TELK OMNIKA Indonesian Journal of Electrical Engineering V ol. 12, No . 10, October l 2014, pp . 7214 7222 DOI: 10.11591/telk omnika.v12.i10.6576 7214 Loc k-in Amplifier as a Sensitive Instrument f or Biomedical Measurement : Anal ysis and Implementation Y .Abd Dja wad* 1 , J . Kiel y 2 , P . Wraith 2 , and R. Luxton 2 1 State Univ ersity of Makassar , Indonesia 2 Univ ersity of the W est of England, United Kingdom Jl. AP P ettar ani, Makassar , Indonesia, telp:0411-865894 *Corresponding author , e-mail: y asser .dja w ad@unm.ac.id Abstract A measur ing instr ument pla ys impor tant role in the biomedical measurement since the biological process in living organism gener ates v er y w eak signal. Theref ore , a reliab le and sensitiv e measur ing in- str ument is needed. In this study , a loc k-in amplifier w as analysed and tested. This paper presents an e xper iment to in v estigate the loc k-in amplifier f or biomedical measurement. An e xper im ent using RC (re- sistor capacitor) tissue model to measure the v oltage cha nge related to impedance change w as perf or med using a loc k-in amplifier to e v aluate the accur acy of the loc k-in amplifier . Three diff erent v alues of the capac- itor in the RC tissue model w ere applied regarding to sim ulate small impedance changes . The measurement results w ere compared with the theoretical calculation and an impedance measurement system. An error analysis w as conducted to in v estigate the accur acy of the measurement. The compar ison result sho w ed that impedance measurement usin g loc k-in amplifier is an eff ectiv e technique , which could ab le to measure v er y small v oltage regarding impedance change in the RC tissue model. K e yw or ds: biomedical, RC tissue model, impedance measurement, loc k-in amplifier Cop yright c 2014 Institute of Ad v anced Engineering and Science . All rights reser v ed. 1. Intr oduction A Loc k-in Amplifier (LIA) is a measur ing instr ument that measures a signal which has similar frequency and phase as the ref erence signal. The LIA can be used to measure a v er y small signal (nano v olts) and is ab le to ignore an y signals that are not synchroniz ed with it, which ma y be a thousand times larger . The LIA consists of 5 b loc ks; a signal amplifier , a p hase shifter to pro vide a ref er ence signal, a phase sensitiv e detector (PSD) or m ultiplier , a lo w pass filter (LPF) and a DC amplifier , as sho wn in Figure 1. An input signal is connected to the amplifier . A second signal is connected to the phase shifter as a ref erence signal. The output is a DC v oltage which is propor tional to the amplitude of the signal being measured. The ref erence signal m ust be similar to the source signal to ”loc k” the signal to the frequency of interest. The LIA has been used in man y applications such as measurement of signal to noise r atio of photother mal signals [1], chem- istr y e xper iments using diode laser [2], in a r ing laser gyroscope [3] and nanoelectromechanical systems [4]. In biomedical resea rch, the LIA is usually used in Impedance spectroscop y (IS) technique . IS is a char acter ization method of mater ials to obtain its electr ical proper ties using electrodes [5]. IS can be divided into tw o categor ies , electrochemical impedance spectrosco p y (EIS) and other techniques . EIS engages measurement and analysis of ionic conduction in the mater ial. EIS is used also to study of fuel cells , rechargeab le batter ies , and corrosion. The resting categor y of impedance spectroscop y is applied to study electr ical char acter istic of die lectr ic mater ial, solid or liquid non conductors , in which electr onic conduction strongly in the major ity[6]. IS has been also applied in biomedical research areas [7, 8, 9, 10, 11]. IS nor mally in v olv es tw o electrodes , counter electrode (CE) and detecting electrode (DE). The sensor detects the chang e of resistance Receiv ed J une 19, 2014; Re vised A ugust 3, 2014; Accepted A ugust 18, 2014 Evaluation Warning : The document was created with Spire.PDF for Python.
