Desig ad Costructio of a Three-phase Digital Eergy Meter D.P.Chadima, V.G.R.G. Jayawardae, E.A.E.H. Hemachadra, I.N.Jayasekera, H.V.L.Hasaraga, D.C. Hapuarachchi (chadima@elect.mrt.ac.lk, geethagaj@gmail.com,era.hem@gmail.com,ishaivaka@gmail.com,lahiru_hasaraga@yahoo.com,diya_elect.uom@gmail.com) Departmet of Electrical Egieerig, Faculty of Egieerig, Uiversity of Moratuwa, Sri Laka. Abstract This project was doe primarily to study the operatio ad costructio of a three phase digital eergy meter. I the Sri Laka cotext maufacturig a digital eergy meter is ot a simple task as testig ad calibratio usig available equipmet is difficult. Throughout the study, the cost ad the complexity as well as usage of tedious calculatio algorithms were miimized. Keywords: accuracy, meter,commuicatio,sesor I. INTRODUCTION Digital meter maufacturig for electricity measuremets is a vast subject that has caught attetio of may professioal segmets. I moder day meterig we ca observe primitive as well as advaced measures beig used i developmet of eergy meters. This project is somewhat a research o the possibility of maufacturig eergy meters i Sri Laka, the major objective beig familiarizig with the process of meter desig ad costructio.. The electricity sector is lookig i to the possibilities of replacig the household electromechaical meters with digital meters ad cosiderig the idustrial sector this is already beig implemeted. The eed to measure electricity usage accurately has reduced the applicatio of electromechaical meters over digital meters. Thus if we ca come up with a low cost ad accurate desig the it ll be a huge opportuity for us to eter the Sri Laka market as well we ca cotribute to the ecoomical developmet. Our aim is to deliver a product that is easy i desigig ad programmig, less i cost, more accurate ad easy to commuicate with. Thus our ultimate goal is to produce somethig that is feasible i the Sri Laka cotext. Curret Sesor Curret Sesor Power Supply Eergy Measuremet Chip LCD Display Microcotroller Serial Commuicatio FIGURE 1: Data Flow of the Meter. II. DESIGN Sesig of voltage ad curret is vital i desigig a digital eergy meter. The actual currets ad voltages ca t be directly fed to the measurig apparatus ad they have to be correctly stepped dow before the aalog to digital coversio of the sigal is doe. The sesitivity ad the accuracy have to be properly chose to ehace the overall performace of the meter. A. Curret Sesig The curret through the power system caot be directly fed to the meter. Therefore a curret sesig mechaism should be used to step dow the iput curret to a desired value ad the covert it to a voltage sigal. The curret to voltage coversio is doe because the eergy measuremet chip ca oly measure a voltage value. To come up with a proper curret sesig mechaism we cosidered four available techologies i the idustry. Low Resistace Curret Shut Hall Effect Sesor Curret Trasformer (CT) Rogowski Coil Low resistace Curret shuts are the lowest cost curret sesig solutio. However they are ot widely used because of the self heatig problems. Hall Effect Sesors (ope-loop implemetatio) eed stable exteral curret sources ad exhibit a large temperature drift: thus very rarely used. Curret trasformers are the covetioal method applied i the curret sesig circuits whereas the Rogowski coils are a ewer techology. Compared with the CTs, Rogowski coils have a higher accuracy because of the air core. Aother beefit of the Rogowski coils is that they are produced with a ibuilt itegrator that outputs a voltage sigal proportioal to the curret i the primary side. Thus there is o eed of a additioal circuitry to covert the secodary curret to a proportioal voltage sigal (The iputs to the eergy measuremet chip have to be voltage sigals). Moder desigs are much cocered o digital itegrators over aalog itegrators i order to operate the meter i hostile operatig coditios ad log life-spas [1].
