~r~ C * DCORR ~ DCORR~r~QlC (26) Where: ,~
~EF~ ;~ deci~al reflectance ::
DCORR ; lnterpolat~d ~nd b~seline corrected sensor daea Decimal refl~ctance d~ta are transformed to optical density (OD) for calibration and subsequent analysis. Prior to this transformation, the reflectance data are smoothed ~rom a 5 nm point spacing to a 20 nm :~
spacing for noise reduction and compatibility with the analysis programs. The operations performed on each ~;
point on the re~lectance curve are: -;

RSI - O.125 * tREFLI2 1 2*RE~Lll + 2*REFLl + 2*REFLl~1 + REFLl~2) (27) ODl ~ - log10 Rsl (28) Where: -REFL - decimal raflectance RS - smoo~hed deci~al refl~ceance OD - optical density ~

Cation exchange capacity t~`EC) and soil moisture content are also predicted with the reflectance data. : ~For ~hese calibrations, 40 nm da~a (1640 to 26~0 nm)~are u~ed, with lO-scan or lOO-scan averaging.
Limited comparisons with predictions using 60 nm data ~ showed no appreciable difference in predicti~e ; 30 capability from the 40 nm data. The best CEC prediction yielded an SEP of 3.59 mEq~lOOg for the combination of 0.~033 MPa and 1.5 NPa moistur~ tensions. Moisture content is predicted~with an SEP of 1.88 percent water for~the data~et including 0.033 MPa, 0033 MPa, 1.5 MPa, an~ air dry soil. In terms of the coe~ficient of ~ -variation, the pre`diction of these twc properties is more accurate than the prediction of soil organic .

.,;

7 ~ 2 WO~ 4X~ PCT/US90/0537 carbon. Thus the invention can also find application in -the measurement of soil moisture and CEC as well as organic carbon.
Several commexcial soil-engaging componen~s 05 were evaluated as mechanisms for preparing a furrow for the reflectance measurement. The chisels and shovels tested did not develop the flat-bottomed furrow reguired for reflectance sensing. A row unit from a Hiniker Econ-O-Till ridge cleaner did create a flat-bottomed furrow with uniform sur~ace texture and was chosen as the sample presentation mechanism for the in-field tests --of the soil sensor.
The selected row unit consists of a horizontal disk row cleaner 100 with attached scraper wings 102 and a leading coulter with depth control bands. The coulter was removed since it would push residue and dry soil into the furrow bottom and possibly corrupt the reflectance measurement. The depth bands 104 were retained to serve as a depth control wheel on the front of the unit. A mounting 106 was fabricated to attach the Hiniker row unit to the three-point hitch 108 of a John Deer 755 compact utility tractor used for the field tests.
The sensor components of ~ig. 2 are mounted to the modified HinikPr row cleaner unit described above.
The filter disk housing 25 and source 20 are contained in a sealed electrical enclosure 115 attached to the vertical mast 110 of the ridge cleaner unit through ;~
vibration isolation mounts 112. The sensor head assembly 65, detector 26, and preamplifier (Fig. 5) are -mounted n another ~ealed electrical enclosure 116, with :
t~e quart window 68 of the sensor head assembly aligned with an aperture in the bottom surface of the enclcsure.
The fiber optic bundle 24 between the~e two enclosures ~ !~
is protect~d inside a length of flexible conduit 118.
The sensor head enclosure 60 is mounted in the light exclusion shield 120, an open-bottomed steel box, WO~ 44~ PCT/US90/05374 2~6~2 ~

which is in turn rigidly attached to the rear of the ridge cleaner row unit. Approximately 20 mm of vertical adjustment in the position of the sensor head relative to the furrow bottom is achieved through mounting slots 05 122 at the connection between the sensor head enclosure and the light exclusion shield. A rubber skirt 124 extends 10 mm below the lower edge of the light exclusion shield to fur~her protect the sensor from ambient light. The portion of the skirt at the front of the light exclusion shield also serves to smooth the furrow bottom after the passage of the ridge cleaner unit.
Three 12 V deep-cycle batteries are mounted on -~
the tractor to power the sensor and data acquisition system. One battery, mounted on the left fender, pro~ides DC power to the lamp and filter disk-timing circuit. The remaining two batteries, mounted on the ~ront of the tractor, provide power to a 24 V invertor which outputs 115 VAC. The invertor output is conditioned by a regulator and used to power the filter disk motor and the precision power supply for the detector and preamplifier. All sensor components can also be DC powered. ~-We have found that data acquisition can also be accomplished with the same MetraByte DAS-16 analog and digital I/O board with the DAS-1~ installed in an expansion chassis attached to a Zenith Supersport Model 20 laptop computer with an 80C8~ processor running at 8 MHz. The computer and expansion chassis are powered by the 115 VAC availahle on the tractor.
The data acquisition software used is somewhat modified from that used with the 12 MHz AT-Compatible computer. The major functional change required is that the program be able to acguire and store individual data -~
scans. Microsoft QuickBASIC is used to compile the data acquisition progr m, as it allows all available memory for array storage. This allows at least 450 scans in a .

