Advertise Here
Two Hour PCBs
Professional website hosting cheaper than doing it yourself

Login    Register Browse Everything


previous article next

Electronics has changed a lot since this research was done.  Today, you would be advised to go buy an Autek RF impedance meter to reproduce this experiment.   The formating was orginally done for a dot-matrix tractor feed printer so that the print would adhere to University requirements for page layout.  So some sections look different than others.  

Water Content Measurement by Reflected Power Method 
Alan Dewey

Associate of Applied Science, Communications-Broadcast Technology, Parkland College, Champaign, Illinois
Bachelor of Science, Engineering Technology, Southern Illinois University at Carbondale
A Thesis submitted in Partial Fulfillment of the Requirements for the Master of Science in Manufacturing Systems in the Department of Technology Southern Illinois University at Carbondale. December 1990


A non-destructive, non-invasive method to measure water content of raw materials and finished products is proposed and tested. A radio frequency oscillator provides a signal to a tuned circuit. The item to be measured is placed inside the inductor of the tuned circuit. Impedance of theinductor is affected by the complex electromagneticproperties of the sample and its water content. The magnitude of change of inductor impedance is determined from its effect on the measured relative transmitted and reflected signals. From the relative transmitted and reflected signals, the numerical value of Standing Wave Ratio Coefficient (SWR), is determined. The circuitry is simple, portable, inexpensive and safe. This method could be used to check water content in raw material or finished product, on-line, off-line, or in storage without need to disturb packaging. Further improvements may develop a device that would provide simple interfacing for closed loop process control.

Acknowledgements The author wishes to thank Dr. Jefferson Lindsey III for his discussions on electromagnetic theory and Dr. Marek Szary for suggestions regarding the procedure. The subject of investigation resulted from suggestions by Dr. Daniel Chavez of SIU Department of Anatomy. The author also appreciates the suggestions and comments from Dr. John Gelderd of Texas A&M University and Dr. Thomas McRae at Laser Imaging Systems, Inc.

Table of Contents

Abstract ..........................................ii
Acknowledgements ..................................iii
List of Figures
List of Tables ....................................vii
Background ........................................1
Invasive vs. Non-Invasive methods ...............2
Invasive and Destructive Methods ................4
Non-Invasive Methods ............................6
1 Two Electrode Conductance Method ..........6
2 Four Electrode Conductance Method..........8
3 Capacitance Methods .......................9
4 Inductance Methods ........................12
5 Microwave Radio Method ....................16
Background Summary ..............................18
Statement of the Problem ..........................20
The Proposed Method ...............................21
Circuit Principles ..............................24
Experiment Apparatus ............................32
Research Procedure ................................36
Calibration of Equipment ........................36
Acquisition of Data .............................38
Results ...........................................42
Description of Graphs ...........................43
Discussion of Results ...........................49
Future Work .......................................54

Bibliography ......................................56

Appendix A Raw Data ...............................

Vita ..............................................60

List of Figures

Figure 1 Experiment Apparatus .....................31
Figure 2 The Directional Coupler ..................33
Figure 3 Impedance Matching Network ...............33
Figure 4 Comparison of Reflected, Forward
Readings and Standing Wave Ratio (SWR) ...44
Figure 5 SWR vs. Sponge Wet Mass;
Data Point Plot ..........................44
Figure 6 Linear Regression Comparison;
Distilled Water ..........................45
Figure 7 Variance Range ...........................45
Figure 8 Prediction Variance ......................46
Figure 9 Slope Variance ...........................46
Figure 10 Standing Wave Ratio vs.
Saline Solution ..........................47
Figure 11 Wet Mass vs. Natural Logarithm
of SWR ...................................
Figure 12 Comparison; Saline Solution to
Distilled Water ..........................47
Figure 13 SWR vs Saline; Unprepared Sample .........48

List of Tables

Comparison of Ice and Distilled Water ............

Background Information

Long range success in manufacturing requires attention to quality. Where water content affects the quality of the finished product, or the processing of raw material, this variable must be quantified for application to process control.  Continuous adjustment to a process based on data from the results of that process is possible only with timely information. This is fundamental to closed loop process control.

The significance of a product's water content varies with situation. Effects may be simple, for example, causingvariation in the curing time of an adhesive, drying time of paint, or the shelf life of finished goods and work inprocess.  Of safety concern might be the strength of ceramic or engineered materials when related to moisture content.Results of moisture may be more serious. For example, nitrogen tetroxide rocket fuel becomes highly corrosive withonly a few tenths percent water content. 1

Water content measurement techniques of varying accuracy, range, speed and application have found widespread use based upon the characteristics of the product or process.  The choice of technique involves the consideration of:

  1. the form of the material, i.e. liquid, solid, or gas; 
  2. the range of measurement capability required of the instrument; 
  3. portability; 
  4. cost; 
  5. time frame allowed to obtain results; 
  6. safety; 
  7. level of operator skill required; and other factors.

Most existing production line inspection sampling techniques are boring and human inspectors are susceptible to tiring and occasions of inattentiveness, especially for repetitive, simple tasks. 2 Machines do not exhibit these characteristics, making reliable, repetitive 100% inspection a feasible undertaking.  Furthermore, destructive and invasive testing allow only a sample of the product to be tested, as in explosives and ammunitions. Thus the ideal combination is machine controlled non-destructive 100% evaluation. With computer control of inspection, and proper data acquisition and interfacing, accurate measurements maybe made in a time frame suitable for closed loop process control. 3 Further, statistical analysis of the process maybe simultaneously generated. 4

A review of basic methods of moisture measurement illustrates the properties of water that enable its measurement. Each method has certain advantages and disadvantages. The method chosen in any particular situation is based upon required accuracy, simplicity, and form of the material. The concepts of conductivity, magnetic permeability and dielectric constant are demonstrated.

Invasive vs. Non-Invasive

Testing methods that remove a sample of the product to be measured may or may not be considered destructive. For example, to remove a one inch core sample from a 12 inchwooden beam may not significantly alter the usefulness or life of the member. Yet, removal of a one inch core from a 2 inch board will render it useless as a structural member. In each case, though, core removal is invasive. In the monitoring of soil moisture, for example, a device may be inserted to the desired depth for measurements. Monitoring humidity in concrete may involve implantation of a thermocouple probe. These also, while not destructive in these applications, do belong in the category of invasive techniques.

This paper will consider methods that are either destructive or invasive in a category distinct from those that are truly non-invasive and non-destructive.


Invasive and Destructive Methods

The most accurate method of water content measurement of any solid is to determine physically the mass of water that may be removed. Such a procedure is always destructive, but this provides a basis to which all other methods are calibrated. Water content as expressed by percent weight may be easily calculated by completely drying the sample and determining mass of the water removed. This may be accomplished by weighing before and after oven drying. Another method involves anhydrous methanol to extract all water from grain. Obtain a sample of the product or the raw material, assume homogeneity, use the methanol to remove the water and then calculate the content as a percentage of total mass.

An invasive method that provides continuous monitoring of moisture content depends on the property of certain materials to expand dimensionally with increase in moisture content. This method was used by Elda De Castro in Portugal for soil measurements. 5 The dimensional changes of woodstrips may be monitored electronically as the change in their length directly increases or decreases the tension in a vibrating wire. Changes in tension cause the resonant frequency of the wire to vary proportionately (as in tuning an electric guitar string). The assembly of wood strips, once calibrated, may be inserted into the soil. With time, the moisture within the wood strips will reach equilibrium with their surrounding environment and the output of the device may be monitored constantly. After data are compensated for temperature, the change in length of the wood strips due to moisture is known, and correlated to the water content of the soil. However this method does require inserting the device into the material to be measured.