TELK OMNIKA ISSN: 2302-4046 7215 hi gh   p ass filter ph ase   sh ift e r A C am p li f i er P S D lo w pa ss filter DC   a m pl ifie r V in (t) V r ef (t) V ac( t) V psd (t) V out (t) Figure 1. Bloc k diag r am of loc k-in amplifier that consists of signal amplifier , phase shifter , PSD and lo w pass filter and capacitance of the mater ial abo v e or betw een the sensor . The basic idea is to apply an input signal via CE and gather the result on DE. These tw o electrodes with cer tain dista nce beha v es lik e a capacitor when small A C v oltage applied to the sensor . Signals with a r ange frequency are applied to the circuit to analyse the char acter istics of the mater ial. An e xample of de v eloped of IS technique is Electr ic Cell-substr ate Impedance Sensing (ECIS). The technique pro vides a v er y useful and elegant approach f or the study of cell spreading, mor phology and micromotion. Thef ore , ECIS could be used f or dr ug screening and to xicology studies in the animal testing [12]. The pr inciple of ECIS is based on Ohm’ s la w that the adhered cell can be treated as an equiv alent of an RC circuit. ECIS uses electrodes that are coated b y protein to pro vide a surf ace f or cell attachment and spreading. When t he cells begin to spread on the electrode surf ace , the cell la y er star ts to aff ect the electromagnetic field betw een the tw o electrodes since the adhere d cells beha v e lik e insulating par ticles due to their plasma membr ane [13]. This small electromagnetic field change on the electrode surf ace modifies the impedance of the capacitance component. Theref ore , se v er al rese archers [12, 14, 15] ha v e used ECIS to predict the impedance of the cells and enab le direct monitor ing of impedance changing of the cells using the LIA as measur ing instr ument. In this study the LIA has been choosen since it has some adv antages lik e it can be b uild with lo w cost, v er y sensitiv e and can remo v e noise . A RC tissue model w as used to replace the or iginal cell tissue . The measurement w as perf or med b y connecting the LIA par allel with the RC tissue model. This RC tissue model is based on an appro ximation of RC v alues of human er y- throcyte [16]. Rd char acter iz es the dielectr ic of the electrode/electrolyte . Rc is the cell resistance and t he capacitor C in this circuit represents the cell tissue . The LIA output w as read b y a PC Oscilloscope and the man ual impedance calculation w as conducted based on the LIA v oltage output. 2. Resear c h Methodology 2.1. A trigonometr y per spective of the LIA Presumab ly a signal with a frequency f 1 and phase of 1 is applied to input channel. It is assumed that signal has a v er y lo w amplitude and it m ust be amplified with gain G ac . The output of the amplifier is a signal with a specific gain,namely : V ac ( t ) = G ac V o cos ( w 1 t + 1 ) (1) In addition, a ref erence signal that has a frequency f 2 and 2 is giv en as a m ultiplicand of the output signal from amplifier , the signal is : V r ef ( t ) = E o cos ( w 2 t + 2 ) (2) The product result of the amplifier output signal and the ref erence signal is an identity product of these tw o signals and can be e xpressed as : Title of man uscr ipt is shor t and clear , implies research results (First A uthor) Evaluation Warning : The document was created with Spire.PDF for Python.
