A change in temperature of 1°C produces the same change in resistance produced by a strain numerically equal to a/G, where a is the temperature coefficient of resistance and G is the gage factor. If a = 1/300°C-1, approximate value for many pure metals, the apparent deformation is in the order of 1/600, which would correspond in steel to a stress of approximately 700 Nmm-2, the order of maximum allowable stress; the magnitude of the error can be reduced considerably by using an appropriate material for wire (for the constantan a и 0°C-1). A low temperature coefficient must take precedence over other desirable qualities for the material and is the main reason for the low popularity of semiconductor strain gages in wind tunnel balances.
In addition to the choice of a suitable material, however, other precautions are necessary to minimize errors due to temperature changes. The simplest method of temperature compensation is to provide a second not stretched strain gage near each strain gage and put them in two adjacent sides of a Wheatstone bridge: in this way the temperature changes do not alter the balance of the bridge. The changes in resistance due to temperature are also negated when the responses of two adjacent active strain gages must be subtracted to obtain a force or a moment.
The balances for intermittent wind tunnels and shock tubes can also be thermally insulated. Balances were also used with air conditioning in wind tunnels for continuous operation, particularly those in which large variations in stagnation temperature occur. These precautions should be regarded as additional to the use of a balanced bridge, using a material with a low temperature coefficient of resistance.
The strain gages are also sensitive to moisture in a fairly random way. One reason for not attempting to waterproof each strain gage is the fact that in this case it is more difficult to dry once they are wet, and if problems are expected, the entire balance can be sealed after placing a desiccant inside the enclosure.
The deformation balances have the advantage of a much faster response: in a null reading balance, even if automatic, the inertia of the system will not allow the measurement of rapidly varying forces. The only limit to the frequency response of a deformation balance is fixed by the natural frequencies of the sting and the model. In addition to the measurement of forces on oscillating models, strain gages can be used to measure the hinge moments of control surfaces: before the introduction of the strain gages hinge moments were measured by connecting the control surface to a balance outside the wind tunnel; the unsteady hinge moments could not be measured by any convenient method.