# Special Calibration

Ophir offers a "special calibration" service so the customer can be provided with higher calibration accuracy than the standard catalog accuracy.

**Thermal Sensors – Power**

A typical catalog calibration accuracy is given below where the power uncertainty (2 sigma) is listed as ±3% and the linearity variation is given separately as ±1% .

If we look at the actual uncertainty budget for this sensor, we see that putting in the frequency, pulse width and energy uncertainty together, we get an expected total uncertainty of 2.51% . If we take out the linearity uncertainty and count it separately as in the catalog spec, then the total calibration uncertainty will be 2.34% Customers requesting special calibration are in general interested in reducing the uncertainty from 3% to 2%. We can accomplish this by the following method:

- The largest component of the uncertainty is the DUT (device under test) repeatability uncertainty. This is the uncertainty of transfer of the calibration from the NIST traceable silver master to the DUT. The standard Ophir criterion for such a transfer is that in the final calibration check, the DUT must read within 2% of the silver master or better. This will result in an uncertainty of 1.74% (2 sigma). If we tighten the requirement that the two sensors must agree within 1.5%, then the uncertainty goes down from 1.74% to 1.3%
- If we calibrate at the same wavelength that the customer will use, then we eliminate the contribution to uncertainty from variation in absorption with wavelength

So, putting the above into practice results in an uncertainty of 1.82% as seen below and we indeed can specify 2% calibration accuracy. This is without taking into account errors of linearity. If it is feasible, the sensor can be calibrated at the same power level(s) that it will be used at, thus eliminating this uncertainty as well.

Thus the customer has the standard option of special more stringent calibration at his wavelength of use. Where feasible, he may also request calibration at a power level or levels of his choice. The cost of this service is of course commensurate with the amount of work that must be done.

**Pyroelectric sensors – energy**

A typical catalog calibration accuracy is given below where the energy uncertainty (2 sigma) is listed as ±3% , the linearity variation is given separately as ±2% and the additional error with frequency is given separately as ±1% or ±2%.

If we look at the actual uncertainty budget for this sensor, we see that putting in the frequency, pulse width and energy uncertainty together, we get an expected total uncertainty of 3.48% . If we take out the frequency, pulse width and energy linearity uncertainty and count it separately as in the catalog spec, then the total calibration uncertainty will be 2.91% Customers requesting special calibration are in general interested in reducing the uncertainty as much as possible. We can accomplish this by the following method:

- The largest component of the uncertainty is the DUT (device under test) repeatability uncertainty. This is the uncertainty of transfer of the calibration from the NIST traceable silver master to the DUT. The standard Ophir criterion for such a transfer is that in the final calibration check, the DUT must read within 3% of the silver master or better. This will result in an uncertainty of 2.38% (2 sigma). If we tighten the requirement that the two sensors must agree within 2%, then the uncertainty goes down from 2.38% to 1.81%
- If we calibrate at the same wavelength that the customer will use, then we eliminate the contribution to uncertainty from variation in absorption with wavelength

So, putting the above into practice results in an uncertainty of 2.29% as seen below and we indeed can then specify 2.5% calibration accuracy. This is without taking into account errors of linearity, pulse width and frequency. If it is feasible, the sensor can be calibrated at the same energy level that it will be used at, thus eliminating some of the additional uncertainties.