7216 ISSN: 2302-4046 V psd ( t ) = 1 2 G ac V o E o ( cos [( w 1 + w 2 ) t + ( 1 + 2 )]+ cos [( w 1 w 2 ) t + ( 1 2 )]) (3) Equation 3 sho ws that the product of tw o sin usoidal signals , which ha v e diff erent frequen- cies , produces tw o sin usoidal signals with diff erent frequencies . V psd 1 ( t ) = 1 2 G ac V o E o ( cos [( w 1 + w 2 ) t + ( 1 + 2 )]) (4) V psd 2 ( t ) = 1 2 G ac V o E o ( cos [( w 1 w 2 ) t + ( 1 2 )]) (5) If Equation 3 is applied to a LPF that has a frequency cut off smaller than ( w 1 + w 2 ) , the high frequency component (Equation 4) is atten uated and the lo w frequency component (Equation 5) is passed the LPF and the final result is sum of tw o sin usoidal signals wit h diff erent frequencies . When the frequency and phase of the input signal and the ref erence signal are similar , the Equation 3 becomes : V psd ( t ) = 1 2 G ac V o E o [1 + cos (2 w t + 2 )] (6) F rom Equation 6, it re v eals that there a re tw o signal components , A C and DC . The ampli- tude of the output signal is a half of the input signal and the frequency of A C component is twice of the amplified input signal. When a LPF which has frequency cut off smaller than frequency of A C signal is applied to the Equation 6, the A C signal is atten uated b y the LPF . The result is the sum of the DC signal and the atten uated A C signal which is a r ipple DC signal. Equation 3 demonstr ates that if the frequencies and the phases of the input and ref er- ence signal are equal, the DC component is maxim um since it produces z ero phase diff erent. If the input signal and the ref erence signal ha v e diff erent phase , the output signal does not reach maxim um output which half of the amplified input signal. It concludes that the LIA only measures signals that ha v e same frequency and phase as ref erence signal. 2.2. Instrumentation F or testing the LIA, a RC tissue model w as used as sho wn in Figure 2. The RC tissue model consists of tw o fix ed carbon resist ors 22 Ohm and a cer amic capacitor as a cell tissue which has three v alues 0.01 F , 0.1 F and 1 F . These v alues of capacitor w ere chosen to analyse the accur acy of the LIA when small impedance of the RC tissue model is changed. T o conduct the impedance measurement, the LIA w as used. The high-pass filter (HPF) w as set to ha v e a frequency cut off 0.7 Hz and the lo w pass filter (LPF) w as set to ha v e a frequency cut off 16 Hz. The HPF w as used to remo v e the DC offset of the sensor . This a v oids DC offset dr iving the chip into satur ation mode . while the LPF w as used to remo v e the noise and suppressed the A C component of the output signal. The f requency cut-off w as chosen based on the interest of the signal. A frequency cut-off of 0.7 Hz w as selected as this is sufficiently high to remo v e the DC component of the input signal. A frequency cut-off of 16 Hz w as chosen based on the obser v ation dur ing e xper iment, since it pro vides smoothest output of DC signal. T o constr uct a v oltage divider , the RC tissue model w as co nnected in ser ies with a 1.5 K Ohm resistor . The 1.5 K Ohm resistor w as designated to allo w the major ity of the v oltage to be dropped across it and lea v e only v er y small por tion of v oltage to be dropped across the RC tissue model. An A C v oltage with amplitude of 250 mV , from PICOScope (PICOScope is a real time PC-based digital oscilloscope , which has function of digital stor age oscilloscope , meter and data logger , spectr um analyser and signal gener ator), and ref erence signal of the LIA with frequency r ange from 10 Hz - 1 MHz w ere applied to the circuit. The LIA w as connected par allel with the RC tissue model to detect small v oltage drop across the RC tissue model. LIA o utput w as measured b y PicoScope . At a specific frequency v alue , the output amplitude of the LIA w as recorded and the impedance TELK OMNIKA V ol. 12, No . 10, October l 2014 : 7214 7222 Evaluation Warning : The document was created with Spire.PDF for Python.
