An engineer or scientist who will be using a wavefront sensor in their work will naturally want to know how good it is. Is it accurate, or repeatably inaccurate? Does it meet specifications? If not, is the machine bad, or were the specifications unrealistic? What is the best operating region? What happens when the boundaries are exceeded? What are the instrument's idiosyncracies (like dust blocking holes in SHAPE)? Wavefront sensors are complex and sophisticated instruments. The information they generate is only as good as one's knowledge of the instrument's strengths and weaknesses. This chapter concentrates on the critical aspects of wavefront sensor evaluation.
When confronted with the task of evaluating the performance of a wavefront sensor, what tests make sense? The wavefront sensor can be essentially treated as a "black box." The evaluation process involves feeding the instrument a series of known phasefronts of increasing complexity. For example, one could start with a flat wavefront with different amounts of X & Y tilt, and then proceed to a curved but rotationally symmetric wavefront such as focus (Sec. 3.3), with both convergent and divergent modes. The next step would be generating non-rotationally symmetric wavefronts such as astigmatism. For the latter case, different clocking angles of the aberration should be incorporated. The machines estimates of phasefront shapes are then compared with the known inputs. This helps pin down accuracy. It is also desirable to perform the measurements several times to determine how repeatable the wavefront estimates are.
The conditions under which one performs the evaluation tests are important considerations. Ideally, this should be a controlled laboratory environment. Realistically, this may not always be an option. Cost and schedule pressures may force evaluation tests in situ at the operational site. This can be awkward, time consuming, and frustrating. In a lab one can exercise more control, and hence obtain more reliable and consistent results. At the operational site control is less certain because you are competing with other people for common resources. (Uninterrupted time is perhaps the most precious one.) As a consequence, the quality of ones measurements can be compromised.
Online access to SPIE eBooks is limited to subscribing institutions.