sensors ISSN

Transcription

1 Sensos 203, 3, ; doi:0.3390/s Communication OPEN ACCESS sensos ISSN A Small and Slim Coaxial Pobe fo Single Rice Gain Moistue Sensing Kok Yeow You, *, Hou Kit Mun, Li Ling You 2, Jamaliah Salleh and Zulkifly Abbas Communication Engineeing Depatment, Faculty of Electical Engineeing, Univesiti Teknologi Malaysia, Skudai, 830 Joho, Malaysia; s: hkmun2@live.utm.my (H.K.M.); jsjamaliahsalleh@gmail.com (J.S.) School of Allied Health Sciences, SEGi Kota Damansaa, 4780 Selango, Malaysia; llyou@segi.edu.my Depatment of Physics, Univesiti Puta Malaysia, UPM Sedang, Selango, Malaysia; za@science.upm.edu.my * Autho to whom coespondence should be addessed; kyyou@fke.utm.my; Tel.: ; Fax: Received: 28 Decembe 202; in evised fom: 7 Mach 203 / Accepted: 8 Mach 203 / Published: 4 Mach 203 Abstact: A moistue detection of single ice gains using a slim and small open-ended coaxial pobe is pesented. The coaxial pobe is suitable fo the nondestuctive measuement of moistue values in the ice gains anging fom fom 9.5% to 26%. Empiical polynomial models ae developed to pedict the gavimetic moistue content of ice based on measued eflection coefficients using a vecto netwok analyze. The elationship between the eflection coefficient and elative pemittivity wee also ceated using a egession method and expessed in a polynomial model, whose model coefficients wee obtained by fitting the data fom Finite Element-based simulation. Besides, the designed single ice gain sample holde and expeimental set-up wee shown. The measuement of single ice gains in this study is moe pecise compaed to the measuement in conventional bulk ice gains, as the andom ai gap pesent in the bulk ice gains is excluded. Keywods: small open-ended coaxial pobe; single ice gain; measued eflection coefficient; elative pemittivity; gavimetic moistue content; tansmission line; micowave measuement techniques

2 Sensos 203, Intoduction Recently, thee has been an inceased inteest in micowave measuement techniques fo the detemination of moistue content, m.c. (o dielectic popeties) of bulk ice gains in the micowave fequencies ange [ 5]. This is due to the fact that the moistue pecentage (wet basis m.c., anging fom 9% to 26%) in the ice gains plays an impotant ole in the ice maketing and stoage aspects. In maketing, the pice of ice is dependent on the weight of the bulk ice, so accumulation of wate in the ice gain will incease the pice of ice. On the othe hand, the moistue content in the gain mass detemines the stoage duation of the ice gains, in which m.c. < 3% indicates a stoage duation of moe than 60 days [6,7]. The m.c. of ice gains can be detemined by eithe diect [8] o indiect methods [ 7]. Diect methods detemine the m.cd by emoving the moistue of the gains using oven dying methods (heating at 30 C) o chemical eaction methods (extacting wate using the eaction of iodine in sulfu dioxide). Both methods emove the moistue detemine the tue wate content by the esulting weight loss. Indiect methods, in contast, equie the measuement of an electical popety of the gain using an instument, a gain moistue mete. Fo vey low fequency (DC) measuements, the desied electical paametes ae conductance, capacitance and esistance of the ice gain. The absoption powe, esonant fequency, attenuation constant, eflection constant and tansmission constant of the ice gain ae the measued paametes of inteest in micowave fequency measuements. Changes in electical popeties (o micowave popeties) that can be diectly coelated with a change in actual m.c. of the ice gain ae obtained fom the oven dying method (diect method). Recently, indiect methods have become moe popula than diect methods due to thei apid tests and use fiendly featues. The pime consideations in the measuement of gain moistue using indiect methods ae the size of the instument senso, which is in diect contact with the ice gain samples and the accuacy of the measuements [5]. Vaious micowave waveguide methods wee poposed fo the above pupose, but some of those methods equie specify dimensions of the ice gain bulk to fit inside the given size of the waveguide [2 4]. Howeve, the ice gain bulk is composed of a mixtue of ai and ice gains. The andom distibution of ice gains and the ai gap causes a low epeatability and low pecision in the measuements. Among the mentioned methods, an open-ended waveguide method is the simplest and a nondestuctive way to measue the m.c. of ice gains. The measuement using commecial open-ended waveguides is suitable fo a specific ice gains size (width and length). This is because the waves scatteed fom the waveguide apetue would penetatd though small single ice gains as the small ice gains faed to entiely cove the apetue aea of commecial pobes. In this study, a millimete size slim open-ended coaxial pobe has been fabicated to measue the moistue in small gains. The coaxial pobe was fabicated fom a 0.86 mm oute diamete (OD) semi-igid coaxial cable equipped with a male-type SMA plug connecto. The coaxial cable was machined flat and polished to fom an open suface end. Then, the coaxial pobe was potected and coveed by a customized stainless steel sheath with a flange, as shown in Figues and 2. Hee, the main focus is put on the apetue of the pobe placed against a ice gain sample anging fom 9% m.c. to 26% m.c. Typically, the open-ended coaxial pobe is calibated by using open ai, shot teminato and liquid load. Diectivity eo, souce match eo, and fequency tacking ae coected by this