:

WO91/04-1X~ ~,JJ 7 ~d W PCT/US90/0537 3 min testing run to be stored in an array as they are acquired. The data are writ~en to disk after the run is completed, using a binary format to conserve disk space.
05 Since this is a slower computer, the data acquisition rate is raduced from the l0 k~z level used in the laboratory to 5 kHz. Tests similar to those described above for determining the sampling rate showed that 5 kHz is a reliable rate with this system.
Hardware triggering is also preferably implem,ented in the field data acquisition program. A
momentary switch mounted within easy reach of the tractor driver is preferably connected to the I/O board digital input port. Once started, the program waits until the swltch is closed before beginning data acquisition, allowing the data run to begin at a defined location in the field.
Dif~erent hardware and software could be utilized without departing from the invention.
While particular embodiments of the invention have been shown and described in detail, it will be obvious to those skilled in the art that changes and modifications of the present invention, in its various aspects, may be made without departing from the invention in its broader aspects, some of which changes and modifications being matters of routine engineering or design, and others being apparent only after study.
As such, ~he scope of the invention should not be limited by the particular embodiment and specific 30 ~ cons~ruction described herein but should be defined by -the appended claims and equivalents thereo~
Accordingly, the aim in the appended claims is to cover all such changes and modifications as fall within the true spirit and scope o~ the invention.
~ ~;
.; . ,: .
.: . . .
"' ,.', ..
''. ;.:-'

Claims

1. A soil test apparatus for field use comprising: means adapting said apparatus for transport over a field for testing the soil thereof; infrared radiation generating means for producing infrared radiation at a plurality of predetermined wavelengths;
light carrying means coupled to said infrared radiation generating means and extending therefrom for directing infrared radiation onto the soil; and light detecting means for detecting infrared radiation reflected from the soil and for producing corresponding electrical signals.

2. Apparatus according to claim 1 wherein said infrared radiation generating means comprises an infrared source and a circular variable filter monochromator.

3. Apparatus according to claim 1 wherein said infrared radiation generating means comprises an infrared source and a grating monochromator.

4. Apparatus according to claim 1 wherein said infrared radiation generating means comprises an infrared source and a prism monochromator.

5. Apparatus according to claim 1 wherein said light detecting means comprises a single element photodetector.

6. Apparatus according to claim 1 wherein said light detecting means comprises an array photodetector.

7. Apparatus according to claim 6 wherein said photodetector comprises a linear array photodetector.

8. Apparatus according to claim 2 wherein said circular variable filter has a wavelength in a range of from substantially 1600 nanometers to substantially 2900 nanometers and a band width of no more than substantially on the order of 60 nanometers.

9. Apparatus according to claim 1 wherein said light carrying means comprises a fiber optic cable.

10. Apparatus according to claim 5 wherein said photodetector comprises a lead sulfide (PbS) near infrared radiation photoelectric detector.

11. Apparatus according to claim 5 wherein said photodetector comprises a lead selenium (PbSe) near infrared radiation photoelectric detector.

12. Apparatus according to claim 5 wherein said photodetector comprises an indium arsenide (InAs) near infrared radiation photoelectric detector.

13. Apparatus according to claim 2 wherein said circular variable filter comprises a quarter-segment circular variable filter having a generally circular disk-like configuration and further including rotating means for rotating said circular variable filter with its center offset from said light source for modulating the infrared radiation from the same by chopping.

14. Apparatus according to claim 13 wherein said means for rotating comprises a servo-controlled motor-generator.

15. Apparatus according to claim 14 wherein aid servo-control motor-generator rotates said circular variable filter at a rotational rate of no more than substantially on the order of 10 rpm.

16. Apparatus according to claim 13 and further including rotational sensor means for sensing the rotation rate of said circular variable filter and for producing an electrical signal corresponding to said rate of rotation thereof.

17. Apparatus according to claim 16 wherein said rotational sensor means comprises optointerruptor means for generating a timing pulse corresponding to interruption thereof, and at least one projecting member on said circular variable filter for interrupting said optointerruptor means at least once per revolution of said circular variable filter.

18. Apparatus according to claim 2 wherein said light source comprises a quartz halogen lamp.

19. Apparatus according to claim 2 and further including a spherical biconvex lens mounted intermediate said light source and said circular variable filter for focusing the light from said light source onto said circular variable filter.

