12 BABB : A RADIO-FREQUENCY ELECTRONIC MOISTURE METER [Vol. 76 A Radio-Frequency Electronic Moisture Meter BY A. T. S. BABB (Presented at the meeting of the Society on Wednesday, February lst, 1950) A moisture meter based on change of permittivity with moisture content and utilising radio frequencies is described. Examples are given of its use for the examination of materials of widely differing texture and moisture content. The meter is shown to be accurate within 0.3 per cent. of moisture over its whole working range, which is from 1 or 2 per cent. to 60 per cent. of moisture, provided that measurements are made under the same conditions as are used for calibration. CONDUCTANCE and permittivity are properties of a substance which, as a rule, are greatly influenced by the moisture content, and many electronic meters have been designed to measure one or both of these properties and hence to indicate the moisture content of the material under examination.Meters based on the measurement of conductance are simpler than those designed to measure permittivity or capacitance, but the range of moisture content over which they are sensitive is usually limited. The conductance of the material is largely due to the solution of ionisable compounds. In many materials of low moisture content the water exists to a very great extent as “bound” water, and in these materials, for a small change of moisture content, the change of conductance is too small to be measured. When the moisture content is high, ionisation may be complete, and then a conductance meter may be insensitive to change of moisture content.A further objection is that for the accurate measurement of conductance the conditions of contact of the material with the electrodes must be reproducible, a requirement not easily satisfied if the texture of the material to be examined varies with moisture content or if it is not a homogeneous product. The conductance method could not, for example, be employed for the examination of lumpy or coarsely-chopped materials, particularly if the sample must be large in order to obtain a result that is representative of the bulk. Meters that depend on change of permittivity and that measure capacitance can be made t o indicate moisture content because the permittivity of water is 81 while that of most dry materials lies in the range 2 to 4.Provided, therefore, that the water is not combined, chemically or physically, with any other constituent of the material, permittivity should increase steadily with moisture content ; but the contribution to permittivity of “bound” water is very much less than that of “free” water, and for any material in which some of theJan., 19511 BABB : A RADIO-FREQUENCY ELECTRONIC MOISTURE METER 13 water is “bound” it would be expected that, as the moisture content is increased gradually from zero, the permittivity would increase slowly at first when the added water becomes nearly all “bound,” but more rapidly as the proportion of the added water that becomes “bound” progressively decreases. A meter for measuring capacitance should therefore be so designed that its sensitivity to change of moisture content increases as the moisture content decreases.Meters based on measurement of permittivity or capacitance are of two general types, those that employ low frequencies of the order of 60 to 5000c/s and those using radio frequencies of upwards of 100 Kc/s (the meter described in this paper utilises a frequency of 10 Mc/s). Low frequency meters are usually insensitive at low moisture contents because of the difficulty of measuring small changes of capacitance with the impedance bridge circuits they usually employ, and at high moisture contents the ohmic resistance of the sample may be affected by texture. A change of ohmic resistance that would have little or no effect on the response of a meter employing a high frequency could introduce serious errors in one operating at a low frequency.There are several radio-frequency moisture meters on the American market but, as far as is known, the only meter available in this country was designed specifically for the examination of textiles. The meter described in this paper was designed for the examination of a very wide range of materials with moisture contents as low as 1 or 2 per cent. or as high as 60 per cent., for powdered, granular or coarsely chopped products and for materials in the form of sheets, shaped pieces or wrapped packets. It was considered that a meter of such great versatility should find many applications, but it was also realised that, as distinct from an “all-purpose” meter, the need would arise in specific cases for a meter that would provide maximum sensitivity over a particular but limited moisture range.The instrument was designed with the following points in mind- (i) The provision of sample cells of varying dimensions to cater for materials of widely differing character. (ii) The ability to suit the dimensions of the sample cell to those of the articles to be tested in special cases. (iii) An electronic circuit that would be sufficiently sensitive to small changes of per- mittivity over a wide range of permittivity. (iv) An electronic circuit whose sensitivity to change of permittivity would be greatest when the permittivity was least. (v) An electronic circuit that, by the mere alteration of the values of some of its com- ponents, could, in a given instance, be made to operate with great sensitivity over any required, limited, moisture range.The “all-purpose” meter employs a frequency of about 10 Mc/s and operates with materials which have an effective permittivity between 1 and 20 and which increase the capacitance of the sample cell by 0 to 25pp.F. Only a brief description of its essential features need be given in this paper.