Another method requiring implantation is the plaster of paris block. Bouyoucos 6 has written several articles describing the use of two electrodes cast in plaster of paris block. The block absorbs and releases moisture in relation to its surroundings. The electrical resistance between these electrodes indicates water content of the soil. Because this method measures water within the block,there is no need to correct for variations in soil density. These and other methods requiring a device to absorb or release moisture to attain equilibrium with the environs exhibit significant hysteresis. Because of the hysteresis, absolute values of moisture content cannot be determined without knowing whether the water content of the substance has been increasing or decreasing.

Invasive moisture measurement also may be made utilizing thermocouples and heaters. 7,8 The device is implanted in the material. The rate at which the temperature will rise in a porous material is related, among other things, to total water content. By measuring the rate at which temperature increases upon application of known heat quantities, or by measuring heat required to raise the temperature a specific amount, the water content may be derived. The thermocouple, plaster block, and wood strip methods of moisture measurement cannot distinguish between water or other permeating liquids.

Because invasive and destructive methods of moisture measurement destroy or alter the item, there is an economic loss for each measurement performed. Further, the manufacturer monitoring incoming, process, or outgoing quality is limited to sampling inspection.

Non-Invasive Methods

Non-inasive water content measurement also encompasses various methods. These include measuring the electrical capacitance effects of the sample, inductance effects, the attenuation of microwave radio energy, or direct measurement of conductivity. More time consuming methods include infra-red spectroscopy. 9 The physical attributes of the material will usually preclude certain methods. Methods utilizing electrodes may fall in the non-destructive category, yet most would not be usable once a product has been packaged, or before raw materials are uncrated at the shipping dock.

1) Two Electrode Method

A two electrode conductivity method similar to that used in the plaster of paris block method described earlier, where the electrodes are placed upon the material, not inserted within, would be non-destructive and non-invasive, though not feasible once the product is packaged. The ohmmeter is a two terminal device that provides voltage on two electrodes. Any non-insulator connected between the electrodes will then allow some amount of current flow. The magnitude of the current flow is directly proportional to conductance. Since conductance increases with increasing water content, a correlation can be derived for this effect.

Direct measurement of conductivity must be accomplished by physical contact of the electrodes and the material under test. The material being tested dictates the type of electrodes used. Loose materials that are powdered or granular might be put into cells. Wood may require the electrodes in the form of clamps. Surface electrodes may be used if the object has sufficient conductivity to provide low contact resistance. However direct measurement of conductance is affected by variations in materials which are not homogeneous. Temperature, packing density, and electrode contact resistance are also problems. 10 However, not only changes in water content affect the conductance. Other factors include mineral content of the water, salt content, temperature, electrode size, pressure with which the electrodes are attached, distance between attachment points, and the frequency at which the measurement is performed. These variables significantly limit the accuracy of the two electrode methods. For liquids, however, many of these factors can be controlled by using electrodes produced for uniform size, shape, and spacing, and by always performing the measurements at a specific frequency.

Two electrode wood moisture meters are available commercially. Plans for a simple two electrode conductance wood moisture meter kit are given and use of such a meter is demonstrated in an article in Popular Electronics. 11 This article shows the simplicity of the equipment and procedure. Operated by battery and a one transistor meter circuit, the electrodes are two nails that are pressed into the wood. However, no mention is made in the article of the variable readings that will be obtained with differing penetration of the electrodes into the wood.

Two electrode methods as described here are not the same as the capacitive method described later. Capacitance measurements do not require physical contact with the sample.

2) Four Electrode Conductance Method

An improvement over two electrodes is the four electrode method. In this case, two electrodes placed upon the material are used to inject a known constant alternating current into the sample. The current is independent of electrode contact resistance, temperature, spacing of electrodes, frequency or even the conductivity of the sample. Two more electrodes electrically isolated from the first pair are used to measure the voltage difference at two points a specified distance apart. Both of these electrodes are placed between the current-supplying pair. Note that these electrodes are placed on the surface of the material, not embedded. Thus, there is no invasion of the sample required. By using a bridge type circuit, no current is required to flow in the voltage measurement electrodes. Thus, this measurement is also independent of electrode contact resistance. To illustrate the non-destructive nature of the four electrode method, consider that measurements were performed on live humans as early as the 1930s. 12   Moreover, this method is still commonly used today. Using a constant 800 microAmp rms current at 100 kHz, measurements from hand to foot correlate to total body water in the subject. 13  Biologists refer to this as Bioelectric Impedance Analysis. When the electrodes are placed closer together, across a finger or leg, the changes in conductivity due to blood pulsations can be measured and recorded. Development of the four electrode method revealed that the magnitude of whole body impedance varies with frequency. Thus measurements using the four electrode method must also be performed at a constant frequency.


3) Capacitance Methods

Use of electrical capacitance to measure water content is based on the significant difference between the dielectric constant of water and that of most solids. Dielectric constant is the nature of the medium between two plates of a capacitor to alter the value of the electron forces between the plates. Any insulating medium within the electric field of a capacitor alters the electron charge to voltage ratio (the capacitance) of these plates. The degree of this effect, expressed as ratio to the capacitance value in a perfect vacuum, is known as the dielectric constant. 14,15 For example, if the air (relative dielectricconstant = 1.0) between two capacitor plates is completely replaced with water, the value of capacitance now exhibited by these same plates is increased by a factor of approximately 80.

An alternating electric field of known voltage may be applied to two plates of a capacitor to excite currents in the item to be measured. Electrically, the resistive component of the induced current (in phase with voltage) and capacitive component of the current (leads voltage by 90 o ) will then exist in parallel within the material. The net phase angle of current is the vector sum of the capacitive and resistive components. This phase angle may be measured, along with the magnitude of the current, to determine the net capacitance.

The magnitude of the capacitive component of the current decreases with decreasing frequency, whereas the resistive component of the current does not, so that below approximately one MHz, determination of the capacitive component becomes difficult. 16 Frequencies above 100 MHz are not suitable either. At frequencies greater than this,the water molecules do not follow the signal. 17  Most capacitance type meters employ bridge circuits. These consist of accurate oscillators and components. The bridge employs a variable capacitor with a calibrated dial. The circuit is adjusted to bring the bridge into balance, as indicated by a meter. The reading from the adjustable capacitor is used to determine capacitance of the external capacitor. The dial may be calibrated to read directly in terms of moisture content of a specific type of material. Another technique utilizes the test capacitor as part of a resonant circuit. Changes in the capacitance affect the resonant frequency of the oscillator which may be determined by a frequency meter. This method is applied in grain moisture meters.

Most solids exhibit dielectric constants less than 8, whereas water is slightly more than 80. 18 This great difference permits a correlation between capacitance and water content. Capacitive methods of moisture measurement are quite popular in measurements of grain and wood. For these methods, in which the material being measured is placed between the plates of the capacitor, the total mass must be known. The capacitive method does not differentiate water molecules from those of the material. What is measured is the total dielectric effect on the capacitor. Therefore, either a specific mass is placed in the test chamber, or a correction chart is used for the actual mass measured.

Similar in construction to the two electrodes in plaster of paris block by Bouyoucos discussed earlier, is one by J.Flectcher 19 that embeds two brass plates of a capacitor in the plaster of  paris block. Fletcher measures the change in capacitance of the block to determine percentage of soil moisture. The moisture content determined in this manner is said to be immune to water conductivity variations caused by minerals in the water. 20 However the block must absorb water from the soil, and therefore must be inserted into the earth, which puts this method in the invasive category. A non-invasive capacitive probe system for detecting moisture is detailed by Outwater which provides relative readings. 21 This device used a capacitive probe made from two concentric tubes of brass. The end of the probe is placed on the surface of the item being tested. The net capacitance of the probe is affected by the dielectric characteristics of the material, and of any moisture present. Since the distance from probe to water affects the measurement as well as the quantity of water, no quantitative data will result. The unique aspect of this device is not the probe (which has performance limitations) but the method of detecting changes in its capacitance. Whereas most capacitive detectors use either Wheatstone or L-C (inductor- capacitor) bridges, this circuit uses two identical resonant circuits with the exception of the capacitive probe. With only dry air near the probe, the circuits are balanced and no meter indication is noted. Any imbalance in the test circuit caused by dielectric effects or variance in conductivity, will alter the balance of the system and deflect the meter.