3 Sensos 203, technique. Besides that, the apid and simple calibation of the coaxial senso without the use of shot and load calibation kits wee also poposed in this study (discussed in Section 3.2). 2. Configuation and Dimensions Coaxial Senso Figue shows the side and the font sectional views of ou milimete size coaxial pobe. The font sectional apetue of coaxial pobe shows 2a = 0.20 mm diamete of inne conducto, 2b = 0.66 mm diamete of coaxial-filled Teflon and 0.86 mm diamete of the oute conducto. The coaxial-filled Teflon suppots the coaxial line between the oute conducto and the inne conducto. Both inne and oute conductos guide the popagation wave in the coaxial line. In addition, an.3 mm total diamete steel flange is used to cove the total finging field at the apetue pobe. Figue 2 shows a pictue and coss sectional stuctue of the coaxial pobe. Figue. The side sectional view and font sectional view mm 7.87 mm 3 mm 0.86 mm 0.66 mm 0.5 mm 4.8 mm 0.2 mm.3 mm Figue 2. The Coss-sectional View. Silve Teflon Stainless steel Coppe z Γ AA Γ BB

4 Sensos 203, Senso Model 3.. Reflection Coefficient Model The eflection coefficient, Γ models of the millimete coaxial pobe ae given as: n 3 n 2 n n n f n f n f n n0 n0 n0 n0 () whee the symbols δ (in unit f 3 ), β (in unit f 2 ), γ (in unit f ), and χ ae the complex coefficients fo the polynomial Equation (). The symbol f epesents the opeation fequency. The complex paametes in Equation () was obtained by fitting the polynomial coefficients with calculated values obtained fom the Finite Element Method using the COMCOL simulato ove a boad ange of pemittivity values. The complex polynomial coefficients with seven decimals fo Equation () ae listed in Table. Equation () is valid fo small coaxial pobes, satisfying the elative pemittivity, ε fom to 40 and the opeation fequency fom 0.4 GHz to 20 GHz. Compaison between the calculated and FEM simulated values fo the eflection coefficient, Γ is shown in Table 2. If the FEM simulation esults ae used as the efeence value, it is found that the pecentage of elative eo between both magnitudes of eflection coefficient will be less than %. Table. Complex Coefficient fo Equation (). δ 0 δ δ 2 δ 3 δ 4 δ 5 δ 6 δ j β j β j β j β j β j β j β j β 7 Complex Coefficients in Equation () j γ j γ j γ j γ j γ j γ j γ j γ j χ 0 + j j χ j j χ j j χ j j χ j j χ j j χ j j χ j

