| Abstract | Precizne mjerne metode malih otpora predstavljaju zahtjevnu tematiku koja se istražuje u nekoliko područja znanosti. Sa stajališta tehničkih znanosti mjerenja malih otpora su značajna u više područja: elektrotehnici, građevinarstvu, strojarstvu a izrazito su bitna u medicinskoj elektronici, distribuciji električne energije te dijagnostici kvarova. U ovoj doktorskoj disertaciji istražene su krajnje mogućnosti analogno digitalnih pretvornika i različitih multimetara, te njihove uporabe u svrhu istraživanja i razvoja precizne metode za usporedbu etalona otpora malih nazivnih vrijednosti. Istraživanja su uključila razvoj i ispitivanje preciznog strujnog izvora velike nazivne struje od 9 A, te razradu i implementaciju metode za usporedbu etalona malih nazivnih vrijednosti. Osim strujnog izvora u istraživanju su se razradili i razvili i ostali programski i sklopovski elementi koji su neophodni za realizaciju metode i automatizaciju mjernog sustava. Razvijeni mjerni sustav se sastoji od uređaja visoke točnosti, pažljivo razvijanih i konstruiranih, te nakon izrade analiziranih i ispitanih. Razvijena nova mjerna metoda IGMMMO je detaljno ispitana, analizirani su mjerni rezultati, te procijenjene mjerne nesigurnosti za pojedina mjerenja za koja je pokazala nisku razinu mjerne nesigurnosti reda 10-7. Razradom ove metode podignuta je razina točnosti u području mjerenja otpora razine 1 mΩ do 1 Ω na FER-u, te je omogućena šira primjena DAQ mjernih uređaja kao zamjene za skupe voltmetre visoke razine točnosti. Također ova mjerna metoda može biti dobra podloga za razvoj budućih mjernih metoda u području mjerenja otpora. |
| Abstract (english) | Faculty of Electrical Engineering and Computing in Zagreb has an internationally accredited laboratory that is part of the Croatian Metrology Institute and holder of national standards of electrical resistance, capacitance and voltage. The laboratory has many years of experience developing high accuracy measurement methods. In this work, an automated precise measurement resistance comparison method for low resistance values in the range from 0.1 mΩ to 10 Ω is presented. The measurement method uses specially built current source, range selector, current reversal module and low cost analog to digital converters. The whole measurement procedure is controlled by LabVIEW program. The realized precise resistance measurement system has achieved precision comparable to more expensive commercial devices. The thesis is divided by chapters. The Introduction explains the tasks and goals of the doctoral work. In second chapter basic concepts of electrical resistance as a unit of SI system of units, standards of resistance and the concept of traceability is presented. In third chapter design, realization and testing of a low noise current source has been described. The careful design and the accurate selection of active and passive components have made it possible to obtain noise levels similar to those which were previously possible only by means of low noise batteries. The current source can operate with the use of mains power supply with negligible loss of quality. The current source will be used in developing the resistance comparison method for low resistance values. Today, standard resistors are compared to each other using various methods, and many of these methods use stabilized current source. In this paper design of precision DC current source intended to be used in precise resistance comparison method is described. Current source with a maximum current of 1 A has already been realized on FER, and the experience gained has been used in designing current source with higher maximum current of 9 A. Several different current source circuits were contemplated and tested, and finally one of the current source circuits has been chosen and carefully modified to obtain larger currents. The most important modification was to choose suitable power MOSFETs, however all other components were carefully selected in order to achieve required levels of low noise, precision and stability. Before building the circuit, National Instruments Multisim software was used for schematic capture and simulation in order to test different design possibilities and circuit behavior. The larger number of MOSFETs allows achieving higher currents with lower heating of each MOSFET, thus maintaining the longer stability of the source. The designed current source is a low-drift current regulator that provides constant current sink using N-channel silicon MOSFETs to drive the load. It is running off a single supply of +12 volts. The operational amplifier OP27 buffers the precision voltage reference (Linear Technology 1027) output and the resistive divider from the gate of the MOSFETs. This provides better stability with higher current levels. All eight MOSFETs are connected on the same large fined heat sink. The self-developed software for testing was made to characterize the current source on its noise level and time stability. The testing was performed by measuring the voltage drop on two precision current shunts connected in series (RL1 = 2 mΩ and RL2 = 5 mΩ), excited from current source, and analyzed with self-written software in order to test the current source as closely as possible to the final application. The voltage noise floor measurements have been performed by short circuiting the high and low inputs of DAQ with short wire, then by measuring the voltage on current shunts to determine the thermal noise of current shunt with current source off, and finally with current source providing 9 A on current shunts. The noise floor measured with these experiments yielded noise level of more than 120 dB below full scale. The conclusion is that the noise measurement depends on the applied A/D conversion hardware, while the noise level of the current source is below these levels, confirming that the current source can be used in resistance measurement application. The power spectrum of the current fluctuations through the load decreases with increasing of the current setting resistance RL squared. Several current ranges have been added for the purpose of determination of current noise floor. In order to test the stability of current source, a measurement of voltage drop on current shunts with maximum current of 9 A has been started as soon as the current source has been turned on with the temperature measurement at two MOSFETs and current setting resistor using the Epcos S863 type NTC thermistors. The settling time is approximately four minutes from the start of current source operation, largely depending on the heating of current setting resistor, while two MOSFETs heats up to around 60 C. The standard deviation of voltage ratio in two channels after stabilization time was around 1 ppm, even though the voltage levels were 20 mV, a small portion of 10 V voltage range. The standard deviation of voltage ratio is actually type A uncertainty of resistance measurement system and this