Despite the lack of quantitative information, this device has proven useful in determining the extent of water penetration in fiberglass rocket motor cases. Graphs connecting plots of identical readings provided indication of water invasion contours. Repeating the measurements and graphing at regular time intervals showed propagation of water in the medium.

4) Inductance Methods

Another method that does not require the attachment of electrodes to the sample utilizes the properties of an inductor. When a conductor is wound in a helical shape, its inductance is increased and may be calculated. Such components are referred to simply as coils. A varying current in the conductor varies the magnetic field about each turn of the helix. This varying current creates a varying magnetic field which affects the current flowing in adjacent turns.

Presta et. al 22 used a large solenoidal coil driven with a 5-MHz radio frequency current and a human subject placed internally. The difference between coil impedance empty and with the subject inside is an index of the conductivity of the subject. It is this conductivity that correlates to the water content of the subject. By combining this information with height, weight and sex of the subject, the total body water can be predicted. The reference method for total body water measurements in live humans is by weighing the subject in air and under water. Correlation of this inductive method with the hydrostatic weighing measurement was 0.903 with a confidence level (1-P) of .0001. The article describing this work does not state by what method the coi impedance is determined.

Tarjan and McFee 23 used electrostatically shielded coils operating at 100 kHz to measure conductivity changes of the human thorax and head due to pulsatile blood flow. The shielding was designed to eliminate the electrostatic field effects, allowing only measurement of magnetic properties of the subject. One coil was driven with the 100 kHz signal.

Two more coils placed coaxially, one on each side of the driven coil, were attached in opposing phase (180 0 ) to detect variations in the magnetic field as it was changed by the varying volume of blood within the subject. Although this system was highly sensitive to motion of the subject, and required very careful construction of the coils, the device provided graphical records of varying blood volume within the body as low as 0.05%.

It is important to understand that the use of conductivity to measure body fluid in humans is possible because the chemical balance of bodily fluids is maintained within narrow tolerances. Thus the conductivity of the body  fluids is fairly constant. Conversely, pure water is an extremely poor conductor and its quantity would not be measurable by conductivity methods alone.

Magnetic induction is also used to measure fluid conductivity for laboratory or manufacturing process chemicals. One type of commercially produced conductivity meter induces alternating current into the solution using toroidal coils. 24 A second coil picks up the magnetic signal created by this current induced in the fluid. Toroidal core coils are self shielding, so only magnetic fields are detected by the pickup coil. So the output ofthe detection coil depends on the magnitude of the induced current. If the fluid within the field of the excitation coil has no conductivity, no current will be induced and therefore no magnetic field will be coupled to the pick upcoil. A conductive fluid will complete a magnetic path between the coils. The magnitude of the detected signal is proportional to the conductivity of the solution. Differences in dielectric constants will not directly affect the magnitude of the detected signal. However, these measurements must be compensated for temperature because conductivity of some solutions may have temperature coefficients as high as 4%. 25

Note that the inductive methods described herein are based upon the conductivity of the solutions, not magnetic permeability. Failure to understand the difference between conductivity effects on a magnetic field and permeability effects may lead one to believe that inductance methods would not be practical. Nyboer 26

"Unfortunately, physicists indicate that the magnetic permeability of tissues is essentially the same as the air space with which it competes. Our conclusion is therefore that there is no support for construction of a direct plethysmograph based upon the inductive characteristics of body tissues which are ionic and not electronic conductors distributed essentially in a parallel relation to the surrounding space."

Halsted,, 27 in his book on aqueous dielectrics, states; "The magnetic properties of the moist substance are rarely of importance." These statements discounting inductive methods not only ignore that the conductivity of the moist sample can affect inductors, but that the effects of dielectrics on the distributed capacitance of inductors can be measured. If an unshielded inductor is used to induce current in a material, and the impedance of the coil is simultaneously monitored, then it is possible to measure effects on the coil due to the dielectric properties of the sample, as well as those due to its conductivity. Such a method would detect not only variations in quantity of water present in a sample, but also variations in conductivity or dielectric constant which might also be caused by contamination of thesample.

5) Microwave Radio Methods

Though capacitance methods use frequencies from 1 to 100 MHz, measurements of moisture content may be performed at frequencies on the order of GHz (GigaHertz = 1000 MHz) by determining the amount of electromagnetic radiation absorbed. Microwave signals are attenuated by water at a linear rate, expressed in decibels, per unit thickness of water. This relationship holds until the water content is so high that there exists free water as well as bound. 28  Thus, for any given frequency, distance of measurement and material, the signal attenuation is proportional to water content.

James et al 29 used 4.81 GHz microwave electromagnetic radiation to measure moisture content of sawn wood. The microwave signal was transmitted through the wood and reflected back to the detector by a rotating dipole antenna.  The strength of the signal detected, after the round trip through the wood, was attenuated in proportion to water content. By spinning the reflecting antenna at 9,000 rpm, the grain angle of the wood could also be determined. This information was used to predict the strength of the wood. Although this method provides consistent measurements, it requires expensive microwave radio generators and careful alignment of the complicated arrangement of equipment. Further problems include the confounding reflections of signal from surrounding objects and complicated data reduction.

Background Summary

Reliability of field use measurement equipment is reduced by complexity. Elimination or reduction of environmentally sensitive circuitry, while maintaining performance levels, will have economic advantages. Reduction in quantity and complexity of operating controls eases operation, reduces operator training time, and lowers the probability of mistakes.

As described here, accurate and repeatable methods of non-destructive, non-invasive moisture measurement include four electrode, capacitance, microwave radio attenuation, and magnetic permeability methods. The two electrode conductance method is inaccurate. Four electrode methods are not suitable for use with packaged items as the electrodes require direct contact with the item. Microwave methods, while accurate, are expensive, bulky, and require reasonable operator skill. While this may be adaptable to automatic operation by machine, there still exists the problems of signal reflection and expense.  Measurements of conductivity by induced current have been applied in medical research to measure blood flow and total body water, but have not been applied to determine moisture content of products, processes or raw materials.  Overall, the performance of the capacitive methods are suited to closed loop process control and unattended operation. Many of the circuits are very accurate with good repeatability. However, whether the circuit uses a frequency counter, or uses the test capacitor in a bridge circuit, requires that the capacitor itself become part of an oscillator. Many components of the circuit, in addition to the test capacitor, require temperature compensation.  This increases design cost and production cost of the instrument. Component aging increases the frequency of calibration checks and alignment.

Statement of the Problem

One hundred percent inspection of finished product or raw material for a manufacturing environment requires non-destructive methods. Sampling inspection, as in the case of warehoused material, is less practical if unit packaging must be disturbed or removed. Methods which can determine moisture content without physical invasion of the packaging preserve integrity of the product or raw material. If the measurement system is also sensitive to fluid conductivity, this would provide the additional benefit of detecting contamination by salts or minerals that alter the fluid conductivity.

Methods which do not require physical contact with the sample, such as the capacitive and microwave methods, require high component count with temperature compensation. Additionally, the microwave method, at present, is expensive. A method which requires fewer temperature dependent components and overall lower cost would expand manufacturers abilities to monitor and improve product quality.