* This has been found to be so in practice. THE ELECTRONIC CIRCUIT The circuit is shown in Fig. 1. Apart from the power pack it comprises three essential (i) T h e cell oscillator, which generates a signal whose frequency is governed solely by the capacitance of the tuned circuit. (ii) T h e frequency changer, which receives the signal from the cell oscillator and generates another signal whose frequency is governed solely by the frequency of the signal received.(iii) T h e indicator, which receives the signal from the frequency changer but which gives a visual response only if that signal has a single, pre-arranged frequency. It gives a visual indication when-and only when-the frequency of the original signal generated by the cell oscillator has a pre-arranged value, i.e., when the tuned circuit of the cell oscillator has a pre-arranged capacitance. sect ions- * It is intended to publish a more technical description of the circuit elsewhere. The meter has been named “The Kappa Moisture Meter” (Prov. Pat. No. 124752/47) and both the “all purpose” meter and meters for special applications are manufactured under licence by Messrs.Toplis Simpson and Co., Ltd.14 BABB A RADIO-FREQUENCY ELECTRONIC MOISTURE METER [Vol. 76 One component of this circuit is the sample cell, or condenser C2; when the capacitance of this condenser is increased, as when a sample is put into the cell, the capacitance of the tuned circuit is increased and the frequency of the generated signal is reduced. If the capacitance of the circuit be now gradually THE CELL OSCILLATOR- Only the method of control of capacitance need be described. E I - - - . . - - - $ .- reduced by the adjustment of a condenser, say the condenser C1A in parallel with the cell, the capacitance of the circuit will eventually reach its former value and the signal will have its former frequency. Thus, provided the meter had first been tuned to that frequency,Jan., 19511 BABB : A RADIO-FREQUENCY ELECTRONIC MOISTURE METER 15 the indicator would show that the capacitance added by way of the sample had been exactly nullified by the adjustment of ClA, and this adjustment would therefore be a measure of the capacitance of the sample. Such a method could not be adopted in practice because the added capacitance is small, and its variation with moisture content is so small that changes of the order of 0.01 ppF would have to be detected by the use of a very fragile and insufficiently stable component.Removal of capacitance by the operation of series condensers instead of parallel condensers has therefore been adopted and the design is such that the ratio of capacitance to be removed to capacitance added by way of the sample is at least 50 when the latter is 1 ppF and 12 when the latter has the maximum value of 25 ppF for which the meter provides. The removal is effected partly by the operation of the tuning condenser C8 which has a working range of 60 ppF and is ganged to a pointer which moves over a scale, and partly by a series of condensers, C4, C5 and C6, which by the operation of push-buttons remove capacitances in five increments each of 50 ppF. The push-button unit therefore functions as a range-selector switch so that, of the total capacitance to be removed, not more than 60 ppF is recorded on the scale, which is marked linearly with well-spaced graduations.The design of the circuit thus combines a very large magnification of the capacitance of the sample with an easily read scale.Two controls of this circuit are provided-the Standard control C3, operated by push- button, and the Set Standard control C9; in conjunction with the control C13, a component of the frequency changer, they are used to align the meter before use. By their adjustment the indicator is made to respond visually when the sample cell is empty and again when a standard capacity is introduced in parallel with the cell and a pre-arranged capacitance is removed by operation of the push-buttons and tuner. In this condition the meter operates correctly over its whole working range. THE FREQUENCY CHANGER- The frequency changer section contains two parts, a control oscillator of the same type as the cell oscillator and a signal mixer. The oscillator generates a signal of fixed frequency, about 10Mc/s (adjustable for alignment purposes by the Set Zero control C13).During the operation of the meter this signal, called the control frequency, is continuously fed to the mixer which also receives the signal from the cell oscillator. These two signals are electronically “mixed,” and the mixer is caused to generate a third signal whose frequency is equal to the difference between the frequencies of the two signals received. The third signal is passed to the last section of the meter, the indicator. The control frequency is made slightly greater than that generated by the cell oscillator when the cell is empty, so that a relatively small change of the cell oscillator frequency induces a relatively large change of the frequency of the third signal.For example, the “all-purpose” meter is designed to have a control frequency of 10.000 Mc/s and a cell oscillator frequency (cell empty) of 9-630 Mc/s; the third signal therefore has a frequency of 0.370 Mc/s, to which frequency the indicator is designed to respond. If the frequency of the