result is satisfactory for the required 10 ppm of total uncertainty for low value resistance standards. The current source was then compared with Fluke 5220A trans conductance current amplifier, which states that for currents of 20 A the output changes less then ±0,005 % of output current ±200 μA in 10 minutes with constant line, load and temperature. The Fluke 5220A was driven by the LT 1027 voltage reference from the current source, the output connected to the same type of current shunts. The voltage drops on shunts were sampled simultaneously on two channels of NI 9239 for ten minutes. It was determined that the voltage ratio has standard deviation of 1.1 ppm this time confirming that current source has comparable, or slightly better time stability than Fluke 5220A. Finally, the output impedance of the current source was determined by adding the Fluke resistance standard 742A-100K with nominal value of 100 kΩ in parallel with the current source and measuring the difference of voltage drop on measured shunt with the Agilent 3458A digital multimetar with long integration time (NPLC set to 100), whose own input resistance have been already determined to be in the range of TΩ. It was thus determined that current source has output impedance of 255 MΩ. However, the influence of current source output impedance can largely be neglected for the designated application as two digital multimetars or two DAQ channels are used simultaneously. In fourth chapter an automated precise measurement resistance comparison system for low resistance values in the range from 0.1 mΩ to 10 Ω is presented. The Measurement resistance comparison method consists of automatic control device IGUS, current source IGSI, NI DAQ USB 9239 device used for voltage ratio measurement, NI DAQ USB 6008 for temperature measurement, and a personal computer to control the system. The automatic control device IGUS is capable of controlling the whole measurement process, to reverse the current polarity and to switch the resistance positions at the DAQ card channels. The controller IGUS is connected to the PC with the LPT port cable. It consists of the following modules: • Power supply module that convers voltage to the needed levels, • Measurement range module is used to regulate current through resistors • Module for current reversal has a function to change direction of the current to cancel thermoelectric voltages in the circuit • Digitizer reversal module for changing the position of the resistors in the circuit is needed to cancel the offset errors of the DAQ device. Measurement characteristics of different channels have been measured using the precise voltage calibrator Fluke 5700 and two channels that had comparable offset errors were chosen for the resistance measurement system. The whole measurement process is controlled using the LabVIEW programming environment. The NI DAQ devices 6008 and 9239 are connected using the USB, while the self-built measurement range, current reversal and module for changing the positions of the DAQ channels are controlled through LPT port and its ability to change the status of each input/output pin using LabVIEW program. The NI UltiboardTM software was used for element placement and actual look of circuit board in 3D to minimize the possible heating effects. Module for measurement ranges regulates the current of IGSI through the connectors AMRS+ i AMRS- by modifying the resistance in the current source circuit. The system is designed for the measurement of the resistance ratio of reference standard R1 and unknown resistor R2. Voltage terminals of each resistor (R+, R-) are connected alternately to the DAQ card channels that measure the voltage drop on these resistances. The NI 9239 DAQ card has four possible channels, but only two are used in each measurement. It is four-channel, 24-bit C Series module that has voltage range of +/-10,5 V at each four channels, input impedance of 1 MΩ per channel, and maximum sampling speed of 50 kS/s/ch. It also provides simultaneous analog inputs. Transfer accuracy as well as the difference between four channels has been analyzed. The channels 2 and 3 have been used for the measurement of resistor ratios as they have the best stability and drift. The current is obtained from a current source that is capable of producing current up to 9 A. In order to optimize the measurement process and synchronize all relays and measurement devices it is necessary to carefully define the cycles of time during the measurement process. The current level is defined by the settings of the current source and the range module. The full period of one measurement lasts 80 seconds to reduce the heating effects. The positive and negative half cycles lasts for 22 seconds and the number of actual samples is defined in the settings of the measurement program IGMMMO which should not exceed 8 seconds. Additional time is needed for settling time for each current reversal and DAQ channel switching. The LabVIEW front panel is used to set up the measurement process and collect all the measurement data. The results of the measurement process are recorded automatically, and the measurement uncertainty is also estimated using the LabVIEW program. In fifth chapter measured results have been analyzed with new measurement system IGMMMO. The measurement method have been tested on different standard resistors and shunts in the range of 1 mΩ - 1 Ω, using the nominal 1:1 and 1:10 ratio measurements. The results were compared with the results obtained with the high precision 8 ½ digit DVMs instead of the DAQ card. As the DAQ card has a voltage range of +/- 10 V, the voltages that are in millivolt range do not provide the needed sub-ppm precision of measured ratio. The results were comparable to the DVM method only when using the current source IGSI as it provides the needed stable high current levels for the measurement with the DAQ card when measuring the low-ohmic resistors. It was determined that the measurement method can achieve the standard deviations (type A standard uncertainties) similar to the one obtained with more expensive 8 ½ digit DVMs. Also, for higher resistances, the DAQ channel input resistance of 1 MΩ must be taken into account. The relative measurement uncertainty type A was usually uA(r) = 1⋅10-7. Measurement procedures with system IGMMMO are described in the sixth chapter. The seventh chapter is the conclusion. Design, realization and testing of precise measurement resistance comparison method for low resistance values has been described. The method used specially designed current source, controller and software developed in NI LabVIEW program. Using NI USB-9239 DAQ device the method can achieve high accuracy measurements of resistance ratios comparable to more expensive methods. The future research will utilize programmable gain amplifier as DAQ front end to better use the DAQ card capabilities and complete testing of the current source in resistance measurement system. |