The Proposed Method

The method proposed here utilizes the relationship between radio frequency energy reflected from the termination of a transmission line and the impedance of that termination. The termination, in this project will be a solenoidal coil. This method reduces the number of components that must be temperature compensated. Overall cost would be lower than resonant frequency measuring circuits or bridge circuits. Operation would be as simple and safe as currently available capacitance methods.  Improvement of the method could reduce to two, the component count affecting calibration. 

The circuit for this procedure uses a solenoidal coil driven by a radio frequency signal fed through a directional coupler and an impedance matching network. The sample to be measured is placed within the coil. The sample, and water within that sample, will affect the net impedance of the coil. A change in the impedance will thus create a mismatch between the transmission line carrying the signal and the impedance matching network feeding the coil. Such a mismatch will reflect signal back toward the source. The relative reading of power from the signal source and the relative reading of reflected power will be measured by a directional coupler and recorded. The ratio of the power from the signal source to the reflected signal power is expected to relate to moisture content of the material under test.

The nature of the reflected power method proposed here does not require routine calibration of the meter or output level of the signal source because the objective of the measurement is solely to determine the ratio of the magnitudes of signal through the directional coupler. A directional coupler is a device that provides a d.c. voltage output in proportion to the signal propagating through the transmission line. Two outputs from the directional coupler are provided. One output signal each represents the signal traveling in each direction through the coupler.

Circuitry is very simple and stable. Only one meter and one measuring circuit are used for both the forward and reverse readings. The user only operates one switch and notes the two readings. Moreover, the circuit can be built with few components and simple controls.

The experimental setup is shown in Figure 1. For this procedure the media investigated was a cellulose sponge.

Water was applied incrementally via a syringe to the sponge for measurement. To permit consistent placement of the sponge within the coil, a carrier was made from foam wrap-around pipe insulation with internal diameter of 13/16inch, outside diameter of 1.75 inches. The outside diameter was trimmed by knife for a sliding fit within the coil form. Alignment marks were cut on the outside surface of the carrier to aid in consistent placement of the carrier within the coil, and of the sponge within the carrier. The cavity within the foam held the sponge securely. This foam was of the closed cell type which prevented absorption of water. A lengthwise cut through the foam provided easy insertion and removal of the sponge.

It is not the use of an inductor that is unique to this research. It is the method of measuring the changes in impedance and circuit loss that is investigated. The reflected power method should be applicable to a circuit utilizing a capacitive test cell replacing the inductor.

Circuit Principles

Changes in the circuit impedance due to the differing net conductivity of the test sample, together with the dielectric effects, will determine the magnitude and phase angle of power reflected. The directional coupler provides a d.c. voltage output that is dependent only upon themagnitude of signal reflected from the circuit. Phase angle does not directly affect the output voltage. A solenoidal winding of wire exhibits inductance related to its diameter, length and number of turns of wire. However, capacitance exists between all turns of the wire throughout the length of the coil. Known as distributed capacitance, this additional reactance is effectively inparallel with the inductor. Proximity of a dielectric to the turns of wire of the coil will increase the value of this capacitance.

The parallel (inductor and distributed capacitance) circuit impedance depends upon frequency. Maximum impedance occurs at the frequency where the capacitive impedance and inductive impedance are equal but opposite. This frequency is known as the self resonance frequency of the inductor. A change in the capacitance of the coil circuit changes the overall impedance of the coil. This also means that the frequency of self resonance is changed. Thus the self resonance frequency could measured as an indicator of water content. Determination of self resonance frequency requires a frequency counter or some other device to monitor the oscillation rate, and that the coil be part of the oscillator circuit. At a fixed frequency, though, a change in the impedance of the coil also changes the impedance at the input to the matching network. This impedance change causes a change in reflected power.

The dielectric and conductivity effects of the foam carrier will be a constant in this experiment, as will be the effects of the acrylic coil form. Salt content has no effect on dielectric constant. 31 However the salt content will affect conductivity, which in turn affects the magnitude of the current induced. Given a fixed dielectric constant and quantity of water within the sample, changes in its conductivity will also vary the inductance of the coil and cause a corresponding change in the reflected power. Conductivity is defined as conductance per unit volume. Thus, a homogeneous sample will have a certain conductivity regardless of its dimensions or volume. However, the conductance of that sample is determined by its dimensions. Therefore, an increase in water content of a sample will increase the net value of conductance. Without dimensional changes, an increase in the quantity of conductive fluid within the internal spaces of the sample, will increase its conductivity. This results in a relationship between water content and conductance. A conductive material within the field of the coil lowers its inductance.

The magnetic field produced by the coil will induce non-uniform currents in the sample. The periphery will have greater current than more internal sections of the sample. (The 'skin effect') If the water within the sample is not uniformly distributed, erroneous data may result. For this reason it is intended for use with materials that are homogeneous.

With proper impedance match, all transmitted power delivered to the matching network will be radiated as radio frequency energy or dissipated as heat by circuit losses, resulting in no reflected power. This will result in an indication of zero reflected power by the directional coupler. Any changes in impedance due to the electromagnetic properties of objects within the electromagnetic field of the coil will alter the line to load impedance match. When load impedance does not match the characteristic impedance of the transmission line, some portion of the arriving signal is reflected back toward thesource. The ratio r of reflected voltage to incident voltage is; 32

(note that the formulas did not reproduce correctly when this file was imported.)

Z r - Z o Z r = Load impedance

r = ------- (1)

Zr + Z o Z o = Line impedance

The amount of reflected voltage, which is determined by the ratio of mismatch in impedance, will relate to the net electromagnetic properties of the sample. The reflection coefficient may only have values from -1 to 1. Whether the sign of r is positive or negative is determined by the relation of the impedance mismatch. When load impedance is greater than the transmission line impedance, r will be positive, and when load impedance is lower than the transmission line impedance, r will be negative. Thus,determination of r requires determining either the true impedance ratio, not just the ratio magnitude, or determination of the phase of the reflected voltage.Another method of expressing impedance mismatch is known as Voltage Standing Wave Ratio (VSWR). It is simple to measure the ratio of signal magnitude travelling toward the source to the magnitude of reflected signal. The formula for VSWR is; 33

power reflected

1+ power forward

VSWR= ------------------- (2)

power reflected

1- power forward


VSWR may also be calculated in terms of voltage; 34


V o +V r V o = forward voltage

VSWR= ----- (3)

V o -V r V r = reflected voltage


The output of the monimatch directional coupler is in terms of volts, so that use of formula (3) is most appropriate. Note that VSWR must always be equal to or greater than 1.Throughout the remainder of this paper this ratio will be called simply SWR.



The monimatch type directional coupler, shown in Figure 2, is a simple device which can determine the ratio of forward to reflected voltages. Construction and circuitry are very simple and low cost. Electrically, the monimatch consists of a section of transmission line with two sections of transmission line for sampling terminated at one end by a resistor matching the characteristic impedance of the line,and at the other by a diode rectifier. Note that the monimatch does not measure the voltage directly. A small portion of the signal is coupled into the sampling lines. With reasonable construction tolerances, the efficiency of each sampling section will be equal. Thus, the true ratio of forward to reflected signals can be determined without knowledge of the actual voltages.

The impedance matching network, Figure 3, is used to match the 52 ohm characteristic impedance of RG-58/AU coaxial cable to the high impedance of the test coil circuit. In the experimental setup, the impedance matching circuit is connected to the test coil by a length of coaxial cable.


The impedance matching network introduces a unknown phase shift into the signal path. Thus, the electrical distance (expressed in terms of wavelength or degrees phase shift) between the directional coupler and the test coil is unknown. Theoretically, the electrical distance between the directional coupler and the test coil will not affect determination of SWR. Whether the directional coupler is located at a voltage maximum or a voltage minimum, the ratio of forward to reflected power remains constant (assuming lossless transmission line). However, nonlinearity of the directional coupler rectifying diodes may produce differing results under such circumstances.

