Page 155 - Mechanical Engineers' Handbook (Volume 2)

144 Temperature and Flow Transducers

elements. These tubes must be mounted farther apart than the size of the ice particles to
assure adequate cooling. One thermoelement and one copper lead wire are put into each
tube.
The principal requirement of the reference bath is that its temperature be known accu-
rately. Any region of known temperature can serve as a reference bath.
Many instruments that provide their output in temperature units contain local reference
regions and compensating circuits that augment the thermoelectric signal to account for the
local reference temperature. Such instruments can be used only with the type of thermocouple
for which they were intended, since the compensating network is specific to the calibration
of the thermocouple being used.

2.8 Obtaining High Accuracy with Thermocouples
The temperature emf tolerances quoted for thermocouples account for two types of deviations
from the expectation values: (1) batch-to-batch differences in average calibration and (2)
point-to-point differences in local calibration along an individual thermocouple.
Calibration of individual thermocouples can account for the batch-to-batch differences
but not the point-to-point variations along the wire.
For highest precision, three precautions should be taken:

1. Calibrate the individual thermocouples.
2. Minimize the working temperature difference (i.e., use a reference temperature near
the working temperature and physically close by).
3. Install the thermocouple so the working temperature difference is stretched over as
long a length of wire as possible.

2.9 Service-Induced Inhomogeneity Errors
When thermocouples are used in unfavorable environments or for very long times, the output
voltage may drift with time.
There are many possible causes for this drift, among which are selective oxidation,
which changes the composition of the alloy; diffusion of one or more of the components
from the thermocouple alloys to the sheath of a mineral-insulated, metal-sheathed assembly;
and local annealing of previous cold work. Most of these are ‘‘high-temperature’’ effects,
occurring mainly between 500 and 1500 C, as described by Campari and Garribba and
7
8
Bentley. Many of these effects are attributable to the complex composition of the thermo-
9
electric alloys, according to Schuh and Frost, who pointed out that the alloys were developed
in the early and mid-1900s to generate an emf that was linearly proportional to temperature.
This constraint, imposed by the widespread use of simple analog instruments, led to the
‘‘tailoring’’ of the emf characteristic by adding trace amounts of several constituents. With
a complex composition, even small changes could significantly change the emf at a given
temperature. Schuh and Frost pointed out that linearity is no longer an important issue since
digital processing does not require linearity. They recommended increased attention to the
use of high-temperature structural alloys as thermoelements, relying on ‘‘smart’’ instruments.
In the low-temperature domain, such as electronics cooling, one of the sources of in-
homogeneity error is cold working of the thermocouple wire by bending. Type K is signif-
icantly vulnerable to this effect. The calibration of a type K pair can be lowered by 1%
simply by bending the wires by hand. The error caused by this cold work depends on the
severity of the cold work, the length of the damaged region of wire, and the temperature