5 Sensos 203, Table 2. Calculated and Simulated Reflection Coefficient, Γ. f (GHz) Reflection Coefficient, Γ Reflection Coefficient, Γ Relative (ε = j 0) (ε =40 j 5) Eo (%) Equation () Simulations Equation () Simulations j j j j Relative Eo (%) j j j j j j j j Calibation Model In this study; the simplest technique of de-embedding of coaxial pobe is by extending the tansmission phase in which the phase of eflection coefficient at measuement plane; AA is extended towads the open end of coaxial pobe; BB using exponential tem of exp (j2k c z). Fist; a full one-pot calibation technique was implemented at the AA plane using a commecial HP 85052D 3.5 mm calibation kit (open; shot and load) which is only fo netwok analyze and cable eo coections. Secondly; unde the assumption of quasi-tem mode; the measued eflection coefficient; Γ AA of the sample at the plane AA can be de-embedded to the end of the pobe connecto which coincide with the calibation plane BB to give a eflection coefficient; by [9]: Γ BB 2 c Γ Γ e jk z BB ' AA (2) whee z and k c = (2πf/c) ε c ae the appaent physical length (in mete) and popagation constant of coaxial line, espectively. Symbols f, c and ε c ae the opeation measuement fequency (in Hz), velocity of light in fee space ( ms ) and elative dielectic constant fo the mateial filled in coaxial line (Teflon: ε c = 2.05), espectively. In this wok, only open standad involved in the calibation, the eflection coefficient measuements, Γ Ai fo ai at the plane AA was taken, while the standad values fo the ai eflection coefficient, Γ Ai _ FEM at the plane BB was simulated by Finite Element Method (COMSOL simulato). Fo the de-embedded pocess, the values of appaent physical length, z fo coaxial line of the pobe ae equied to detemine. Once obtaining both values ( Γ AA and ), the appaent physical length, z can be calculated as: Γ A _ FEM j Ai _ FEM z ln 2kc Ai zjz Simultaneously, the attenuation constant, α in the coaxial line can be found fom the optimized length, z as: 2 f z (4) c z (3)

6 e( BB,) m( BB,) e( BB,) m( BB,) Sensos 203, Teflon and methanol liquid have been tested to veify the accuacy of the poposed calibation method. Figue 3 shows the compaison between the calibated eflection coefficient measuement and the simulation esults. In simulation, the elative pemittivity, ε of Teflon was While, the elative pemittivity, ε, of methanol was computed by Cole-Cole model with paametes: ε s = 33.7, ε = 4.45, τ = s and α = [0]. The Cole-Cole model is: s j In compaison, the calibated measuement esults using only the open standad have given a satisfy accuacy measuement up to 5 GHz. Howeve, the noise inheent in both calibated measuements ae due to andom eos contibuted by the instuments o envionmental measuement contol. Moeove, the andom eos ae not taken into account in eithe calibation technique. The effect of the standing wave in the measuement of the open standad calibation becomes inceasingly obvious when the opeating fequency, f, is above 5 GHz. Figue 3. Vaiation in eal and imaginay pats of eflection coefficient, (Re (Γ BB ) and Im (Γ BB )) at plane BB with fequency, f at (25 ± ) C. (5) Simulation Data Polynomial Equation (This study) Open-Shot-Wate Calibation Open Calibation (This study) Teflon Teflon Fequency, f (GHz) Simulation Data Polynomial Equation (This study) Open-Shot-Wate Calibation Open Calibation (This study) Fequency, f (GHz) Simulation Data Polynomial Equation (This study) Open-Shot-Wate Calibation Open Calibation (This study) Simulation Data Polynomial Equation (This study) Open-Shot-Wate Calibation Open Calibation (This study) Methanol 0.8 Methanol Fequency, f (GHz) Fequency, f (GHz) The incident wave fom the plane AA is tansmitted to the plane BB by shifting phase of k c z, and is eflected back to input AA with the same shifting phase. Thus, the apetue eflection coefficient, Γ BB at the plane BB can be found by the phase delay of 2k c z with espect to the measued Γ AA at the plane