Initial investigation of apparatus performance used commercially produced hermetically sealed quartz crystal oscillators. These oscillators are available from numerous sources under $10. A 7 pole Chebyshev low pass filter was constructed and connected to each oscillator. Output from the low pass filter was found to be approximately 4 mW. This circuit provided a stable, low cost, battery powered signal source. However this power level was found to be much too low for use with an experimental setup of this sensitivity. It was discovered that the coil impedance matching network could not be adjusted accurately or with any repeatability. Also, oscillator output this low required direct connection to the directional coupler with no isolation possible. Changes in load impedance caused signal voltage variations at the filter output. By using a transmitter adjusted to 14 watts output as a signal source,a 10 dB attenuator could be added, reducing signal power to1.4 watts nominal delivered to the impedance matching network.

The attenuator provides a degree of isolation between the transmitter and the varying net impedance of the load. This isolation, in addition to automatic level control within the transmitter, provided constant forward signal level.

Following are some possible causes of variance in the reflected power method of water content measurement. Some factors may be easily controlled in the laboratory, but all should be given due consideration before application to field use.

1. Heating; Some portion of the electromagnetic energy radiated by the coil will be absorbed in the sample undertest. Resistance of the sample transforms current flowing within into heat. (Resistance is that portion of the inverse of conductance which causes current to be in phase with voltage.)

Incidental (forward transmitted) power is 1.4 Watts. Thus, the maximum possible rate at which heat may be added to the sample can be calculated. Converting to calories;1 Joule = 4.1902 calorie 35 1.4 W= 1.4 Joule/second (1.4 J/s)x(4.1902 cal/J)= 5.866 cal/second If all power from the radio frequency source were to be absorbed by the sample, heat would increase by 5.86 calorie/second. Although not all power delivered to the coil will be absorbed by the sample under test, noticeable heating occurs with small amounts of water.
The effect of heating may be kept small by obtaining the reading quickly. The magnitude of this variable will be relatively smaller with increases in mass of the sample. Temperature changes of small amounts would not put this procedure in the destructive method category. Also, it is the conductivity of the sample that will be affected. At the power level used here, temperature changes will not be great enough to yield significant effects on dielectric constant. Conductivity changes, however may be significant with certain fluids.

2. Mechanical changes; Changes in ambient temperature will alter mechanical dimensions of the device, which in turn will affect the electrical characteristics of the circuitry. Much of this can be compensated by "re-zeroing" the device periodically. This requires only a trimming adjustment of the capacitor in the impedance matching circuit. Dimensional changes due to ambient temperature fluctuations should not be a problem in the laboratory.

3. Ambient atmosphere; Changes in atmospheric humidity will vary the amount of total water in the test coil magnetic field. This effect should be small compared to the volume of water within some types of products. Such a change would also be inherently slow enough to permit compensation by periodic recalibration or use of correction tables. Effects of ambient atmosphere humidity are not discussed in literature relevant to moisture content measurement of solids.

4. Motion; this technique is sensitive to motion of the sample. Within the laboratory environment this is of no concern. However, use of the inductor to measure moisture of a product on the production line would require that each sample be brought to a complete stop, or that the detection coil move with the sample.

5. External effects; Items outside the coil of wire, but within the magnetic field will also affect the net inductance of the circuit. These effects can be minimized by proper shielding of the device.

6. Cables; Once an equation or graphical method of determining moisture is generated for the device, neither cable length nor type may be changed. These factors will affect the reflected power ratio for a given impedance change at the coil.


Experiment Apparatus

Dimensions of the coil were chosen by experimentation to provide achievable impedance matching and overall physical size suitable to test small samples. A lathe was used tocut 40 pipe threads per inch for a length of 6.0 inches on an acrylic form of 2.25 inches outside diameter and 2.00 inches inside diameter. Enameled wire of gauge 28 was secured in the groove for a total of 240 turns.

Calculating the inductive impedance of this coil by standard published formula 36 :

d 2 n 2 d= diameter in inches

L (microHenry)= ------- l= coil length in inches

18d+40l n= number of turns

= 1039 microHenry

AWG 28 wire is 0.01264 inch in diameter. 37 The threads cutin the coil form space each turn of wire uniformly 0.025 inches apart.

Self resonance frequency of the empty coil was found by feeding a sine wave signal from a function generator (B & K model 3011A) through a 100 kOhm resistor to the empty coil. Using an oscilloscope to monitor coil voltage, the signal generator frequency was increased from 100 kHz to 2 MHz. Maximum voltage, which indicates resonance, was found to occur at 712 kHz.

Signal source for the experiment was a 10-100 Watt adjustable output transmitter set to 7.0260 MHz (see Figure1). Automatic level control circuitry within the transmitter stabilized output power. The frequency was chosen somewhat arbitrarily during preliminary trials for a balance between overall sensitivity of the directional coupler and sensitivity of the coil circuit. Sensitivity of the directional coupler increases with frequency. However attempts to use significantly higher frequencies resulted in excessive sensitivity of the system to positioning of the cables and components, and proximity of the experimenter.

Additionally higher frequencies (above 11 MHz) resulted in impedances beyond the range of the impedance matching network.

The 7.026 MHz radio frequency signal was then fed via 52ohm solid dielectric coaxial cable (Belden 8240 RG-58/AU) through a 10 dB attenuator to the directional coupler and an impedance matching network. The directional coupler was constructed of a 11.2 inch segment of brass rod with two11.2" terminated sections of brass rod placed 0.125" distant in parallel and diametrically opposite the transmission line. The brass lines were supported by and passed through plexiglass spacers. The coupling lines were terminated by 510 ohm 1/4 watt carbon film resistors in series with 500 ohm potentiometers to provide for precise null adjustments.

Detection was by two 1N34A germanium diodes, one for each sampling line. The directional coupler assembly is mounted within a one inch square brass chassis 12 inches in length. The impedance matching network is comprised of a T-network high pass filter with adjustable capacitors and a continuously variable inductor. The output of the T-networkis fed to a 1:4 (input impedance to output impedance) ratio balanced to unbalanced toroidal transformer for connection to the balanced feed line which carries the signal to the test coil. Maximum value of inductance is 18 microHenry and the capacitors maximum values are 340 picofarad.

Research Procedure

Calibration of Equipment

The directional coupler was matched to the impedance of the transmission line by adjustment of the sampling line termination resistors. A 52 ohm non-inductive coaxial load resistor was connected to the output of the directional coupler, and the input of the coupler connected to the transmitter output. The transmitter was set for 30.0 MHz and power output of approximately 14 watts (75 volts p-p) as measured by oscilloscope. The terminating potentiometer for the reflected direction was then adjusted for 0.00 volt indication. Signal frequency was varied from 1.0 to 30.0 MHz in 1 MHz increments while the digital voltmeter was monitored to ensure no frequency dependent effects.

Connections to the coupler were swapped and the procedure repeated for adjustment of the other sampling line. After removing the terminating resistor from the line, the 10 db attenuator, impedance matching network and test coil were connected to the transmitter. At this point the impedance matching network was adjusted by trial and error until 0.012 volts reflected reading was obtained. Forward power sample reading was 1.567 volts. Once proper impedance match was found, as indicated by minimum reflected signal, no further adjustments were made.