cell oscillator decreases to 9.620 Mc/s, a change of 1 in 963, the frequency of the third signal increases to 0.380 Mc/s, a change of 1 in 37. In this way the sensitivity of the meter to change of frequency of the cell oscillator is magnified 24 times. THE INDICATOR- The indicator is designed to respond to a signal of fixed frequency (0.370 Mc/s in the “all-purpose” model) and only if the third signal has this frequency does the neon lamp respond.In practice the frequency band over which the indicator operates is about 0-02 Mc/s, from 0.360 to 0.380Mc/s, and the peak intensity of illumination can be judged to within 0.002 Mc/s. Thus the indicator shows that the frequency of the third signal lies between 0.368 and 0.372 Mc/s and therefore that the frequency of the cell oscillator signal lies between 9.632 and 9-628 Mc/s. It has not been found necessary to provide any stabilising circuits; this is because the cell and control oscillators are of the same type and, although mains fluctuation or temperature change might cause frequency drift, any drift would be common to each oscillator and the frequency difference (the third signal) would be unaffected. The meter is therefore extremely sensitive to change of frequency.16 BABB : A RADIO-FREQUENCY ELECTRONIC MOISTURE METER [Vol.76 SAMPLE CELLS To enable materials of widely different texture to be placed in the cell in a standardised manner, and for convenience and cleanliness, the sample is not put directly into the meter but into a container, made of Perspex, that slides smoothly into position between the plates of the condenser C2. According to the nature of the material, one of three sets of cells and containers should be chosen. The small container holds 100 to 200 g of material (depending on its bulk density) and is designed for use with powdered or finely granular material. The medium-sized container holds 400 to 800 g, and is for coarse granules, pellets,. tea, material in sheet form, biscuits, small tablets and similar articles. The large container is intended for larger irregularly shaped pieces, or for use when a very large sample must be taken to be representative of the bulk; this container holds 1000 to 2000 g.The small container is provided with a ‘compression block fitted with an adjustable stop so that, for materials of variable bulk density, the volume as well as the weight of the sample can, if necessary, be controlled. Within certain limits, cells and containers can be designed to suit the dimensions of shaped articles or packaged goods and to fit into the standard meter. OPERATION The meter having been aligned, the testing of a sample is performed as follows- A pre-arranged, convenient weight of sample is taken; weighing must be accurate to &O-5 per cent., or to &O-2 per cent. if the moisture content is low and the greatest accuracy is required.The sample is transferred to the container in a standardised manner and com- pressed if necessary. The container is placed in the meter and the push-button and tuner set to give a peak intensity of the indicating light. The position of the pointer on the scale card is recorded. The temperature of the sample is taken and the scale reading corrected for temperature if necessary. The moisture content is read from a calibration curve, or directly from the scale card if this has been graduated for the material under test. Tuning the meter to the peak intensity takes about one minute if all the moisture ranges have to be searched, but only a few seconds in routine work when the approximate moisture content is known.CALIBRATION- This change is governed by the permittivity of the dielectric, i.e., the sample under test, but whilst the permittivity of the sample generally increases with moisture content there is no simple relationship that enables one value to be expressed in terms of the other. Therefore, in common with all electronic moisture meters,, the meter here described must be calibrated for use with the particular material that it is required to examine. The meter is calibrated by taking three or more readings for each of a sufficient number of samples of known but different moisture contents so that a calibration curve to cover the required moisture range may be drawn. The sets of three or more readings show the degree of reproducibility of results and give an indication of the accuracy to be expected.The temperature correction is obtained by calibrating with the samples at two or three different temperatures. It will be appreciated that the routine followed in the subsequent determinations must be the same in all details as that adopted for the calibration. The scale card may be calibrated directly in terms of moisture content from the calibration curve. The meter measures indirectly the change of the capacitance of the condenser C2. PERMITTIVITY AND MOISTURE CONTENT The permittivity of the material depends mainly on its moisture content, but also on a (a) The proportion of the total moisture that is “bound” or chemically combined with the material. ( b ) The composition of the dry material.