7 Sensos 203, AA and the tansmission line elationship is given as in Equation (2). Howeve, the tansmission line is impefect and a finging field occus nea the apetue pobe. Hence, a phase shift between the fowad wave and the eflected wave occus, and poduces the standing wave due to the supeposition between the incident wave and the eflected wave inside the coaxial line []. The standing wave effect can be ignoed if the opeation quate wave length, λ/4 in the coaxial line is lage than the physical length, z of coaxial line. Fo instance, 5 GHz of opeation fequency will give λ/4 = 5 mm which is smalle than the physical length, z 22 mm, thus, the standing wave effect was significant when the opeation fequency, f, was inceased Invese Model Fo invese solutions, the pedicted values of the elative dielectic constant, ε, of a ice gain sample is obtained by minimizing the diffeence between the measued eflection coefficient, Γ BB and Equation (), Γ by efeing to the tial function, : Data Finding the zeo outine was ealized using the MATLAB fzeo command. The initial appoximate value in the numeical pediction was equal to 5 j The m tems in Equation (6) denote the mbb weighted paamete fo atio of. The weighted paamete, m is suggested due to the e measuement with calibation using Equation (2) (only open standad: ε = ) does not conside the finging effects fo the medium loss samples. The pedicted dielectic constant, fo Teflon and methanol liquid ove fequencies 0.5 GHz to 2 GHz at oom tempeatue (25 C ) ae validated and compaed with the FEM simulation, Cole-Cole models and the Agilent 85070E dielectic pobe as shown in Figue 4. The diamete of oute adius fo the Agilent 85070E pobe (2b = 3 mm) is appoximately 4.5 times lage than the studied coaxial pobe. m m m e e BB BB BB (6) Figue 4. Vaiation in dielectic constant, methanol at (25 ± ) C. with fequency, f fo Teflon and liquid Dielectic Constant,, Teflon Dielectic Constant,, FEM Data (m = ) Cole-Cole model Open-Shot-Load calibation data (m = ) Agilent dielectic pobe Open calibation data (m = ) Open calibation data (m = 0.97) 2.75 FEM Data (m = ) Open-Shot-Load calibation data (m = ) Agilent dielectic pobe Open calibation data (m = ) Open calibation data (m = 0.97) Fequency, f (GHz) Methanol Fequency, f (GHz)

8 Sensos 203, Expeimental The measuement eflection coefficient using millimete coaxial pobe that consists of the Agilent E507C netwok analyze in the fequency ange between 0.5 GHz to 2 GHz was caied out at oom tempeatue. The open end of the coaxial line was teminated by single ice gain sample. Nomally, the length and width fo vaious kinds of ice gains was in the ange of mm and mm, espectively. Thus, the finging field (sensing aea 2b) [9] fom the coaxial pobe apetue was sufficiently coveed by the single ice sample. This can be applied based on the pinciple that the diffeent signals ae eflected fom the teminal suface of the moist gain though the coaxial opening. Jati long gain white ice gown on the fetile soil of Kedah, the Rice Bowl of Malaysia, was used as the expeimental sample. The ice gains wee divided into diffeent goups of 200 g pe goup. Each goup of gains was spayed with diffeent estimated quantities of distilled wate to achieve desied moistue levels. The bulk gains in each espective goup wee stied and sealed in a containe at 4 C fo 72 h to ensue a unifom wate distibution within the bulk gains. The gains wee conditioned to oom tempeatue fo 0 h pio to the measuements. Finally, 0 g of each goup of bulk gain ice was died in an ai convection oven at 30 C fo 24 h [8]. The aveage moistue content, m.c. (in unit %) of each goup of bulk ice gain was calculated on a wet basis as: mwate mc. % 00 (7) m m whee m wate and m dy bulk gain ae mass of wate and dy bulk gain, espectively. wate dy bulk gain Figue 5. Expeimental Set-up. Coaxial senso Connected to VNA Rice gain sample Nylon Sping Aluminium Stainless steel Thumb scew In this measuement, a specific holde was customized fo the measuement of a selected single ice gain. The customized holde has a movable nylon platfom which was mounted on a etot stand as shown in Figue 5. A single ice gain was andomly selected fom each 0 g of bulk gain and placed