Therefore Rogowski coil is the best optio for this applicatio but sice it is a fairly ew techology i the idustry it was difficult fidig a maufacture supplyig i the required rage. Therefore curret trasformers were used to sese the curret. B. Sesig The voltage of the power system caot be directly fed to the Eergy measuremet chip. Therefore a suitable mechaism should be used to reduce the iput voltage to a milli volt value before takig it ito the Eergy measuremet chip. Whe decidig upo a voltage sesig mechaism three available optios were cosidered. Trasformer Resistor divider Capacitor divider FIGURE 3. Power Supply Circuit. trasformers require a lot of space ad the primary cocer i desigig a meter is reducig the space requiremets. Thus a electroic compoet based voltage reductio techique is preferred over a voltage trasformer. Therefore as the voltage sesig mechaism we used the resistor divider method due to its simplicity ad low cost. I this circuit a capacitor ad metal oxide varistor is used i additio to the resistor divider. This is used to provide surge protectio to the eergy meter. The metal oxide varistor is a voltage-depedat resistor. Its resistace decrease as the voltage icrease. The metal oxide varistor is coected i parallel with the equipmet which requires protectio which is i our case is the voltage sesig circuit. Whe there is a situatio of a overvoltage it forms a low resistace shut ad prevets ay rise i voltage i the voltage sesig circuit. C. Power Supply A cotiuous power supply is essetial for the operatio of the digital meter. The desig draws the power requiremet from the icomig lies ad i the absece of 3-phase supply; a rechargeable battery is used to power up the microcotroller, memory ad the essetial compoets. A additioal trasformer is coected to cater for high curret demads. The power is supplied from the voltage sesig circuit; just after the metal oxide varistor thus protectio from over voltages is applied. FIGURE 2. Sesig Circuit. D. Eergy Measuremet Chip The sesed values of curret ad voltage is the take i to a eergy measuremet IC which coverts the aalog sigals ito digital sigals. The Eergy measuremet IC is MCP3909. The eergy meterig IC has two distict fuctios. FIGURE 4. Implemeted Power Supply Circuit Active Power Pulse Output Waveform Output via SPI Iterface I the active power pulse output fuctio the device supplies a frequecy output proportioal to active real power ad higher frequecy output proportioal to the istataeous power for meter calibratio. I the waveform output operatio, it ca be used serially to gather 16-bit voltage chael ad a 16-bit curret chael digital output. This has aother mode where a 20-bit multiplier output is give. For the purpose of our applicatio we use dual chael output mode, where the values of voltage ad curret ca be take separately. I the dual chael output mode the 16-bit outputs of the two chaels is sychroized to form a 32-bit word. Each chael is 16 bits (15bits + sig) ad it is
represeted i 2 s complimet. The voltage chael output is the first 16 bits ad the curret chael is the followig 16 bits. I this mode the data is clocked ito the device o the risig edge of the clock sigal ad out of the device i the fallig edge. The sesed values of the curret ad voltage is take ito two chaels of the eergy meterig IC. The maximum differetial voltage iput for the voltage chael ±660 mv ad the maximum absolute voltage is ±1V. Therefore the voltage sesig circuit was desiged so that the maximum differetial voltage will ot exceed ±660 mv to prevet ay damages to the IC. The curret chael also has a differetial voltage iput of ±470 mv ad a maximum absolute voltage iput of ±1V. The desig of the curret sesig circuit esures that that the curret chael iput will ot exceed this value for the maximum curret of the system [2] E. Microcotroller The microcotroller used for the eergy meterig system is a PIC18F4520 by microchip cooperatio. The mai tasks of the microcotroller are Calculate the rms voltage, rms curret, active power, apparet power, reactive power, power factor ad the system frequecy of the coected power system. Display the calculated parameters through the LCD display. Commuicate the calculated parameters with the PC usig RS232 protocol. FIGURE 5. Mai Meterig PCB. FIGURE 6. Side-view of the Completed Hardware Module. RMS Reactive Power F. Calculatio Algorithm Active Power Apparet Power FIGURE 7: Calculatio Algorithm The calculatio algorithms are desiged to optimize the usage of the microcotroller. The eergy measuremet chip samples the iput aalog sigals at a rate of 128 samples per lie cycle (for a 50Hz sigal). Time iterval betwee two samplig poits is 156.25µs. Accordig to the data sheet of MCP3909 it takes about 3.2µs for the data to arrive at the buffer; thus we have about 150µs left for the calculatios before the ext data set is ready. The istataeous voltage ad