20. Apparatus according to claim 13 and further including a light shield having a restricted slit intermediate said light source and said circular variable filter for directing light onto a single controlled area of said circular variable filter as the same rotates relative to said light source in such a manner as to obtain a substantially monochromatic light transmitted by said filter at any given position thereof relative to said slit.

21. Apparatus according to claim 9 wherein said fiber optic cable comprises an elongate fiber optic bundle-type section converter, wherein the fiber area thereof changes from a substantially rectangular cross-section at one end thereof to a substantially circular cross-sectional configuration at the other end thereof.

22. Apparatus according to claim 21 wherein said rectangular cross-sectional end of said fiber optic bundle is mounted to receive light from said circular variable filter and wherein said circular cross-section end thereof is mounted to direct light onto said soil.

23. Apparatus according to claim 1 and further including processing means for receiving said electrical signals from said detector and for determining a predetermined property of said soil therefrom.

24. Apparatus according to claim 23 wherein said electrical signals are AC coupled to said processing means and wherein said processing means includes dynamic baseline correction means for converting said AC coupled signals to a baseline corrected data comprising a DC signal with a baseline level of zero A/D counts.

25. Apparatus according to claim 23 wherein said processing means further includes means for developing a step response curve from a series of said DC signals and further including wavelength calibration means for converting A/D point numbers on said curve to their corresponding wavelengths and for compensating for wavelength variation at least at initial step response portions of said curve, for normalizing the initial portion of the response curve with respect to wavelength.

26. Apparatus according to claim 24 wherein said processing means further includes data point interpolation means for interpolating the baseline corrected data to a standard wavelength spacing.

27. Apparatus according to claim 24 wherein said processor means further includes reflectance calculation means for converting baseline corrected data to decimal reflectance data by comparison against data obtained from a standard ceramic disk reference element with a known reflectance at said predetermined wavelengths.

28. Apparatus according to claim 27 wherein said processing means further includes optical density calculation means for transforming said decimal reflectance data to optical density data.

29. Apparatus according to claim 1 wherein said adapting means comprise means for mounting said apparatus to a vehicle, and further including furrow opening means mounted to said frame means for opening a furrow in the soil as said vehicle transports said apparatus thereover.

30. Apparatus according to claim 29, wherein said infrared radiation generating means is mounted to said frame, said light carrying means extend along said frame for delivering infrared radiation to a furrow opened by said furrow opening means, and said light detecting means is mounted to said frame means adjacent said furrow opening means.

31. A soil test apparatus for field use comprising: frame means for mounting to a vehicle for transporting the apparatus over a field for testing the soil thereof, furrow opening means mounted to said frame means for opening a furrow in the soil as said vehicle transports said apparatus thereover; infrared radiation generating means mounted to said frame for producing infrared radiation at a plurality of predetermined wavelengths; light carrying means coupled to said infrared radiation generating means and extending along said frame for delivering infrared radiation to a furrow opened by said furrow opening means for directing infrared radiation onto the soil; and light detecting means mounted to said frame means adjacent said furrow opening means for detecting infrared radiation reflected from the soil and for producing corresponding electrical signals.

32. A soil test apparatus comprising:
infrared radiation generating means for producing infrared radiation at a plurality of predetermined wavelengths; means for directing light from said infrared radiation generating means onto a sample of soil; and light detecting means for detecting infrared radiation reflected from the soil and for producing corresponding electrical signals; wherein said infrared radiation generating means comprises an infrared source and a circular variable filter monochromator.

33. A soil testing method comprising:
producing infrared radiation at a plurality of predetermined wavelengths; directing said infrared radiation onto a sample of soil; detecting infrared radiation reflected from the soil, anbd producingt electrical signals corresponding to the detected radiation, wherein the step of producing infrared radiation comprises rotating a circular variable filter monochromator in the path of infrared radiation from a fixed source.

34. A test apparatus for determining constituents of a sample of material comprising:
infrared radiation generating means for producing infrared radiation at a plurality of predetermined wavelengths; means for directing light from said infrared radiation generating means onto said sample of material; and light detecting means for detecting infrared radiation reflected from the sample and for producing corresponding electrical signals; wherein said infrared radiation generating means comprises an infrared source and a variable filter.

35. A testing method for determining constituents of a sample of material comprising:
producing infrared radiation at a plurality of predetermined wavelengths; directing said infrared radiation onto said sample of material; detecting infrared radiation reflected from the sample, and producing electrical signals corresponding to the detected radiation; wherein the step of producing infrared radiation comprises placing a variable filter in the path of infrared radiation from a fixed source.