TELK OMNIKA ISSN: 2302-4046 7217 w as cal c u lated to allo w a compar ison betw een theoretical v alues and C60 (impedance measur ing instr ument made b y Cypher instr uments Ltd) output. C Ro = 1.5 K Rd = 22 Rc= 22 cel l m od el   = Ze q Vi Vo LIA PICO scope out PC US B Figure 2. The circuit diag r am of measurement system with the RC tissue model 3. Result and Anal ysis 3.1. Experimental Results Theoretically , the cell impedance can be defined b y f ollo wing equations : Z eq = R d + R c 1 + j ! R c C Z eq = R d + R c 1 + j ! R c C : 1 j ! R c C 1 j ! R c C Z eq = R d + R c j ! R 2 c C 1 + ! 2 R 2 c C 2 (7) and can be represented in the rectangular f or m as : Z eq = R d + R c 1 + ( ! C R c ) 2 j ! C R 2 c 1 + ( ! C R c ) 2 (8) in the polar f or m can be represented as f ollo w : Magnitude : Z eq = p ( R d + R c + ! 2 R 2 c R d C 2 ) 2 + ( ! R 2 c C ) 2 1 + ( ! C R c ) 2 (9) Phase: = ar ctan ! C R 2 c R d + R c + ! 2 R 2 c R d C 2 (10) When the frequency is v er y lo w , the imaginar y par t of Equation 8 is relativ ely lo w . As the imaginar y par t of cell impedances retains only a minor v alue , the impedance of the cell model is appro ximately equal to Rc+Rd. When the frequency g r adually increases , the imaginar y par t also slo wly increases until reaching a maxim um at a specific frequency and then steadily decreases to z ero again. Thus , the v alues of the RC tissue model pro vide v er y small eff ect to the whole impedance as sho wn in Figure 3. Figure 4 sho ws the impedance change when the capacitor is changed. It sho ws that the impedance change is relativ e ly small. The g r aph re v ealed that the impedance change betw een Title of man uscr ipt is shor t and clear , implies research results (First A uthor) Evaluation Warning : The document was created with Spire.PDF for Python.
7218 ISSN: 2302-4046 Figure 3. Contr ib ution of imaginar y par t to the impedance of the RC tissue model with the capac- itor of 1 F capacitor 0.01 F and 0.1 F is v er y small which is less than 0.1 Ohm. While the impedance change f or capacitors 0.1 F and 1 F is e xponentially increased. Similar result is happened with capacito r v alues of 0.01 F and 1 F . The compar ison of impedance change w as limited until frequency around 3.3 kHz since this is the smallest frequency cut off among three v alues of capacitor . Abo v e this frequency , the impedance change increased because the smaller the capacitor v alue the bigger the frequency cut off . Figure 4. Impedance change as the capacitor v alues change un til frequency cut off around 3.3 kHz Using the LIA v oltage output, the impedance calculati on is perf or med across a v oltage divider as f ollo ws : TELK OMNIKA V ol. 12, No . 10, October l 2014 : 7214 7222 Evaluation Warning : The document was created with Spire.PDF for Python.
TELK OMNIKA ISSN: 2302-4046 7219 (a) (b) (c) Figure 5. Compar ison of the impedance measurement of the RC tissue model with capacitor of (a) 0.01 F (b) 0.1 F (c) 1 F Title of man uscr ipt is shor t and clear , implies research results (First A uthor) Evaluation Warning : The document was created with Spire.PDF for Python.
