A variety of methods are available for making automated resistance measurements. There are three methods available in the OhmRef software. Each method is capable of achieving parts-per-million results, and some can do better than one ppm depending upon the resistance level. The method chosen will depend upon the equipment available, resistance level and accuracy required. Each method can be used to scale resistors over a range of 10 to 1.

OhmRef provides very good results with a low cost system using equipment already available in most laboratories. The program is easy to setup and use.

Method 1: Direct four-terminal DMM connection
For this method the DMM is set to the ohms function and makes four-terminal connections through the scanner to measure the resistors directly. This method does not require an external source; and thus, is simplest method, requiring only a Low Thermal Scanner and an 8 digit digital multimeter. The method is useful for measuring higher values of resistance (above 100 kilohms) where the DMM loading reduces the accuracy for methods 2 and 3.
It accommodates any value of resistor between 0.1 milliohm and 100 megohms. The following describes the three methods used in OhmRef with typical results that can be achieved by each method and recommended ranges.

For all three methods the program creates a statistically balanced design for the resistors being compared. Any number of resistors can be compared with OhmRef between two and eight. The resistors are measured in pairs, and the difference between each pair is recorded. The software will control the scanner and source, and get readings from the multimeter.


The voltage or current to be applied is determined from the current limit entered on the OhmRef setup screen for each resistor. 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 fit calculation and the average of the values of the reference resistors in the test.

Readings are taken as rapidly as possible for each pair of resistors in the design to reduce errors caused by instabilities in the current source. Only the average of the difference readings from both forward and reverse measurements is recorded. In this way, the source needs to be stable for the duration of a pair measurement rather than for the entire measurement. Statistical monitoring of the meter readings is used to reduce measurement time by not taking more samples than necessary. Readings are taken rapidly until five successive readings fall within a preset deviation.

Method 2: Direct connection using external source
This method uses an external current source that provides several advantages over the first method. First, the source polarity can be reversed which reduces errors caused by circuit thermals and meter zero offset. Also a higher current can be used which allows greater resolution. With this method the multimeter is set to measure voltage rather than resistance. In the voltage mode scaling between different values of resistors can be done with greater accuracy. Be sure to consider the loading effect of the multimeter input impedance when measuring resistors above 100 kilohms.

OhmRef limits the maximum current to 100 milliamps when Data Proof scanners are used. Although the relay manufacturer specifies a maximum current limit of 0.5 amps, life tests run at that level have shown contact discoloration which could lead to increased thermal offsets. No such effect was noted at the 100 milliamp level. For higher current operation  method 3 is recommended.

Page 8



For all three methods OhmRef measures pairs of resistors and records the differences. When all readings are completed OhmRef does a least-squares-fit calculation to find best-fit values for all the resistors in the test. The best-fit values are based on the least-squares-fit and the reference values of the traceable units in the test. A report can be printed that shows the difference readings along with the best-fit resistor values. The report is similar to the report shown in an earlier section of this paper titled VRMP: Voltage Maintenance Program. Also as in the VRMP the data can be stored in a data file so that control charts and plots can be made at a later time. The stored data can also be exported to a DOS file for analysis using
Method 3: Series connection using external source
In this method all the resistors in the test are hard-wired together with the current source. The current path is not switched through the scanner. This has the advantage that any current level can be used limited only by the source. Also the current is not interrupted by the switching process. The meter reads the voltage drop across the resistors through the scanner. During the measurement process the current source is electrically reversed to reduce the effects of circuit thermals. This method is useful for lower value resistors where higher currents are required. However it has the disadvantage that the resistors to be tested must be wired together for each test. Whereas with methods 1 and 2 any resistors connected to the scanner can be compared simply by choosing them from the program startup screen .
a spread sheet program.

Standard resistors can be compared in an automated statistically controlled process. By using the scanner to make many measurements rapidly, and by reversing the source and resistance connections, system errors are greatly reduced. Typical uncertainties for each method are as follows.

Method 1 can be used to compare nominally equal resistors over the range of 1 ohm to 100 megohm with uncertainties in the range of 1 to 10 ppm. The uncertainty increases near 1 ohm due to the meter resolution, and above 1 megohm due to leakages in the meter circuit.


Method 2 is recommended over the range of 1 ohm to 1 megohm. This method usually gives better results than method 1 up to 1 megohm because higher currents can be used. Uncertainties in the midrange can be about 0.5 ppm. Uncertainties when scaling 10 to 1 will be in the range of 1 ppm to 7 ppm.

Method 3 works well from 1 milliohm to 1 megohm. It achieves similar results over the midrange and better results at the lower values because more power can be applied to the resistors. Uncertainties of about 0.5 ppm can be obtained when measuring nominally equal values, and about 1 ppm when scaling 10 to 1.

Page 9

Home Page Contents (Page1) Previous Section Next Section