9 Sensos 203, into a naow and depth concave suface at the cente of the platfom. The steel flange of the coaxial pobe was igidly enteed fom the holde edge into a naow space inteval. The ice gain on the nylon platfom was moved fowad to the apetue coaxial pobe by using a thumb scew. The concave cicula suface on platfom was used as a pobe guide to ensue that the pobe apetue is exactly touching the ice sample. The fou spings wee employed to ensue that the apetue pobe contacts fimly with the inteface single ice gain and to avoid the eos in the measuement of inteface ai gaps. 5. Results and Discussion Figue 6 shows the eflection coefficient, Γ BB, fo 0 selected single ice gains fom a bulk ice gain sample, whee the bulk has a cetain aveage wate content, measued using the millimete coaxial pobe. As known, the m.c. fo a bulk ice gain sample (which consists of thousands of single ice gains) is a statistical mean value due to the slightly diffeent m.c single of each single gain. Thus, the deviation of the actual m.c single distibution in a single ice gain is highe as compaed to the aveage m.c. of the entie bulk ice sample. This will lead to a scatteed measuement of the eflection coefficient, Γ BB, of the single gain in efeing to the aveage m.c. of the bulk sample, as shown in Figue 6. Howeve, Figue 6 does show a significant change of the measued eflection coefficient, Γ BB of a single ice gain with the m.c. of bulk ice gain. The black solid line in Figue 6 is a egession fitting line fom the aveage of measued eflection coefficient data (blue point-line). The aveage of the measued eflection coefficients was obtained fom the 0 data points of the measued eflection coefficients fo each bulk moistue content, m.c. The ice moistue calculations ae based on a gavimetic method, thus diffeential effects of density between the bulk ice and the single ice gain in the measuement wee ignoed. In this study, the 2.44 GHz and the 5.8 GHz fequencies wee chosen due to the fact those fequencies coespond to a fee unlicensed band which is specifically fo industial, scientific and medical (ISM) measuement puposes. The 0.02 GHz band was also chosen, since the dielectic loss fo the wate is geate at aound 0 GHz and thus povides a compaative appoach fo the ice measuements at highe fequencies. Figue 6. The elationship between eflection coefficient (magnitude, Γ BB and phase, ø) and the moistue content, m.c at 2.44 GHz, 5.8 GHz and 0.02 GHz, espectively BB, f = 2.44 GHz (ad) f = 2.44 GHz Moistue Content, m.c (%) Moistue Content, m.c (%)

10 Sensos 203, Figue 6. Cont BB, (ad) f = 5.8 GHz Moistue Content, m.c (%) f = 5.8 GHz Moistue Content, m.c (%) BB, f = 0.07 GHz Moistue Content, m.c (%) The polynomial egessions fo the eflection coefficient, Γ BB (magnitude, Γ BB and phase, ø) with espect to m.c. (in % units), ae given in Equation (8) as listed in Table 3. Subsequently, Equation (8) was used by the coaxial pobe to pedict the m.c. (in % units), in a ice gain. It was found that, at highe fequencies, the point of measued eflection coefficient, Γ BB shows a lage vaiation of m.c. (in % units) of ice gains. The vaiations in elative dielectic constant,, of ice gains with the pecentage of m.c. at 2.44 GHz, 5.8 GHz and 0.02 GHz, espectively, ae plotted in Figue 7(a). The solid line of elative dielectic constant,, in Figue 7(a) was the invese of the eflection coefficient and efes to the tial function of Equation (6) with m =. The eal pat, Re(Γ BB ) and imaginay pat, Im(Γ BB ) in Equation (6) wee calculated by using elationship of Re(Γ BB )+jim(γ BB )= Γ BB exp(jø), whee the values of Γ BB and Ø wee obtained fom Equation (8), while, the efeence values of Re(Γ) and Im (Γ) in Equation (6) wee computed using Equation (). The measuement of dielectic constant points,, of the single ice gain in Figue 7(a) wee caied out using the studied coaxial pobe with the Agilent E softwae in ode to veify the data obtained fom the dielectic invesion wok. The toleance, Δ of dielectic constant pediction between both techniques is shown in Figue 7(b), which gives the maximum deviation, Δ 2 fo all fequencies anging 9.5% m.c. to 28% m.c. (ad) f = 0.07 GHz Moistue Content, m.c (%)