7220 ISSN: 2302-4046 Z eq = V o R o V i V o (11) F rom Figure 5, it sho ws that the LIA perf or ms adequate measurement compared with theoretical calculation and measure ment using the C60 impedance measur ing inst r ument. It also demonstr ates that the LIA has limited measurement of appro ximately 210 KHz, as be y ond this frequency the v oltage decreases r apidly and pro vides less accur ate results . The figures illustr ate that the measurement of the LIA pro vides results close to the theoretical and the C60 output in the frequency r ange from 10 Hz - 100 Hz. Abo v e this frequency , the impedance is slightly increased. Meanwhile at the slope , the measurements from the LIA pro vides impro v ed outcome . It also can be obser v ed that when the capacitor v alues are increase 10 times , the frequency cut off decreased appro ximately 10 times compared with pre vious frequency cut-off . 3.2. Err or anal ysis In this study , an error analysis w as perf or med. The error analysis w as conducted in the r ange frequency of 0 - 210 kHz since this r ange is the limited frequency r ange of the LIA. Be y ond this frequency r ange the measurement pro vide unacceptab le results . In the error analysis , the relativ e error and standard error w ere used and defined as f ollo ws : = x o x x (12) = S D p N (13) Where x o is t he measurement v alue , x is the theoretical v alue , S D is the standard de via- tion of measurement and N is the n umber of data in the measurement. Figure 6a illustr ates the relativ e error g r aph of the LIA. It sho ws that relativ e errors of measurement using capacitor 0.01 F are linear ly increased from 4% until 100 kHz and steadily decreases until frequency 210 kHz. While measurement using capacitor 0.1 F , the g r aph sho ws that the relati v e errors are almost constant until frequency 30 kHz and at this point the v alues are decreased linear ly until frequency 100 kHz and almost constant abo v e 100 kHz. Diff erent result sho w ed b y the measurement of 1 F . A spik e of relativ e error sho w ed at frequency 1.5 kHz. Abo v e this frequency , the measurement sho w ed constant v alues and increased until 210 kHz. While , Figure 6b depicted the relativ e er- ror g r aph of C60 measurement. The g r aph sho ws constant relativ e error f or measurement using capacitor 0.01 F until around 60 kHz. The error r ises abo v e 60 kHz and almost constant abo v e 100 kHz which is around 2%. The measurement using capacitor 0.1 F pro vided better relativ e error which around 2% belo w 50 kHz and constantly 0.5% abo v e 50 kHz. Similar with the LIA measurement, the result of using capacitor 1 F pro vides a spik e of relativ e error sho w ed at 1.5 kHz and decreased slo wly until 50 kHz an d after this frequency the result slo wly r ises until 210 kHz. The relativ e errors of bot h measurements presented v ar ying relativ e error . The g r aphs depicted that relativ e error mean of C60 is slightly better than the LIA, where the diff erences about 1-3%. Ho w e v er , T ab le 1 descr ibes the standard error of tw o measurements . It can be seen from T ab le 1, that both mea s u rement results pro vide similar trend which as the capacitor v alue is increased the standard error is also r ise . The error r ises with margin about 1.5 as the capacitor v alue increases 10 times . T ab le 1. Standard error of measurement 0.01uF 0.1uF 1uF LIA 0.34 1.85 3.01 C60 0.10 1.61 3.00 TELK OMNIKA V ol. 12, No . 10, October l 2014 : 7214 7222 Evaluation Warning : The document was created with Spire.PDF for Python.
TELK OMNIKA ISSN: 2302-4046 7221 (a) (b) Figure 6. Standard error of measurement using (a) the LIA (b) C60. Dotted lines are mean of the standard error 4. Conc lusion The theor y of loc k-in amplifier has been re vie w ed and an e xper iment to measure the RC tissuel model using the LIA has been perf or med. The main idea of the LIA w as descr ibed mathe- matically to gain deep understanding about ho w the LIA w or ks and to obser v e the char acter istics of the circuit. An e xper iment of impedance measurement of RC tissue model using diff erence v alues of capacitor has been perf or med. The e xper iment re v ealed that the LIA w as ab le to detect small v oltage change in the RC tissue model as the impedance change . The compar ison of the LIA with theoretical calculation and impedance measurement (C60 ) sho w ed that measurement using the LIA pro vides an accur ate measurement result which ab le to detect small impedance change . This ability is sho wn b y the error analysis , where the relativ e error of the LIA measure- ment is relativ ely small. Theref ore , this study sho w ed that LIA can be used as an alter nativ e lo w cost and an eff ectiv e measur ing instr ument f or biomedical measurement. Ref erences [1] A. Mandelis , “Signal-to-noise r atio in loc k-in amplifier synchronous detection: A gener aliz ed comm unications systems approach with applications to frequency , time , and h ybr id (r ate Title of man uscr ipt is shor t and clear , implies research results (First A uthor) Evaluation Warning : The document was created with Spire.PDF for Python.