Acquisition of Data

The media for investigation was a section of Arrow Brand "Genuine Cellulose" household sponge. No further information regarding the make-up of the sponge was given on the package, and no tests were performed to determine it sactual composition. From this was cut a section of rectangular shape with dimensions 4x1x1 inches. The sponge had been rinsed repeatedly during preliminary trials. One of the purposes of these trials was to determine a useful frequency and technique for inserting the carrier. Before each run of the experiment, the sponge was wrung damp dry and then placed in a microwave oven for further drying. Care was taken to allow the sponge to cool between each heating cycle of 30 seconds, so that the sponge would not be damaged. The sponge was dried in this manner until stiff.  Mass of the sponge was then determined by weighing on a torsion balance (Torbal model IL-11) indicating in 0.1 gram increments.

The sponge was then inserted into the foam carrier and aligned with the placement marks. The carrier, with sponge, was then inserted into the coil until centered, as indicated by a groove cut into the carrier which aligned with the edgeof the coil form. The meter is then switched to the reflected signal output from the directional coupler. This reading is recorded and the switch toggled so that the meter indicates forward signal. This indication was likewise recorded. To reduce unnecessary heating of the sponge, the carrier and sponge are immediately withdrawn from the coil. A 3 cc syringe, graduated in 0.1 cc increments was used to draw 1.0 cc purchased sodium free distilled drinking water. All water for the experiment was taken from the sameone gallon container. The water was applied to the sponge across its length through the longitudinal slit opening in the foam carrier. The water in the sponge was allowed to disperse for a minimum of 30 seconds. Then the carrier was re-inserted into the coil and reflected and forward measurements again recorded. Water was observed escaping from the sponge into the carrier after application of 21 cc total during trials 1 and 2. To avoid collection of data beyond the water capacity of the sponge, subsequent runs would cease at approximately 20 grams total mass of the wetsponge.

The procedure was repeated for a total 12 runs. Each time the sponge is dried, weighed, and the mass recorded. Recorded data were entered into a computer spreadsheet for calculation of  SWR and "wet mass", the mass of the sponge at beginning of trial, plus number of cc's water added. Data were also obtained for 0.9 percent saline solution applied to a similar size sponge cut from the same larger sponge as the distilled water runs. Saline solution was used to investigate the effect of conductivity on the system. Rather than use some arbitrary salinity, the value chosen (0.9%) is that of normal physiological saline. By simulating the conductivity of body fluids, one may infer from the data obtained, the possible application of the SWR method to biological measurements as in the manner performed by Presta et. al. 39

To prepare the saline solution, 100 grams water was weighed and put into a 1 liter container. This was repeated 9 times to provide 1 kilogram of distilled water. This method was necessitated by the 110 gram maximum capacity of the balance. To this was added 9 grams of iodine free table  salt. The container was shaken vigorously initially and 3 more times at one hour intervals, then left undisturbe dovernight.

This saline solution was used on the next sponge in the same procedure as that described for distilled water, with the exception that drying was by compression in a 4" bench vise only. First the sponge was rinsed with saline solution and compressed in the vice, this cycle then repeated twice more. The sponge was not dried by microwave oven and each trial was begun immediately after compression so that evaporation was kept at a minimum. This modification of procedure would reduce the likely increase in saline contentof the wet sponge with each following trial.  The remainder of the procedure was the same as for the preceding trial using distilled water. But drying the sponge by compression in the vice so damaged the sponge that only 3 runs were practical.

A third sponge was prepared with the same dimensions as the previous two. However, this sponge was not prepared by rinsing with saline solution, and the experiment was repeated a total of 6 times despite the extreme deterioration of the sponge.

For comparison of the effect of dielectric constant, the apparatus was rearranged so that the axis of the coil was vertical. A 3 inch square of 0.1 inch thick plexiglass was glued to the bottom end. This rearrangement necessitated a minor readjustment of the impedance matching network. After readjustment, and a 2 minute wait to insure stability, the coil form was filled to the top with distilled sodium free water. At this time the forward and reflected voltage indications were recorded. The coil form and water were then placed in a freezer for 9 hours. Upon removal from the freezer, the coil was reattached to the circuit and the SWR again determined. However, condensation soon covered the coil and connections. The coil form was warmed with a hot air gun until the ice could be pushed out in one piece. The ice was kept in the freezer until the test coil returned to room temperature (26.8 o C) as indicated by digital thermometer. At this time the cylindrical block of ice was removed from the freezer and place in the coil. The forward and reverse readings were recorded immediately.


Raw data were entered into a computer spreadsheet for calculation of SWR and wet mass (sponge mass at beginning of trial plus number of cc's added.) The spreadsheet output was used to manually enter data coordinates into AutoCad v2.18 for creation of the plots. Lotus 1-2-3 was used to perform linear regression on data obtained from trials using distilled sodium free water. A printout of the reqressionis included in the appendix. The formula calculated for aline was;

SWR = (0.2403329)(wet mass) + 0.1603431 (4)

Standard Error of Y estimate = 0.3235442

R 2 = 0.9380692

Total number of observations = 211

Standard Error of Coefficient= 0.0042714

From this the X intercept is calculated by setting SWR = 1

in equation 4 (No reflected power gives a ratio of 1).

1=0.2403329 x (wet mass) +0.1603431

1-0.1603431 = 0.2403329 x (wet mass)

3.4937 = mass

SWR calculated from data for all trials using sponges were plotted. After graphing results of the experiments with 0.9% saline solution, the relationships were obviously non-linear so no statistical analysis were performed on these data. Results of the water versus ice comparison were tabulated.

Description of Graphs

The following are descriptions of the data presented in graphical and tabular form. A discussion of the results is found in the following section. The dependent variable is represented by the horizontal axis. In each graph, this term is expressed as total wet mass of the sponge. Because the sponge did not contain the same amount of water at the beginning of each run, plotting volume of water added versus SWR would not be appropriate. Although water content measurements are usually expressed as a percent of total mass, the investigation here is concerned with the relationship between water and SWR.

Figure 4 is a plot showing voltage measured at the output of the directional coupler for the forward and reverse directions, and SWR calculated from these measurements for the first trial run using distilled sodium free water. Voltage is read from the left vertical axis and SWR is read on the right vertical axis. The horizontal axis indicates volume of water added to the sponge.

Figure 5 is a plot of SWR calculated from data obtained using distilled sodium free water. Each circle represents SWR calculation for one data point. Data from all 12 trial runs using distilled sodium free water are plotted. The X-axis is the wet mass of the sponge calculated as the sum of the mass before addition of water, plus the number of cc's water added. The Y-axis of the plot shows the SWR calculated from reflected and forward signal readings on the digital voltmeter.

Figure 6 once again shows all data points obtained using distilled sodium free water. Also shown is the prediction line determined by linear regression performed by Lotus1-2-3 computer program. In Figure 7 straight lines have been added to enclose all data points. The upper boundary is one line. The lower boundary was created with two lines intersecting at 13.5 grams. These boundary lines are shown in Figure 8. Projection lines have been added to show what values of total wet mass of the sponge can produce an SWR of 2.0 and 4.5 respectively.

Figure 9 is presented in the same manner as Figure 7. For this graph however, data are plotted only for trials 1,5, and 12. Data points are connected by lines to illustrate progression of each trial.

Figure 10 shows SWR versus wet sponge mass as calculated from data collected using 0.9% saline solution. This is plotted in the same manner as Figure 6 although note the change in scale for SWR. All three trial runs are shown. Here, in Figure 11, the same data as Figure 10 are used to plot wet mass versus the natural logarithm of SWR.

Figure 12 provides for comparison of data from distilled sodium free water and 0.9% saline solution. Scales are the same as the Figure 10, with expansion of the mass axis allowing for plotting of one complete trial of each fluid. Figure 13 presents SWR calculated from data obtained using 0.9% saline solution and the unprepared sponge. Trial sequence numbers identify individual trials for comparison.

Table 1 shows the signal samples and calculated SWR obtained when the apparatus was rearranged for measurement of distilled sodium free water and this same water after freezing. Only the first readings obtained for the ice are presented, though the SWR increased rapidly as the ice melted.