11 Absolute Eo,, Sensos 203, Table 3. Polynomial functions fo the eflection coefficient magnitude, Γ BB, and phase, ø, (in ad units) espect to moistue content, m.c., of the bulk ice (in unit%) Fo f = 2.44 GHz, Γ BB = m.c m.c , R 2 = (8a) Ø = m.c m.c m.c , R 2 = Fo f = 5.8 GHz, (8b) Γ BB = m.c m.c , R 2 = (8c) Ø = m.c m.c m.c , R 2 = Fo f = 0.07 GHz, (8d) Γ BB = m.c m.c , R 2 = (8e) Ø = m.c m.c m.c , R 2 = (8f) The symbol m.c efes to the pecentage of bulk ice moistue content. Figue 7. (a) The vaiations in of single ice gain with its bulk moistue content, m.c. at 2.44 GHz, 5.8 GHz and 0.02 GHz, espectively. (b) The absolute deviation, Δ of dielectic constant pediction between the use of studied invesion technique and the Agilent 85070E softwae computation. Dielectic Constant,, GHz (Using tial function (6)) 5.8 GHz (Using tial function (6)) 2.44 GHz (Using tial function (6)) 0.02 GHz (Using Agilent 85070E softwae) 5.8 GHz (Using Agilent 85070E softwae) 2.44 GHz(Using Agilent 85070E softwae) GHz 5.8 GHz 2.44 GHz Conclusions Moistue Content, m.c (%) (a) Moistue Content, m.c (%) (b) The poposed coaxial pobe has a small sensing aea which coves the size of single ice gain and povides a nondestuctive and eal time moistue measuement fo single ice gains. Moeove, the single gain measuement does not depend on the bulk density of the ice gains, thus the uncetainty

12 Sensos 203, of bulk density in the ice measuement (due to diffeent ates of boken ice in the bulk gain) can be ignoed. In this study, moistue and dielectic models wee ceated to suit the studied coaxial pobe. The poposed simple de-embedding technique povides a apid and low cost calibation pocedue fo the coaxial pobe. Howeve, the technique does not conside the systematic and andom noises along the coaxial line, and thus it is suitable fo a shot coaxial pobe, (z<λ/4) due to the significant standing wave that is poduced inside the long coaxial line. Acknowledgments This study was suppoted by the Reseach Univesity Gant (GUP) fom Univesiti Teknologi Malaysia unde poject numbe Q.J J55. Refeences. Yagihaa, S.; Oyama, M.; Inoue, A.; Asano, M.; Sudo, S.; Shinyashiki, N. Dielectic elaxation measuement and analysis of esticted wate stuctue in ice kenels. Meas. Sci. Technol. 2007, 8, You, T.S.; Nelson, S.O. Micowave dielectic popeties of ice kenels. J. Micow. Powe Electomagn. Enegy 988, 23, Jafai, F.; Khalid, K.; Yusoff, W.M.; Hassan, J. The analysis and design of multi-laye micostip moistue senso fo ice gain. Biosyst. Eng. 200, 06, Kim, K.B.; Kim, J.H.; Lee, S.S.; Noh, S.H. Measuement of gain moistue content using micowave attenuation at 0.5 ghz and moistue density. IEEE Tans. Instum. Meas. 2002, 5, Nelson, S.O.; Kandala, C.V.K.; Lawence, K.C. Moistue detemination in single gain kenels and nuts by f impedance measuements. IEEE Tans. Instum. Meas. 992, 4, You, K.Y.; Salleh, J.; Abbas, Z.; You, L.L. Cylindical Slot Antennas fo Monitoing the Quality of Milled Rice. In Poceedings of the PIERS, Suzhou, China, 2 6 Septembe 20; pp Thaku, S.K.P.; Kolmes, W.S. Pemittivity of Rice Gain fom Electomagnetic Scatteing. In Poceedings of the 4th Intenational Confeence on Electomagnetic Wave Inteaction with Wate and Moist Substance, Weime, Gemany, 3 6 May 200, pp Moistue Measuements-Ungound Gain and Seeds; ASABE Standads S352.; ASABE: St. Joseph, MI, USA, You, K.Y.; Abbas, Z.; Khalid, K. Application of micowave moistue senso fo detemination of oil palm fuit ipeness. Meas. Sci. Rev. 200, 0, Blackham, D.V.; Pollad, R.D. An impoved technique fo pemittivity measuements using a coaxial pobe. IEEE Tans. Instum. Meas. 997, 46, You, K.Y.; Salleh, J.; Abbas, Z. Effects of length and diamete of open-ended coaxial senso on its eflection coefficient. RadioEngineeing 202, 2, by the authos; licensee MDPI, Basel, Switzeland. This aticle is an open access aticle distibuted unde the tems and conditions of the Ceative Commons Attibution license (