A Poor Man`s Resistance Bridge
Contents
1. thermal emf errors less than 20 nanovolts typical with minimal leakage better then 10 ohms A driver and procedure was written into the OhmRef Automated Resistance Maintenance Program 6 to take advantages of the features of the 8508A 01Reference Multimeter With the OhmRef software the operator can now select from several test methods including this new test method For those of you familiar with OhmRef you ll recognize this method as similar to Method 1 but because we are using the ratio mode the new method is called Method 1B This software allows the operator to store information for each of their resistors Any resistor on the list can then be selected to be include in the test run up to eight at a time within a 1 10 ratio The software will set up the test pattern control the scanner and meter to take the readings then compute and display the results in a report The computed value of a resistor from the previous run is stored so it can be used as the known value in the next test run This makes it easy to do build ups beginning with known values for only the first run A new feature was also added to store a run or sequence of runs so with a few clicks of the mouse a series of test runs can be repeated Test runs can be set up to run repeatedly to make it easier to run multiple tests in a day or during the night to accumulate sufficient data over a shorter period of time The resistors used in this experiment were Fluke 742A standard r2. A Poor Man s Resistance Bridge Speaker Gary Bennett Fluke Corporation PO Box 9090 Everett WA 98206 Phone 425 446 6365 FAX 425 446 5649 Email Gary Bennett fluke com Authors Gary Bennett Fluke Corporation Thomas A Marshall Data Proof 2562 Lafayette Street Santa Clara CA 95050 www DataProof com 1 Abstract Using only a single DMM scanner and software a lab can have their own automated resistance bridge capable of accuracies from sub ppm to a few ppm over its entire range This paper describes how we created a simple resistance bridge using off the shelf test equipment Software was created to take dual four terminal DMM resistance ratio readings and control a scanner to automate the process Multiple forward and reverse measurements are taken to reduce the effects of thermal EMFs and to provide statistical information This bridge can be set up to automatically scale up or down over a wide range of resistance values Results from various 1 1 and 10 1 measurements between 1 ohm and 10 kilohm will be presented 2 Introduction Bridges have been widely used for precision resistance measurements for many years Automated Current Comparator Bridges are commercially available today with the capability of providing very low uncertainties However their high cost prevents many smaller laboratories from owning these new bridges Our objective was to create a simple lower cost system for making standard resistor comparisons
3. Improving Automated Measurements for Voltage Resistance and Temperature NCX Proceedings August 199 I 3 Thomas A Marshall and James A Marshall Developing Resistance Measurement Software MSC Proceedings January 1993 4 Fluke 8508A Digital Reference Multimeter Users Manual July 2002 5 Data Proof Operating Manual amp Service Manual For 160B 164B 320B Low Thermal Scanners 1984 2005 6 Data Proof Automated Resistance Maintenance Program User Manual 1994 2005 2005 NCSL International Workshop and Symposium
4. UE EXPANDED UNCERTAINTY FROM NOMINAL OHM FOR MEAN OF DATA 1 kilohm s 4 4 PPM 1 000 0044 0 49 PPM 1k_ohm_U 4 400 PPM STANDARD DEVIATION 0 108 PPM STARTING DATE ENDING DATE HIGHEST VALUE IN DATA LOWEST VALUE IN DATA UPPER CONTROL LIMIT LOWER CONTROL LIMIT 27 Apr 2005 29 Apr 2005 4 512 PPM 4 282 PPM 2 Sigma 4 890 PPM 2 Sigma 3 910 PPM Comparison average verses certified value 0 000 PPM Figure 4 1 kQ Build Up from 100 Q 2005 NCSL International Workshop and Symposium RESISTOR DATA NOMINAL RESISTANCE 10 kilohm s ACTUAL CERTIFIED VALUE 4 2 PPM 10 000 042 EXPANDED UNCERTAINTY 0 50 PPM DATA STATISTICS DEVIATION FROM NOMINAL OHM FOR 10K_OHMU MEAN OF DATA 4 593 PPM STANDARD DEVIATION 0 044 PPM STARTING DATE 27 Apr 2005 ENDING DATE 30 Apr 2005 HIGHEST VALUE IN DATA 4 297 PPM LOWEST VALUE IN DATA 4 210 PPM R CONTROL LIMIT 2 Sigma 4 700 PPM R CONTROL LIMIT 2 Sigma 3 700 PPM UPPI LOW Cl GI Comparison average verses certified value 0 393 PPM Figure 5 10 kQ Build Up from 1 KQ 2005 NCSL International Workshop and Symposium 6 Uncertainty Analysis While the results show the build ups from 1 Q to 10 KQ this uncertainty analysis will show the uncertainties of measurements starting at 10 kQ and building down to 1 Q which gives the best uncertainties over the g
5. ace and we hope to report on them in the future 2005 NCSL International Workshop and Symposium RESISTOR DATA NOMINAL RESISTANCE 10 ohm s ACTUAL CERTIFIED VALUE 36 1 PPM 10 000361 EXPANDED UNCERTAINTY 1 37 PPM DATA STATISTICS DEVIATION FROM NOMINAL OHM FOR 10_ohm_U MEAN OF DATA 37 187 PPM STANDARD DEVIATION 0 195 PPM HIGHEST VALUE IN DATA 37 384 PPM LOWEST VALUE IN DATA 36 954 PPM UPPER CONTROL LIMIT 2 Sigma 37 47 PPM LOWER CONTROL LIMIT 2 Sigma 34 73 PPM Comparison average verses certified value 1 087 PPM Figure 2 10 Q Build Up From 1 Q 2005 NCSL International Workshop and Symposium vs ESISTOR DATA NOMINAL RESISTANCE 100 ohm s ACTUAL CERTIFIED VALUE 3 3 PPM 100 00033 EXPANDED UNCERTAINTY 0 62 PPM DATA STATISTICS J EVIATION FROM NOMINAL OHM FOR 100_ohmU MEAN OF DATA 3 254 PPM STANDARD DEVIATION 0 133 PPM HIGHEST VALUE IN DATA 3 400 PPM LOWEST VALUE IN DATA 3 015 PPM UPPER CONTROL LIMIT 2 Sigma 3 920 PPM LOWER CONTROL LIMIT 2 Sigma 2 680 PPM iE Comparison average verses certified value 0 046 PPM Figure 3 100 Build Up From 10 Q 2005 NCSL International Workshop and Symposium vs ESISTOR DATA DATA STATISTICS DEVIATION NOMINAL RESISTANCE ACTUAL CERTIFIED VAL
6. esistors Values from 1 Q to 10 kQ were used in an environment that was kept at 23 C 0 6 2005 NCSL International Workshop and Symposium 5 Results We began with a 1 Q resistor with a known traceable value All of the other resistors were treated as unknown values resistors with just the nominal value entered in the appropriate screen of the software While the value of the 10 k resistor was also known it was entered as an unknown so we could compare the results of the sequence of runs from 1 to 10K ohms with our target value For the values between 10 Q through 10 KQ the results show the measurements within 1 uQ Q of the certified values The 1 to 10 Q measurement shows results just over 1 uQO Q More work is continuing on getting better readings from 1 to 10 Q 10 Q measurements should be able to be improved by changing some parameters in the measurement process We will continue to investigate this and make improvements to the measurement process Figures 1 through 5 show the results of a series of build ups from 1 Q to 10 KQ These figures are taken from the reports available in the OhmRef Automated Resistance Maintenance Program software Information included in the reports show the measurement results along with the statistics of the measurements Overcoming difficulties with the 1 Q to 10 Q measurements caused delays in our work that did not allow us to make the planned measurements from 10 kQ to 1 MQ These measurements will take pl
7. for buildups of various resistor sets between 1 ohm and 1 Megohm that would fit in a smaller laboratories budget Our parameters were that the system would be fully automated and would require a minimum number of pieces Each piece of equipment should be off the shelf and multipurpose with other standard functions Our goal for this system was that it would be capable of making 10 to 1 measurements with about 1 part per million or better accuracies for each step While Current Comparators are certainly capable of better accuracies 1 ppm may be suitable for a majority of standards laboratories In this paper we will present the results of a system developed with the combined efforts of researchers at Fluke Corporation in Everett Washington and Data Proof in Santa Clara California All of the equipment used in this system is available off the shelf and may already be available in many standards laboratories The measurement software was developed in California and the testing and evaluation was performed in Washington State in the Fluke Primary 2005 NCSL International Workshop and Symposium Standards Laboratory A description of this measurement system and the results from our testing of the system are presented 3 Measurement Setup and Procedure This system requires only two pieces of test equipment a DMM capable of making precision resistance ratio measurements and a Low Thermal Scanner The DMM is set up to make four terminal resistance comparis