Table 1

Comparison of Ice and Distilled WaterSample Reflected Forward SWR Distilled Water 2.432 V 2.616 V 27.4Ice 1.878 V 2.290 V 10.1

Discussion of Results

Studying Figure 4 illustrates the principles of the SWR  method. First consider the graph of the reflected voltage versus water added to the sponge. The impedance matching network could not be adjusted to provide a perfect match, so that even when the coil was empty the reflected signal reading was 12 mV. With addition of water to the sponge,reflected signal indication increased. Next observe the forward reading. Although the transmitter output level was maintained by automatic internal circuitry, the indication of forward power was not constant. This is due to the change in impedance at the point of measurement. The impedance of the line at any point is a function of the characteristic impedance of the line, distance from the load, and the impedance of the load. With constant power travelling in the transmission line, an increase in the magnitude of the impedance results in an increase in the magnitude of the voltage of the signal.

Next note the graph of SWR calculated from the observations. The SWR line is straighter than either the forward or reflected curves. This illustrates that use of either reflected or forward voltage reading alone would not be as meaningful as SWR. In addition, determination of reflected signal sample alone would give erroneous results, were true forward power to change. Ideally, SWR is independent of power level. However, non-linearity of the directional coupler diodes may cause results to vary with changes in forward power level. The extent of this effect was not investigated.

By observing the plot of all SWR data points calculated from all twelve trials with distilled sodium free water, shown in Figure 5, a distinct relationship becomes apparent.Within the range of water content tested, there appears tobe a relationship of a linear nature. Figure 6 compares the linear regression line to the SWR calculated from observed data. Overall, a straight line appears to fit fairly well. However, note from this graph the discrepancy between the prediction line intercept and some observed lower values. The prediction line intercept implies that if all water were removed from the sponge its mass would be 3.49 grams. Massof the sponge before beginning the trials, at 2.35 grams, was nearly one-third lower.

Returning to the premise that the SWR method can be a tool for prediction of water content, the point of interest is the accuracy. As an investigation of accuracy, lines were found that contain all data points of observations with distilled sodium free water. These lines, shown with data points in Figure 7, are reproduced in Figure 8. As an example, an SWR measurement of 4.5:1 could have resulted from total wet sponge mass from less than 15 grams to nearly 21. A calculation of water content would involve measurement of total mass, then determination of water content as a percent of this. This example SWR measurement could have resulted from wet sponge mass differences of approximately 6 grams, which is a range of 30 %. A second example is illustrated choosing an SWR of 2:1 and once again inspecting the X axis. An SWR of 2:1 could have resulted from sponge wet masses from just under 6 to approximately 9.5 grams. This is a +/- 1.75 gram range from a value of 7.75 grams, a tolerance of +/- 23%.

The nature of the variance of the plotted SWR values may lead to its identification and reduction. Figure 9 plots data from three individual trials. Note how well trials 5 and 12 fit a smooth curve. This was typical of the majority of trial runs. Trial 1 has been included to show its abrupt change in slope at approximately 8 grams. The cause of difference in slope between runs is not known, nor is it understood. Variance in readings due to inconsistent placement of the carrier within the coil would be expected, if significant, to appear as deviations from aline or smooth curve on each trial. Inconsistent distribution of water within the sponge would be expected to exhibit a similar variance. All water used in th eexperiments came from a single one gallon container. The frequency of the transmitter is derived from a crystal oscillator locked synthesizer. Output level from the transmitter is controlled, and any residual variation should not be significant as SWR depends solely on the ratio of forward to reflected signal. However, the transmitter was turned off between trials, as sometimes drying the sponge could take more than one half hour. Although the transmitter frequency and output are stabilized, circuit losses in the impedance matching circuit, or perhaps within the test coil itself, may have caused heating of one or more components, causing a change in value. These temperature induced component drifts might explain the change in slope.

The results of the experiment with 0.9% saline solution shows that changes in fluid conductivity have dramatic effects. The three trial runs shown graphically in Figure10 exhibit an very non-linear relationship with the wet mass of the sponge. As seen in Figure 11, the wet mass has a more linear relationship with the natural logarithm of SWR .However, if a straight line approximating this relationship would be quite erroneous for low values of water. When one of the trials of saline solution is compared to a trial of distilled sodium free water, as in Figure 12, the slope is seen to be much steeper even for low amounts of saline solution. The rapid rate of increase in slope exhibited above 13 grams wet sponge mass becomes so steep at 14 grams, that it appears to be asymptotic at some value slightly greater than those investigated here. Below this point, the effect of conductivity shows that the inductance method should be suited to applications in biology or foodprocessing.

By studying the comparison of SWR between distilled sodium free water and the saline solution in Figure 12, one might initially conclude that this graph shows the differing effects of dielectric constant and conductivity. But the difference seen here necessitates exclusion of such an inference. Because of the clearly demonstrated effect of increased water conductivity, it can not be assumed that the results of the experiments using distilled sodium free water were due to dielectric effects alone. Although distilled water has very poor conductivity, its value is unknown for the water used in these experiments. Furthermore, the conductivity may have increased upon contact and reaction with the molecules of the sponge.

The water versus ice comparison was performed in an attempt to answer the conductivity vs. dielectric question. The coil, when filled with distilled sodium free water, presented an SWR of 27.4. After freezing the water, the SWR was initially 10, and was rising rapidly as the ice on theperiphery of the block melted in the coil form. As water has a dielectric constant of 80.37 and that of ice is reported to be 3.7 at 10 MHz, 38 this difference might have been used as a indicator of the effect of dielectric constant on the coil impedance. However, the ice had visible fracture lines and air bubbles. The bubbles and fractures were obviously not uniformly distributed within the block of ice. Thus the ice was not homogeneous and these factors would affect the conductance of the ice block as a whole. Therefore, the degree to which the dielectric properties of water affect the SWR can not be determined from this experiment.

This uncertainty does not, however, detract from the credibility of the reflected power method itself. The reflected power method would provide a relationship between dielectric constant, and quantity of that dielectric, in the electrostatic field of a capacitor.

The SWR method is simple. If sensitivity can be increased sufficiently, the method could be accomplished with very simple circuitry, and only two temperature sensitive components. Thus, cost, reliability and stability could be significantly better than current methods.

Future Work

The reflected power method could be applied to a capacitor type system. As much experience has been gained in these type detectors, perhaps greater accuracy and range could be perfected. The capacitance method may be necessary for moisture measurements in situations where conductivity of the fluid is not constant, or where its value is very low.

In its current configuration, this circuit required useof one switch and one meter. The two desired measurements could be reduced to one by developing electronic circuitry to output a signal corresponding to SWR directly. Perhaps the two individual signals could be converted to digital signals, then mathematically corrected for directional coupler diode non-linearity. This would eliminate the need for calculations and provide a single signal suitable for closed loop process control of a product manufacture. Accuracy of the directional coupler could also be improved which may lead to improved repeatability. Also,coil parameters may be adjusted for optimum performance. Proper design of the coil would require only a single variable capacitor to obtain the empty coil impedance match. Different frequencies might improve the range of the instrument. Measurement of intrusion by other fluids,perhaps with higher conductivity, also may be informative. Finally, with proper coil dimensions, or use of a capacitor as the test chamber, this test system might be usable for measurements of fluid volume in small animals.

The great difference between curves for distilled sodium free water and saline solution warrant further investigation. A series of experiments using incremental salinities might show possible application of this technique to conductivity measurements of fluids. Current methods for non-contact measurement of fluid conductivity requires theuse of two toroidal coils. Mass of water in a test tube can be measured accurately and this information combined with SWR could determine conductivity.

