Sect. II Page 4
tuations, variations, and drift in the input signal. *2% of full scale value for any range in use. In
It displays these changes more accurately than this discussion the term full-scale, whether or not
would have been possible on the 1 KC rangeun- the scale is expanded, always relates to the fre-
3
expanded. quency value shown under the arrow marker on
the RANGE switch dial skirt. This accuracy ap-
In the example above, suppose ai50 cps variation plies only to single frequency measurement. When
was to be observed. X10 expansion could have been differential measurements are made two readings
used to present an even greater pointer excursion are implicit in the process, and the double liability
per cycle of variation than that for X3. In the X10 of error doubles the possible error.
case, the segment would contain only 100 cps (0-10
scale), and the meter pointer (700 cps) would be
placed at ?5? to allow the variation to be observed. As shown in Table 2B-1, the unexpanded scale ac-
curacy is *2% full scale for single frequency mea-
The amount of variation anticipated on the input surement and is *3.5% full scale for differential
signal governs the amount of expansion chosen. measurement. In examining the various sources for
Paragraph 2B-6 discusses instrument accuracy for error in the first two columns of the table, it is
each measurement condition. seen that all errors double in differential measure-
ment, except the circuit calibration error which
2815 ADDITIONAL SCALE CALIBRATION
is linear, always in the same direction, and constant
METHODS for single or differential frequency measurement.
In the example of paragraph 2B-4 the expanded Moving to the third column, X3, it is seen that the
scale was calibrated with the input frequency for sources for differential error are dividedby three-
differential measurement . except the phantastron timing error which remains
constant. This error is not reduced upon ex-
The accuracy of such a calibration is limited by panding because it arises in the phantastron circuit
the basic accuracy of the unexpanded scale used with line variation. The timing error varies the
for the initial measurement, Le. -i2 percent full length of the constant current pulses to produce
scale, but the measurement accuracy of change an erroneous meter indication which is expanded
in frequency is improved (see paragraph 2B-6). along with the scale expansion so that it remains
constant percentage when related to the full-scale
7
Another method of calibration, useful in random range value.
counting, for example (paragraph 2B- 7), consists
of placing both OFFSET controls fully clockwise. All other errors are related to ?pin to pin?
This action calibrates the expanded scale ?0? meter behavior and remain constant in percentage
as zero cycles, and calibrates full scale to the when related to the number of scale divisions on the
magnitude of the segment. For example, the 10 KC meter. The effect is to reduce the percentage full-
range expanded X10 with the OFFSET Controls scale error when the scale divisions are related
fully clockwise would calibrate the 0 to 10 meter back to the unexpanded full-scale value. For
scale from 0 to 1 KC. example, consider meter tracking error on the
100 KC range as *l%. This is *1/2 scale division
Calibration of expanded scale with a known fre- or * 1000 cps. Expanding the 100 KC range to
quency, such as a standard or calibrated oscil- X10, *l% then becomes *1/2 scale division or
lator, can also be accomplished to improve the *lo0 cps. Relating this 100 cps back to full-scale
accuracy of expanded scale calibration. In this value, 100 KC, meter tracking error becomes
method an accurate frequency, close to the unknown *0.1%: doubled for differential error it becomes
in magnitude, would drive the instrument to establish *0.2%. Thus, expanded scale error is determined
a convenient calibration on the expanded scale. This in all cases, except phantastron timing, by dividing
signal would be removed and the unhown signal the basic range error by the factor of expansion,
would be measured on the instrument without and then doubling this quantity to obtain total dif-
changing the OFFSET control. ferential error.
The last three columns of Table 2B-1 show maxi-
2B-6 ACCURACY OF MEASUREMENTS mum possible errors when differential measure-
ments are made with constant line voltage. In
The basic accuracy of measurement on an un- these cases the errors arising from line voltage
expanded range of the Model 500B is better than variation become zero. |