(c) The bulk density of the sample. (d) The disposition of the sample particles or pieces with respect to the condenser plates and to each other. (e) The temperature of the sample. number of other factors the most important of which are-Jan., 19511 BABB: A RADIO-FREQUENCY ELECTRONIC MOISTURE METER 17 It follows that unless these factors are constant, or change steadily with change of moisture content, the meter cannot give reliable results. It also follows that for any one material the method of filling the container must be standardised in such a way that “packing errors” are minimised. In this respect no difficulty is experienced with materials of uniform particle size unless texture varies largely with moisture content. With irregularly shaped pieces “packing errors” can be sufficiently reduced by taking large samples, and uniformly shaped articles should be packed in pre-arranged and regular formation.APPLICATIONS Without experience with similar materials it is impossible, in a particular example, to predict either the accuracy with which moisture content could be determined or the sensitivity of the meter to change of moisture content. This is because of the widely differing properties of materials-texture, composition, grain-structure, water-combining powers, etc.-and trial alone will show whether the meter can usefully be applied to the examination of a particular material. Some typical examples of the use to which the meter has so far been put are shown graphically in Fig. 2.The curves illustrate, for a great variety of materials and moisture MOISTURE CONTENT, PER CENT Fig. 2. Relationship between meter reading and moisture content. Curves for- (1) Biscuits (7) Tea (2) Spray-dried whole egg (several different (8) Fish meal brands) (9) Powdered gelatin (3) Dried ground chicory (4) Sulphamerazine granules ( 5 ) Cops of cotton (6) Cornflakes (ioj Rye grain - (11) (12) Dried chicory roots, sliced Flour of one type only - rn - rn - of several types, strong, weak, chlorinated, etc. __ + - + - ranges, that the meter is adequately sensitive to change of moisture content. As an example of a substance with a moisture content greater than 20 per cent., alumina granules gave a curve with an average slope of 7 scale divisions for 1 per cent. of moisture over the range 13 to 38 per cent.of moisture; determinations were made with an error not exceeding 0.3 per cent. of moisture over the whole range. The author wishes to thank the Directors of Messrs. J. Lyons & Co., Ltd., for permission to publish. J. LYONS & Co., LTD. THE LABORATORIES LONDON, W.1418 BABB : A RADIO-FREQUENCY ELECTRONIC MOISTURE METER [Vol. 76 DISCUSSION DR. J. D. DERMOTT HARDING asked what was the effect of the temperature of the samples on the determination of moisture by the instrument described. Could the method be applied to inorganic sub- stances, in particular t o stable inorganic salt hydrates, such as aluminium fluoride ? MR. BABB replied that a small temperature correction to the apparent moisture content had in some cases been found necessary.Apart from alumina, with which good results had been obtained, inorganic substances had not as yet been examined. DR. H. J . CALLOW asked whether the method could be applied to determine the average moisture content of a bulk sample such as a bale of jute which measured approximately 4 f t . x 2 f t . x 2 ft. and weighed about 400 lb. MR. BABB said he thought i t would be possible to design a cell for such a purpose provided there was no irregularity of shape. It might be found advisable to use a rather lower frequency with such a cell. MR. R. L. STEPHENS said that in view of the small capacitance involved, the frequency should be increased and not decreased. MR. BABB replied that the cell could be designed t o keep the capacitance about.the same as those met with in the normal cell.He had referred to the possibility of using a lower frequency, not to allow for a smaller capacitance, but to ensure that the wavelength would be many times greater than the dimensions of the cell. DR. D. W. KENT-JONES asked what the meter actually measured when used for testing cereals that contained both free and combined moisture. Would a meter which made an accurate measurement of the natural moisture in grain also measure the moisture in a freshly damped sample? DR. D. E. HAWKINS asked a similar question and stated that he had found electronic moisture meters unreliable for dried wheat, presumably because of uneven distribution of moisture in newly dried samples. MR. BABB replied that he thought he had made it clear that the meter reading obtained was governed solely by the permittivity of the sample, and any property of the sample which affected permittivity- one such property being the distribution of moisture--would affect the meter reading. If the material was not in equilibrium with its moisture or could not be brought to the same condition as obtained when the calibration curve was constructed, an error would be introduced. DR. PAGE asked whether any special packing precaxtions had to be taken for such materials as paper and biscuits. MR. BABB replied that the sample must be placed in the cell in a reproducible manner. The results on paper had been obtained with sheets cut to a suitable size, and biscuits had been stacked in regular formation in the cell. Biscuits could be ground if that were convenient. MR. H. T. SUTCLIFPE asked whether moisture could be accurately determined in a sample of soap chips taken directly from the drier and being dry on the outside and damp inside, and also in soap powder in which uncombined water is rapidly absorbed as water of crystallisation by sodium carbonate. MR. BABB referred to his previous reply on the question of uneven distribution of moisture and said that in such instances only a trial, using samples taken under working conditions, would show whether the meter reading could be accurately related to mois,ture content. Would the technique apply to biscuits of different shape and texture?