4 example 2: calibrating an sprt, Example 2: calibrating an sprt – Fluke 1595A User Manual

Page 22

Advertising
background image

1594A/1595A Super-Thermometer

Specifications

12

of the SPRT at 157 °C which is 0.1 Ω/°C (see tip above). This results in a standard temperature uncertainty of

0.001028 °C.

2.2.5.3.11 Measurement Noise at 157 °C

During measurement at 157 °C, the standard error of the mean (as reported by the 1595A) is observed to be

0.00004 °C.
Note: The user must monitor measurement noise and use the actual measured measurement noise in the uncer-

tainty calculations.

2.2.5.3.12 Uncertainty of the Calibration Report RTPW

In this example, the RTPW is not measured and entered into the 1595A. The RTPW from the SPRT calibration

report is entered into the 1595A. The standard (k = 1) uncertainty of the RTPW value listed on the calibration

report must be included. In this example, it is 0.0001 °C. As explained in the previous example, all uncertain-

ties related to RTPW must be multiplied by W

T90

of the measured temperature. Multiplying by 1.612 yields

0.000161 °C.

2.2.5.3.13 Drift of the RTPW of the SPRT

Since an SPRT tends to drift, the long-term drift should be included as a source of uncertainty. In this example

the SPRT is allowed to drift 0.002 °C. The assumed distribution of this uncertainty is rectangular. To convert

to a standard uncertainty divide by 1.732 (square root of 3). The result, 0.001155 °C, multiplied by 1.612

yields a standard uncertainty of 0.001861 °C.

2.2.5.3.14 Combining the Uncertainties

At this point, all of the uncertainties can be combined by root-sum-square (RSS) since they are uncorrelated.

The RSS sum produces a combined standard uncertainty of 0.002133 °C. Multiplying by the coverage factor

(k = 2), and rounding, results in a total expanded uncertainty of 0.0043 °C.

2.2.5.4 Example 2: Calibrating an SPRT

As explained in “How the Super-Thermometer Measures” at the beginning of this section, the calibration of an

SPRT is performed by measuring the resistance at some required fixed-point temperature and then at the triple-

point of water. The two measurements are combined by division to get a W

T90

value. The uncertainty of W

T90

is

based primarily on the ratio accuracy of the Super-Thermometer.
In this example, an SPRT is calibrated at 419.527 °C (FP of Zinc). The RTPW is measured directly afterward.

The uncertainties resulting from the 1595A in this example are:

Resistance ratio accuracy of the 1595A at 419.527 °C

Measurement noise at 419.527 °C

Resistance ratio accuracy of the 1595A at 0.01°C (triple-point of water)

Measurement noise at 0.01°C

Reference resistor drift

The uncertainties in this example are calculated and combined as described in Example 1. However, there is a

slight difference in the reference resistor drift component.

2.2.5.4.1 Reference Resistor Drift

In Example 1, the 24-hour stability specification of the internal 25 Ω is used. This may not be necessary when

calibrating an SPRT. When calibrating an SPRT typically both measurements of W

T90

are taken in close prox-

imity of time (< 8 hours elapsed time). It is possible for the reference resistor drift to be negligible, especially

if the Super-Thermometer is in a stable temperature environment.
To be sure reference resistor drift is correctly estimated, the user should perform a test to determine actual

reference resistor drift over the elapsed time. One way to perform this test is to measure a very stable exter-

nal reference resistor over the actual time period using the internal reference resistor. If it is not possible to

measure the reference resistor drift, it may be necessary to use the 24-hour stability specification resulting in a

slightly larger total uncertainty. Another alternative is to use an external reference resistor of very low drift.

Advertising
This manual is related to the following products: