A new model of the Decade Voltage Divider, with an input resistance of 100,000 ohms, has been made available. In addition, the accuracy specification for these dividers has been improved by a factor of 2.5. The factors affecting the accuracy are discussed in this article.
The Type 1454-A Decade Voltage Divider was introduced in 1955' with accuracy specifications of 0.1% in voltage ratio ±0.000001. The intent was to take advantage of the inherently good characteristics of the Type 510 Decade Resistors and to keep the voltage divider in the same price class by avoiding expensive adjustment to closer limits. Since the tolerance on the resistance decades is ±0.05%, it is evident that the worst combinations can produce an error in the voltage ratio of twice this value, or ±0.1%. The specification limits on the voltage divider were set accordingly.
It was recognized that the catalog limit of error was very seldom approached in actual service and most instruments would readily meet tighter tolerance specifications. A recent study has shown that only a few of the component resistors are critical and that proper selection of these during assembly would enable us to guarantee a considerably higher accuracy, 0.04%, for ratios
1 Ivan G. Easton, "An Accurate Voltage Divider for DC and Audio Frequencies," General Radio Experimenter, Vol. 30, pp. 1-5, August, 1955.
above about 0.1 without appreciably increasing the price of the instrument.
The additive constant term of one part in a million in the expression for the error, however, was still the limiting factor at low settings. This could amount to 1% at a ratio of 0.0001. This error is caused by the contact resistance of the switches and is introduced in the manner shown in Figure 1. In the Kelvin-Varley circuit used in this divider, the resistors of the second decade parallel two adjacent resistors of the first decade, and so on. It will be seen that, in the present arrangement of four decades, the voltage drop of three switches in series appears in the output circuit. This causes a residual output voltage at the zero setting and an increased error at the lower outputs.
The cure for this second accuracy limitation was suggested by the fact that the switch contact resistance was found to be remarkably constant. The resistance may increase b3r a factor of two or three when the instrument is not in use but returns quickly close to its initial value after a few operations. The
accuracy of the divider at low settings can be greatly improved if a bucking voltage equal to the average switch drop is introduced into the output loop.
The contact-drop balancing arrangement is shown in Figure 2. A small resistor, R, is placed in series with the first decade at its low end and the low output terminal is connected to the high side of the resistor. The low output terminal thus differs in potential from the low input terminal by the voltage across the resistor, which is made equal to the switch contact drop. The voltage at the output terminals can be balanced to a negligible value by this arrangement. The balancing resistor is very small, about 5 milliohms, so that its presence is never noticed in ordinary use of the divider. When highest accuracy is needed at low ratios, the user must remember to keep the input and output circuits separate. The user must also remember to turn each switch back and forth several times whenever the instrument is first used. The tendency is to forget the first or second decade when these are left at the zero setting, but these contribute the most to the contact drop, since they carry more current, and must be restored to normal resistance if the compensation scheme is to be effective.
Although the selection and matching of the component resistors should insure meeting the new tolerances, a check to one part in 10*' is made of each ratio of each decade in the final inspection of the instrument.
When maximum accuracy is required, temperature effects must be allowed for and the input voltage must be reduced considerably below that corresponding to the dissipation limit of the resistors. Since all resistors are of similar construction and have more or less equal temperature coefficients, the effects of ambient temperature variations are very small. The effects from self-heating are not balanced out, however. Referring to Figure 1, it will be seen that in the first decade between points 7 and 9. which are bridged by the second decade, only half of the input current is carried. The resistors between these points will have only one-quarter of the temperature rise of the others of the decade, causing an error in the output voltage. The temperature rise of the second and following decades is much smaller and can be neglected. The temperature effect is largest at the zero position of the first decade. It has been found that, to keep the self-heating error at this first position within the specification limits, the input to the Type 1454-A Decade Voltage Divider should be limited to 120 volts. The normal dissipation limit is
Was this article helpful?