Page 160 - Mechanical Engineers' Handbook (Volume 2)

2 Thermocouples 149

Figure 15 Electrical network model of the galvanic emf effect on a wet thermocouple. (Reproduced
from Ref. 2, with permission.)

electric signal. The net effect, then, is that the thermocouple reads low. If the galvanic
connection is near the junction, as it would be if certain soldering fluxes were not completely
removed, the corrosive ‘‘necking down’’ of the iron wire increases the resistance to the
galvanic current more and more as the wire is eaten away. This in turn means that the
thermocouple readout circuit picks up a larger and larger fraction of the 250-mV galvanic
signal. In bench tests of this mechanism, 14 it has been shown possible to have an iron–
constantan thermocouple read negative (i.e., below the reference bath temperature).
Neither copper–constantan nor type K material shows significant galvanic effects.
The equivalent circuit is shown in Fig. 15. It is similar to the shunt problem, but the
active emf source in the shunt dominates it.
The output of the system will be the thermoelectric emf corresponding to T T
hot ref
plus or minus the IR drop in the circulating current loop.
Typical values for E and R are 250 mV and 1 M /cm of wetted length (24-
shunt shunt
gauge, duplex-fiberglass-insulated iron–constantan wires, wet with distilled water). The error
induced by this signal depends on the resistance in the thermocouple loop and the location
of the wet spot along the thermocouple. Wet iron–constantan thermocouples can produce
large error signals.
Although type K materials do not display appreciable galvanic emf, they are strain
sensitive (i.e., cold work causes a change in calibration), and during the act of straining,
they generate appreciable emf. Temperatures measured on vibrating equipment may appear
to fluctuate as a result of flexing of the thermocouple wires. Copper–constantan and iron–
constantan are less active than type K, with copper–constantan least active. Spurious signals
on the order of 50 C have been observed using type K materials on a vibrating apparatus
(e.g., whole engine tests).

2.12 Self-Validating Thermocouples
Most thermocouple users blame the thermocouple for any ‘‘bad’’ data when, often, the fault
lies elsewhere in the system—unrealistic expectations, poor installation of the sensor, and
so on. As a consequence, there has been considerable interest in methods of ‘‘validating’’ a
thermocouple’s output while it is still in service.
One approach was to install the thermocouple junction in a capsule containing a metal
or alloy whose melting temperature was known and was below the operating temperature of
the thermocouple at steady state. On each ‘‘fire-up,’’ as the indicated temperature rose, the
thermocouple trace would display a plateau at the melting point, thus confirming that the
output was correct. A recent example of this approach was reported by Sachenko et al., 15
who outfitted a Chromel–Alumel thermocouple with a capsule containing lead.