8. ght The readings are recorded as the ppm ratio to the nominal value of the highest value resistor in the run This was necessary because various resistance values may be included in the run When all the intercomparisons are complete the program computes a least squares fit of the data Values for each of the resistors are assigned based on the least squares fit results and the value s of the reference resistor s in the test If 8508A 01 2 READING OHMS multiple reference values are included the resistor values will be based on the average of these references Since only the average of the difference readings from both forward and reverse measurements is recorded the DMM needs to be stable only METHOD 1B DUAL FOUR TERMINAL CONNECTION TO DMM for the duration of a pair Figure 1 measurement rather than for the entire measurement 2005 NCSL International Workshop and Symposium Statistical monitoring of the meter readings reduces measurement time by avoiding taking more samples than are necessary Readings are taken rapidly until five successive readings fall within a preset deviation 4 Equipment The meter we used is the Fluke 8508A Reference Multimeter 4 with option 01 which has the rear input terminals in addition to the front input terminals The reason this meter was selected is because four terminal resistance measurements can be taken with both the front and rear panel input terminals at the same time This ratio m
9. ndard deviation of the six reported values This shows the excellent repeatability of the entire system Another approach would be to use the standard deviation listed on the software report The column labeled Transfer Combined Uncertainty is the combination of all the components we ve discussed so far Essentially this column has the normalized relative specification of the resistance measurement system Table 3 Final combination of uncertainty components Standard Resistor Combined Effective Expanded Uncertainty Degrees Uncertainty Degrees Uncertainty HO Q of HO Q of HO Q sro Freedom Uc Freedom m Type B Distribution Normal Normal Normal C L or C F 95 1 95 Table 3 shows the combination of the Transfer Combined Uncertainty with that of the Standard Resistor This brings the uncertainty from a relative relation to an absolute value In our example we are using the uncertainty of an ESI SR 104 10 kQ resistor calibrated by the Fluke Primary Lab as our standard This resistor is the starting point for all measurements If measuring a 1 KQ resistor the Transfer Combined Uncertainty for the 1 kQ UUT is combined with the Standard Resistor Uncertainty However if you are going to measure a 10 Q resistor then you must combine the Transfer Combined Uncertainties of the 10 Q UUT 100 Q UUT 1 kQ UUT and the Standard Resistor Below is the mathematical model for the 1 kQ measurement The components listed match those in table
10. ode has the current from the meter passing through both resistors simultaneously reducing errors caused by current fluctuations and environmental effects since both resistors are affected the same way at the same time For resistors below 20kQ a stimulus current of alternating polarity is applied through both resistors simultaneously and the potential difference measured across the resistors is scanned between the Front and Rear Terminals The alternating polarity greatly reduces the affect of thermal EMFs on the measurements Since the ratio mode uses both the front and rear terminals of the 8508A it is critical that a good range zero be performed to cancel any zero offsets in the meter A Low Thermal Scanner 5 is used to fully automate the system and to reduce offsets in the metering circuit Besides switching between resistors the scanner connects each resistor alternately to both inputs of the DMM The 164A Quad Scanner is ideally suited for this purpose because it is specifically designed for making direct four terminal comparisons between two sets of four output connectors Two conventional Low Thermal Scanners can also be used in this system The simplest way to set up two scanners is to connect one scanner to the potential terminals of the DMM and resistor and the other to the current terminals and then set them both to the same address This will function the same as a 164B Quad Scanner The Low Thermal Scanners we used contributed minimal
11. ons The four Front terminals are connected to the A Line output of a Quad Scanner and the Rear terminals are connected to the scanner B Line In this experiment two 320A Low Thermal Scanners were used instead of a Quad Scanner since these were available in the Fluke Laboratory Either arrangement can be used All resistors to be compared are then connected to the scanner inputs The software 1 2 and 3 used enables the comparison of from 1 1 up to 10 1 values For this experiment we first compared a known Ohm to an unknown 10 Ohm In the next run we compared the 10 Ohm to an unknown 100 Ohm carrying each value up to 10K ohms Whether making a run with two resistors or more than two resistors in a run all readings are made with two resistors at a time until all the resistors in the run are included in the measurement The resistors are measured first in the forward direction with one resistor connected to the front DMM terminals then the measurement is reversed so the other resistor is connected to the front terminals The DMM is set to read the ratio between the two resistors and the average of the forward and reverse reading is recorded With two resistors in the run this is repeated four times half with the first resistor in the second position If more then two resistors are to be compared the software sets up the resistors in pairs in a statistically balanced design Any number of resistors can be compared with this software between two and ei
