Calibration Accuracy of Ophir’s High Power Sensors

The 5000W and 10K-W sensors work on the thermopile principle, where the radial heat flow in the absorber disk causes a temperature difference between the hot and cold junctions of the thermopile which, in turn, causes a voltage difference across the thermopile. Since the instrument is a thermopile voltage-generating device, it must be linear at low values of output. Therefore, if it is shown to be linear at powers which are a significant fraction of the maximum power, it will necessarily be linear at very low powers and if the calibration is correct at low powers, it will remain correct at high powers as well.

On the other hand, although the output may be linear at low powers, there may be a zero offset that, due to the relatively low output at low powers, will cause an error in calibration. For example, if calibration is performed at 200 W and the output of the sensor is 3.5μV/W (a typical value) and there is a zero offset of only 0.1μV, this will cause a calibration error of 7%. Ophir’s calibration method always measures the difference between the reading with power applied and without power applied, thus eliminating error due to zero offset. This measurement is taken several times to insure accuracy.

The above measurement method assures that the calibration inaccuracy due to measurement errors is less than 1%, comparable to the expected errors in our lower powered meters. This accuracy will hold at higher powers up to the maximum specified for the sensor if the sensor is linear within spec up to its maximum power.

In order to verify this, models L1500W, 5000W and 10K-W against an L1500W whose linearity was tested at PTB National Laboratory in Germany. The sensors were measured at lower powers directly against the master standard and at higher powers using a beam splitter to split off a percentage of the beam to the master standard and the main beam going to the unit under test. These measurements have shown that these Ophir sensors are well within the claimed limits of linearity as shown by the graphs below.

Besides the accuracy of the initial calibration, there arises the question of the change in calibration due to damage or discoloring of the absorbing coating. This problem has been tested by focusing a high-powered beam down onto the absorber until damage occurs. The damage threshold that appears in the sensor specification is defined as the damage to the surface that will cause a change in reading of more than 1%. Below the defined damage threshold, the surface may become visibly discolored but the reading does not change by more than 1%.

Thus if the power density does not exceed the stated damage threshold power density, the calibration should not change significantly. Note, however, that with a Gaussian beam, the average power density may only be half as much as the peak power density in the center of the beam.