7222 ISSN: 2302-4046 windo w) photother mal measurements , Re vie w of Scientific Instr uments , v ol. 65, pp . 3309– 3323, 1994. [2] J . Whitten, “Blue diode lasers:ne w oppor tunities in chemical education, J . Chemical Educa- tion , v ol. 78, pp . 1096–1100, 2001. [3] B . W ang, W . Zhang, Z. W ang, and P . Zhu, “Loc k-in amplifier technology in laser gyroscope nor th finder of constant r ate biasing, Mathematical Prob lems in Engineer ingl , v ol. 2013, pp . 1–11, 2013. [4] Y . Y ang, C . Callegar i, X. F eng, and K. Ekinci, “Zeptog r am-scale nanomechanical mass sens- ing, Nano Lett. , v ol. 6, pp . 583–586, 2006. [5] J . Macdonald, Impedance Spectroscop y Theor y , Exper iment and Applications , 2nd ed. Wile y-Interscience , 2005. [6] ——, “Impedance spectroscop y , Annals of Biomedical Engineer ing , v ol. 20, pp . 289–305, 1992. [7] C . Xiao , B . Lachance , G. Sunahar a, and J . Luong, “Assesment of cit oto xicity using cell- substr ate impedance sensing:concentr ation and time response function approach, Analyti- cal Chemistr y , v ol. 74, pp . 5748–5753, 2002. [8] T . Houssin, J . F ollet, A. F ollet, E. Dei-Cas , and V . Senez, “Label-free analysis of w ater- polluting par asite b y electrochemical impedance spectroscop y , Biosensors and Bioelectron- ics , v ol. 25, pp . 1122–1129, 2010. [9] M. McCo y and E. W ang, “Use of electr ic cell-substr ate impedance sensing as a tool f or quantifying cytopathic eff ect in influenza a vir us inf ected mdc k cells in real-time , Jour nal of Vir ulogical Methods , v ol. 130, pp . 157–161, 2005. [10] I. Iv ano v , “Impedance spectroscop y of human er ythrocyte membr ane: Eff ect of frequency at the spectr in denatur ation tr ansition temper ature , Bioelectrochemistr y , v ol. 78, pp . 181–185, 2010. [11] L. Ar ias , C . P err y , and L. Y ang., “Real-time electr ical impedance detection of cellular activities of or al cancer cells , Biosensors and Bioelectronics , v ol. 25, pp . 2225–2231, 2010. [12] J . Luong, “An emerging impedance sensor based on cell-protein inter actions: Applications in cell biology and analytical biochemistr y , Analytical Letters , v ol. 36, pp . 3147–3164, 2003. [13] C . K eese , N. Karr a, B . Dillon, A. Goldberg, and I. Giae v er , “In vitro mol. T o xicol , v ol. 11, pp . 83–192, 1998. [14] L. W ang, L. W ang, H. Y in, W . Xing, Z. Y u, M. Guo , and J . Cheng, “Real time , label-free monitor ing of the cell cycle with a cellular impedance sensing chip , Biosensors and Bioelec- tronics , v ol. 25, pp . 990–995, 2010. [15] C . Xiao and J . Luong, “A simple mathematical model f or electr ic cell-substr ate impedance sensing with e xtended appl ications , Biosensors and Bioelectronics , v ol. 25, pp . 1774–1780, 2010. [16] J . Bao , C . Da vis , and R. Schm ukler , “F requency domain impedance measurements of er y- throcytes , Bioph ysics , v ol. 61, pp . 1427–1434, 1992. TELK OMNIKA V ol. 12, No . 10, October l 2014 : 7214 7222 Evaluation Warning : The document was created with Spire.PDF for Python.