It says here (and elsewhere) that during mirror alignment the James Webb Space Telescope will take $18$ “individual images” of a star, “one from each mirror segment”. I’m wondering how this is possible. Can they direct light from one mirror segment at a time onto the sensors, and if so, how? Or do they mean that the data from the sensors is analyzed into $18$ separate images, and if so, how?

  • 1
    $\begingroup$ Wild guess; the other 17 are moved deeply out of focus, and/or tilted so that there are no bright objects in a small field of view somewhere on the sensor. There can be diffuse light but it will not interfere with seeing how sharply focused a target object is by one mirror. $\endgroup$
    – uhoh
    Commented Jan 20, 2022 at 13:16
  • 3
    $\begingroup$ I have work to do, so I can't take the time to provide an answer the question, at least not right now. This paper on the JWST wavefront sensing and control might help. $\endgroup$ Commented Jan 20, 2022 at 13:52
  • 1
    $\begingroup$ My first thought was that first light from the JWST when looking at a guide star could have up to 18 distinct images of that star. An obvious brute force solution to make those 18 images converge is to tweak each segment individually and see what results. After tweaking all 18, tweak again so as to (one would hope) make all 18 images converge on the same point in the imager. Non-linearities mean this won't be perfect, but repeated applications should eventually converge. $\endgroup$ Commented Jan 20, 2022 at 13:58
  • 1
    $\begingroup$ This apparently is exactly what is done in the coarse correction phase. The fine correction phase addresses phasing and astigmatism, and that's a bit beyond my depth of knowledge. $\endgroup$ Commented Jan 20, 2022 at 13:59
  • $\begingroup$ A not-so-brute -force method is to take a picture with mirrors as they are. Leaving one mirror alone, move all 17 others, but each in a different direction/distance. A second picture will identify which image belongs to which mirror, and the distance/direction to adjust each so it aligns with the mirror which wasn't moved. $\endgroup$
    – Woody
    Commented Jan 21, 2022 at 1:44

1 Answer 1


Starting with this quote,

After launch and deployment, the primary mirror segments, secondary, and science instruments will be misaligned relative to each other by up to several millimeters. An iterative process using several types of wavefront sensing and control will bring these mirrors into alignment within tens of nanometers. The large dynamic range (millimeters to nanometers) means that several distinct stages and types of sensing are necessary. This commissioning process is necessarily iterative, due to finite sensing precision and also to mechanism uncertainties inherent to the coarse stage actuator design. As a result, Optical Telescope Element (OTE) commissioning will be iterative at both small scales (a given step may need to be performed several times to converge) and at much larger scales (mechanism uncertainties will likely require looping back to repeat entire sections of the commissioning plan).

My interpretation is that they will move one segment at a time and observe the change in image PSF as well as changes in the measurements made by the local wavefront measurement tools. Due to the magic of matrix math, after every mirror segment has been moved a few times, a maximum a posteriori estimate of the optimum position for all segments can be made.
I am reasonably sure this is the approach being taken, rather than hard-drive each segment to a FOV which excludes all other segments. [background: I'm a former employee of AOA-Xinetics and was involved in the design of many wavefront sensors, including the system provided for JWST alignment]


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.