Perhaps most important work would be to increase the detector sensitivity. If the detection circuit sensitivity can be increased sufficiently, the circuit could utilize low power, single chip oscillators for a signal source. The apparatus used for this investigation required power levels on the order of one watt. The cost of such signal sources would negate the advantages of the reflected power method. Secondly, if the power requirements were to be reduced 20dB, this would enable the production of a truly portable battery powered circuit. Before systems utilizing this method can be mass produced, the radiation levels must be reduced by power reduction and proper shielding to comply with Federal Communications Commission guidelines for incidental radiators of electromagnetic radiation.


1. Clark, Wm. L., Anthony Nudo and Peter Yin. TheDetermination of Water in Storable RocketPropellants. A Comparison of Several AnalyticalTechniques . Humidity and Moisture. Measurement and Control in Science and Industry. 1965 ReinholdPublishing New York. volume 4, page 55-61

2. Besterfield, Dale Quality Control. 2nd edition.1986 Prentice-Hall, New Jersey

3. Nondestructive Evaluation; Application to Materials Processing. Conference Proceedings American Society of Metals Oct 3-4, 1983 Publishers Choice Book Mfg.Co. Mars, Pennsylvania 1984 forward, page v

4. Groover, Mikell P. Automation, Production Systems and Computer Integrated Manufacturing Prentice Hall1987

5. De Castro, Elda Hygrometric Method of Measuring Moisture Contents in Porous Materials. Humidity and  Moisture. Measurement and Control in Science and  Industry. 1965 Reinhold Publishing New York. volume 4, pages 7-12

6. Bouyoucos, George John. Plaster of Paris Block Electrical Measuring Unit for Making a Continuous Measurement of Soil Moisture under Field Conditions. Humidity and Moisture. Measurement and Control in  Science and Industry. 1965 Reinhold Publishing New York. volume 4 pages 105-112

7. Richards, L.A. A Thermocouple Psychrometer for Measuring the Relative Vapor Pressure of Water in Liquids or Porous Materials. Humidity and Moisture.  Measurement and Control in Science and Industry.  1965 Reinhold Publishing New York. volume 4, pages 13-17

8. Vos, B.H. Non-steady-state Method for the Determination of Moisture Content in Structures.  Humidity and Moisture. Measurement and Control in  Science and Industry. 1965 Reinhold Publishing New York. volume 4, pages 35-47

9. Clark, Wm. L., Anthony Nudo and Peter Yin. The Determination of Water in Storable Rocket Propellants. A Comparison of Several Analytical Techniques . Humidity and Moisture. Measurement and  Control in Science and Industry. 1965 Reinhold Publishing New York. volume 4, page 55-61

10. Geary, P J; Measurement of Moisture in Solids. A  survey based on Scientific and Technical Literature.  1970 Chiselhurst: Sira Institute

11. Benry, Ronald M. Is That Wood Dry Enough? Popular  Science July 1971 page 88

12. Nyboer, J. Electrical Impedance Plethysmography  Springfield, Il: CC Thomas, 1970 2nd edition

13. Lukaski Henry C. et al. Assessment of fat-free mass using bioelectrical impedance measurements of the human body. The American Journal of Clinical  Nutrition 41: April 85 pp 810-817

14. Handbook of Chemistry and Physics. 53rd edition 1972-1973 The Chemical Rubber Company, Cleveland, Ohio page F-75

15. ARRL Handbook 1988 . American Radio Relay League. Newington, Connecticut, page 2-13

16. Geary, P J; Measurement of Moisture in Solids. A  survey based on Scientific and Technical Literature.  1970 Chiselhurst: Sira Institute

17. Leroy, R.P. Moisture Measurements by High-frequency Currents. Humidity and Moisture. Measurement and  Control in Science and Industry. 1965 Reinhold Publishing New York. volume 4, page 135-140

18. Geary, P J Measurement of Moisture in Solids. A survey based on Scientific and Technical Literature.  1970 Chiselhurst: Sira Institute page 19

19. Ground Moisture Meter Invented by J. Fletcher Business Week May 15, 1948 page 69.

20. Fletcher, Joel E. The Use of Capacitance Methods for Determining Quantities of Materials in Mixtures. Humidity and Moisture. Measurement and Control in  Science and Industry. 1965 Reinhold Publishing New York. volume 4 page 113

21. Outwater, John O. A Portable Electronic Moisture Detector for reinforced Plastics and Its Application . Humidity and Moisture. Measurement and  Control in Science and Industry. 1965 Reinhold Publishing New York. volume 4, pages 99-104

22. Presta, Elio and Karen Segal, Bernard Gutin, Gail G. Harrison, Theodore B. Van Itallie. Comparison in Man of Total Body Electrical Conductivity and Lean Body Mass Derived from Body Density: Validation of a New Body Composition Method. Metabolism Vol 32 No. 5 (May 1983)

23. Tarjan, P.P., and McFee, R.: Electrodeless Measurements of the Effective Resistivity of the Human Torso and Head by Magnetic Induction. I.E.E.E .  Transactions on Bio-Medical Engineering , 15:266-288, 1968

24. The pH and Conductivity Handbook vol 27 Omega Engineering Inc., Stamford Connecticut. page G-3

25. The pH and Conductivity Handbook vol 27 Omega Engineering Inc., Stamford Connecticut. page G-3

26. Nyboer, J. Electrical Impedance Plethysmography CC Thomas, Springfield, Il. 1970 2nd edition page 60-61

27. Halsted, J.B. Aqueous Dielectrics Halsted Press, New York 1973 page 251

28. Geary, P J MSc Measurement of Moisture in Solids. A  survey based on Scientific and Technical Literature.  page 20. 1970 Chiselhurst: Sira Institute

29. James, W.L., You-Hsin Yen, and R.J. King: A Microwave Method for Measuring Moisture Content, Density, and Grain Angle of Wood . United States  Dept. of Agriculture, Forest Service, Research Note  FPL-0250 March 1985

30. Handbook of Chemistry and Physics. 53rd edition 1972-1973 The Chemical Rubber Company, Cleveland, Ohio page E-43

31. Leroy, R.P. Moisture Measurements by High-frequency Currents. Humidity and Moisture. Measurement and  Control in Science and Industry. 1965 Reinhold Publishing New York. volume 4, page 135-140

32. Sinnema, William Electronic Transmission Technology.  2nd edition 1988 Prentice-Hall Englewood Cliffs, N.J.

33. ARRL Handbook 1988 . American Radio Relay League. Newington, Connecticut, page 16-2

34. ARRL Handbook 1988 . American Radio Relay League. Newington, Connecticut, page 25-14

35. Reference Data for Engineers: Radio, Electronic,  Computer and Communications. 7th edition 1986 Howard W. Sams, Indianapolis

36. ARRL Handbook 1988 . American Radio Relay League. Newington, Connecticut 06111

37. Handbook of Chemistry and Physics. 53rd edition 1972-1973 The Chemical Rubber Company, Cleveland, Ohio page F-138

38. Dielectric Materials and Applications 1954 Technology Press of M.I.T. and John Wiley & Sons. New York page 301

39. Presta, Elio and Karen Segal, Bernard Gutin, Gail G. Harrison, Theodore B. Van Itallie. Comparison in Man of Total Body Electrical Conductivity and Lean Body Mass Derived from Body Density: Validation of a New Body Composition Method. Metabolism Vol 32 No.  5 (May 1983)


Graduate School Southern Illinois University

Parkland College, Champaign, Illinois; Associate of Applied Science  Communications-Broadcast Technology

Southern Illinois University at Carbondale;  Bachelor of Science Engineering Technology

Thesis Title:

Water Content Measurement by Reflected Power Method

Major Professor: Dr. Jefferson F. Lindsey III  K5AAK

Advertise Here