Isotech Gold, Platinum User Manual
GOLD/PLATINUM THERMOCOUPLE AND THE
DC MEASUREMENT REQUIREMENTS
by John P. Tavener
Isothermal Technology Limited
Pine Grove, Southport, Merseyside, PR9 9AG, England
E-mail info@isotech.co.uk Internet www.isotech.co.uk
NOVEMBER 2000
INTRODUCTION
Thermocouples are nasty complicated ill understood things that measure temperature differences badly
and should be avoided at all cost.
Calibration of thermocouples - if possible at all - is a topic fraught with measurement problems. This
author has avoided them as far as possible for the past 30 years and refuses to calibrate most types of
thermocouple in Isotech’s UKAS Laboratory.
A great friend of mine once said of a fellow scientist “what a pity that such a great man should have
devoted so much of his life to such an inferior thing.”
ITS-90 removed them in favour of the Standard Platinum Resistance Thermometer.
The limiting factor with all normal thermocouples is that one or both of the thermo-elements is an alloy
and an alloy can not be produced that is homogeneous. This means that, because the Emf of a
thermocouple is generated along the wire where there exists a thermal gradient, if the thermal gradient is
moved along the thermocouple the Emf will change.
This limits the best thermocouple accuracy to about ±0.3°C, 2 sigma.
If both thermo-elements were of pure metal, then this limitation would not exist, giving the possibility of
more accurate thermocouple measurement.
For good stability noble metals are preferable to base metals and in particular Gold/Platinum and
Platinum/Palladium have been investigated.
The first published tables for Au/Pt and Pt/Pd thermocouples dates to 1941 (Roesner & Wensel), however
it was not until the 1980's with better purity metals available that the Au/Pt thermocouple was
reconsidered.
In Canada the research was carried out by McLaren & Murdoch
he thermocouple that of expansion mis-match. By adding a fine spring of Platinum to the junction
t
[2]
who solved one of the problems with
between the Gold and Platinum the Gold could expand freely without straining the Platinum thermoelement.
McLaren & Murdoch considered all aspects of construction and performance publishing a lengthy 2 part
article in 1987 which is still the standard work and essential reading to all who are interested in
Gold/Platinum thermocouples.
To summarize and simplify their work they found that this thermocouple if made according to their
prescription was capable of accuracies of ±0.02°C from 0°C to 962°C.
It’s output increases from some 6V/°C at 0°C to about 25V°C at 962°C.
During the 1990's Burns et al of NIST have confirmed and extended their work. A reference list is
attached at the end of this article
[1]
.
During the 1990's, reluctantly at first, this author has been looking at the Gold/Platinum thermocouple to
see from a Laboratory and Industrial point of view its advantages and limitations.
To construct such a thermocouple is a sophisticated project in itself.
Following the prescription of McLaren & Murdoch, wires of 99.999% purity are required and a long
complex process of annealing taking many days.
However, making the thermocouple is only a small part of the project.
As the introduction mentions thermocouples measure temperature difference and so both the temperature
at the reference and measuring ends must be considered.
Before selecting equipment to turn the thermocouple into a measurement system it is necessary to
consider some electrical quantities.
At the Silver point the Au/Pt thermocouple generates 25V/°C and may be capable of accuracies
approaching ±0.01°C. To match this, the total measuring system should have, in terms of voltage an
uncertainty of ±0.25V or in terms of temperature ±0.01°C.
Considering the temperature constraints, and firstly focussing on the reference junction, a good quality
Stirred Ice Bath made using distilled water would have an uncertainty of ±0.005°C, or half the total
uncertainties. If a Water Triple Point Cell is utilized this reduces to ±0005°C and so a Water Triple Point
must be used.
Next, considering the upper calibration temperature, a pure (99.9999%) Silver Cell in a heat pipe
apparatus can realize a temperature of 961.78°C within 1 to 4mK. This is adequate.
The third component of the measuring system is the voltage measuring device. After research, the
Wavetek 1281 7½ digit voltmeter was chosen, however the best UKAS uncertainty that could be issued
with the meter was ±0.5V 2 sigma, which is equivalent to ±0.02°C.
Combining these uncertainties it is possible to calculate the overall uncertainties at each calibration Fixed
Point - 4 are used: Water Triple Point, Zinc, Aluminium and Silver.
PHYSICAL CONSTRAINTS
The length of the thermocouple produced was two metres from measuring to reference junction and 3
metres from reference junction to digital voltmeter. This means that the reference junction and digital
voltmeter must be mobile as the thermocouple measuring junction is calibrated. A mobile computer
trolley was found to be ideal.
PLATINUM/GOLD THERMOCOUPLE - CALIBRATION AT FIXED POINTS
METHOD:
The thermocouple is made according to the descriptions in the literature.
Both the reference end and the measuring end are placed in Water Triple Point Cells and the digital
voltmeter reading noted with connections made, and then reversed.
The thermocouple reference junction is then raised 1,2 and 3cm to check immersion characteristics.
Next the measuring junction is placed after pre-warming in a freezing Silver Point Cell and after 20
minutes the Emf is noted over 10 minutes, connections are reversed and the readings repeated.
The measuring junction is withdrawn 2 and 4cm to measure homogeneity then moved to the Al and Zn
Points in turn then the Ag process is repeated.
The complete calibration is repeated to check reproducibility.
RESULTS:
See Attachment.
Loading...