curret values are used to calculate the Active Power. The Active Power calculated for a lie sample will be averaged ad displayed accordigly. These curret ad voltage values ca be used to calculate the root mea squared values ad those values will be used to calculate the apparet power. Whe you have active power ad the apparet power you ca calculate the reactive power ad power factor. The eergy cosumed for each phase ca be calculated by itegratig the istataeous value of active power with respect to the time iterval betwee two eergy measuremet sigals. We have already calculated the active power ad apparet power ad the time iterval is obtaied as 156.25µs. Thus eergy cosumptio (kwh) ad maximum demad (kva) ca be calculated. The istataeous values of curret ad voltage are squared ad added to the previously squared values to fid out the rms value. 2 2 2 V is,rms = V is,rms + V i 2 2 2 I is,rms = I is,rms + I i The istataeous active power is calculated usig the istataeous voltage ad curret. P active,is = V i I i After a predefied umber of lie cycles (assume ): Average active power for a lie cycle is the average of the istataeous active power that is calculated for every sample. i=0 V i I i P active,avg = The average of the square of the istataeous values of curret ad voltage is calculated ad the square root is take to fid out the root mea squared values. Curret RMS Curret Power Factor
Output (mv) V rms,cycle = i=0 2 V is,rms,i i=0 2 I is,rms,i I rms,cycle = The rms values calculated above ca be used to calculate the average apparet power per cycle. S apparet,avg = V rms,cycle I rms,cycle Whe apparet ad active power for a lie cycle is available average reactive power ad power factor for the lie cycle ca be computed. 2 2 Q reactive,avg = S appare t,avg P active,avg P active,avg Power Factor = S apparet,avg The above values are calculated for each phase ad the additio of the values i each phase is the total for the cosidered system. P active,total = P A,active,avg + P B,active,avg + P C,active,avg S active,total = S A,active,avg + S B,active,avg + S C,active,avg Q active,total = Q A,active,avg + Q B,active,avg + Q C,active,avg G. Computer Software Commuicatio software was developed usig C# programmig laguage. Other cadidate laguages were Java, Matlab, VB.et ad C++. Java serial port programmig features are ot advaced eough ad sice large data processig i the host computer is ot ecessary Matlab is out of the picture. C++ user iterface desigig is comparatively tedious ad compared with VB.et C#.et provides ehaced port programmig features ad user iterface desigig features. C# supports serial port programmig ad the ibuilt libraries ca be maipulated to commuicate with the microcotroller. Except for retrievig calculated data from the microcotroller this software ca be used to fulfill may other requiremets. It is used to view the curret settigs of the meter as well as to sed the user requiremets to the microcotroller. However this feature is limited to the advaced users ad oly the authorized users ca chage the meter parameters III. TESTING AND RESULTS The purpose our eergy meter is to measure voltage, curret, active power, reactive power, apparet power, power factor ad the frequecy of the power system. All these values are calculated by measurig the voltage ad the curret of the system. Therefore it is essetial that these values are measured very accurately to icrease the accuracy ad the reliability of the meter. Therefore lot of time was spet o desigig ad testig these iput circuits. From the testig carried o the voltage sesig circuit we foud that there is a almost liear relatioship betwee the iput voltage ad the output voltage. Therefore the accuracy of voltage sesig circuit is high. The curret sesig circuit was also tested for its accuracy. The curret variatio of the system depeds o the load coected to it at a give time. Therefore to test the accuracy of the curret sesig circuit we coected differet loads to the power system ad measured the output voltage variatio with the primary curret of the curret trasformer. 600 500 400 FIGURE 9: Software Iterface A spread sheet applicatio (GemBoxSpeadsheet) is itegrated with the mai commuicatio module to log real time data received from the microcotroller. Usig this applicatio the user could log meter data without utilizig microcotroller s EEPROM. Yet there is a limitatio to the logged data as the free versio of the plug-i used as well as Microsoft Excel allows oly 150 rows of loggig. Purchasig a commercial versio is the solutio if more data is eeded to be logged. The sole purpose of the software is achieved by providig security to sesitive data ad makig the 300 200 100 0 0 50 100 150 200 250 Iput (V) FIGURE 10. Iput vs. Output Microcotroller Software Serial Port Serial Port Moitor Commuicatio Software Display Parameters FIGURE 8. Commuicatio betwee the Microcotroller ad PC.
Output (mv) 400 350 300 250 200 150 15 bits for oe chael 215 = 32768 data values. For sigal 1 uit (660*230)/ (32768*523) = 0.0088577 V For Curret sigal 1 uit (470*30)/ (32768*330) = 0.001312256 A B. Techical Barriers 100 50 0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Primary Curret (A) FIGURE 11. Output vs. Primary Curret. These results were the aalyzed to derive the relatioship betwee the iput curret ad the output voltage.the combied circuitry ( sesig circuit, Curret sesig circuit ad the power supply circuit) was used to test the accuracy of the iput data to the eergy measuremet chip. Actual parameters of the measured system have to be liearly stepped dow i order to be fed for calculatios. I the iitial stages we foud out that Shut Resistor method is the least cost way to sese the curret. However the maximum absolute voltage limit of the measuremet chip was violated whe it was used. The we further studied the remaiig techologies ad foud out that Rogowski coils, which are the latest method of curret sesig is the most accurate method. Thus to optimize the accuracy of the module we were hutig for Rogowski coils. Rogowski coils are ot for sale i Sri Laka ad the iteratioal market prices are comparatively high. I Uited Kigdom the prices are high ad i the Chiese ad East Asia markets they are ot familiar with the air cored techologies. Fially we had to get Curret Trasformer samples from a Chiese maufacturer ad oce agai provig the lower quality of the Chiese products oe CT was mechaically destroyed while beig trasported. Still the best optio prevails to be Rogowski coils. L ma Wattmeter Vout Curret Sesig Circuit Load Sesig Circuit Power Supply Circuit +5V The measuremet chip has to be provided with reduced voltages i the rage of millivolts for proper operatio. Eve though the digital multi-meters i Machie ad Power Systems laboratories could measure i the mv rage the accuracy of the meter is very less i the rage 1mV - 50 mv. There is a small voltage shift which makes the measured values ivalid ad erroeous. The digital oscilloscopes as well show a shift aroud 10mV-20mV ad we could t measure the startig values of the curret sesig mechaism due to these reasos. Vout N FIGURE 12. Circuit Arragemet for Testig. IV. ANALYSIS A. The Smallest Uit Values The maximum curret ad voltage sigal iputs to the measuremet chip are 470mV ad 660mV respectively. These data values are digitized i to a 15 bit data stream, thus 32768 (2 15 ) ca be represeted by the measuremet chip. With help of the stadard desig values of voltage ad curret we ca calculate the miimum values that ca be sesed by the meter. It is very importace to calculate these parameters sice they directly cotribute to the system fuctioality FIGURE 13. Error Sigal of the Oscilloscope.
C. Problems i Codig We did the codig usig CCSC programmig laguage ad we ecoutered with a problem while codig. Whe the parameters were calculated we foud that the calculatio algorithms are ot providig the expected values. We aalyzed the problem ad foud out that the time take by the calculatio fuctios causes loss of data. We used CCSC laguage (A basic deviatio of C laguage) for programmig the microcotroller ad it degrades the operatio of the algorithms i time-wise. The time differece betwee two samples from the measuremet chip is 156.25µs ad the istataeous calculatios have to be doe withi this period. The calculatio program developed usig CCSC takes more time to process the data from the measuremet chip. This causes a data loss ad that causes the deviated results. If a high level machie laguage like Assembly is used the we ca optimize the time take for calculatios. REFERENCES [1] William Koo, Curret Sesig for Eergy Meterig. [2] Microchip, MCP 3909 Eergy Meterig IC with SPI Iterface ad Active Power Pulse Output.