12. r the Reference resistance measurement and another column for the UUT resistance measurement The degrees of freedom for these specifications were infinite In the Fluke Primary Standards Lab we enter 200 for any degrees of freedom that are infinite Table 2 OhmRef uncertainty components Effect of 1E 12 Q OhmRef leakage Method 1B Transfer Effective Ref UUT Degrees Type A Degrees Combined Degrees UUT HO Q uO Q of of Uncertainty of Nominal UR UuL Freedom Ua Freedom u Freedom 1 10 eo eee a ee Type B Distribution Normal Normal Normal Normal C L or C F 95 95 1 1 2005 NCSL International Workshop and Symposium Table 2 shows the uncertainty components associated with OhmRef Method 1B Data Proof Scanners have a thermal EMF specification of 50nV but is typically 20nV Because the 8508A is in the TruOhm mode for measurements below 20 kQ the 8508A reverses the current through the resistors during the measurement process reducing the thermal EMFs to negligible levels The first two columns contain the uncertainty caused by the specified leakage resistance of 1E 12 Q Again like the 8508A section the uncertainty for measurements of reference resistors and UUT resistors are listed separately The next component is the Type A component In this case we have chosen to show the standard deviation of repeated measurements For example the 100 Q UUT Type A was determined by repeating the measurement six times and determining the sta
13. reatest range The descriptions of the uncertainty components are divided into three parts The components associated with the transfer uncertainty of the 8508A The components associated with OhmRef Method 1B Finally the reference resistor and the combination of build up or build down transfers are discussed 8508A Range 2 Table 1 Fluke 8508A uncertainty components 8508A Transfer 8508A Measurement Uncertainty 20min Uncertainty Reference UUT Reference UUT wO O wO O wO O wO O Degrees of Nominal Nominal Reading Range Urm Uum Freedom 1 10 n 20 10 1 f o8 07 200 10 10 f 02 015 Type B B Distribution Normal Normal Confidence Limit C L or Coverage Factor C F 95 95 Table 1 shows the uncertainty components coming from the Fluke 8508A The 8508A is set up to use the ratio mode for these measurements so the Transfer Uncertainty was used The 8508A Transfer Specification uses both uQ Q of reading and uQ Q of range specifications These were combined and computed separately for both the reference and UUT resistors For example the 10 Q UUT is measured using the 200 Q Range which has a 0 15 uQ Q of range specification and 0 2 uQ Q of reading specification The 10 Q UUT will have an uncertainty of 3 2 uQ Q 0 15 uQ of 200 is 0 00003 Q 0 2 uQ Q of 10 is 0 000002 Q 10 1E 6 3 2 The uncertainty contribution for the 8508A is listed under the columns labeled 8508A Measurement Uncertainty one column fo
14. s 1 through 3 above The expanded uncertainty is 2 times the combined uncertainty in this case because the effective degrees of freedom of the combined uncertainty is so high Uc Combined uncertainty 2 2 2 2 2 Uc U km U im U i Uy U sop U 2 2 2 2 2 2005 NCSL International Workshop and Symposium Ue Expanded uncertainty K 2 U k 7 Evaluation Conclusion The results of these measurements were well within the 1 wQ Q goal for 10 through 10 KQ measurements and just above 1 uQ Q for the 1 Q to 10 Q measurements While the data so far is very encouraging we should exercise caution before drawing conclusions from such a small number of data points The uncertainty of the OhmRef Method 1B is about half the uncertainty of the meter itself with the major contributor of the uncertainty the transfer uncertainty of the Fluke 8508A With some work to characterize the transfer uncertainty of the 8508A in the OhmRef system the uncertainty could be reduced to improve the overall uncertainty of the system While the system s uncertainty could never approach the uncertainty of a current comparator these initial measurements lead us to believe that achieving uncertainties close to 1 pOQ Q between 1 Q and 10 KQ are possible with a little additional work References 1 Kenneth L Garcia AUTORES Automated Resistance Maintenance Program MSCProceedings January 1992 2 James A Marshall and Thomas A Marshall
Download Pdf Manuals
Related Search