NASA is currently estimating the that the upcoming wavefront process to align the JWST's mirror segments will take around three months.

This description of the WFSC process paints a picture of numerous stages that involve variants of processes of a similar nature:

  • Acquire an image using the current hardware positions
  • Perform image processing incorporating new and previous images to understand what positional changes to make next
  • Actuate changes to one or more hardware positions
  • Repeat until alignment is satisfactory

Of these three processes (image acquisition, image processing and actuation) each can sometimes be very fast and sometimes very slow depending on the engineering application.

If they were all very fast but consist of a lot of iterations, perhaps communication round trips between JWST and mission control would be important too (there is some suggestion that image processing will rely on high-performance computing clusters that would presumably be ground-based too).

If there are many significant human decisions to be made in between steps of the process then significant portions of the time may elapse when none of the above are taking place.

So what be occupying the most of the JWST's time during these months?

  • $\begingroup$ It may not take that long, space agencies have long learned to work in some fudge factor into their time estimates so that they set expectations. If they say 6 weeks and it takes one minute more then it's national news and a torrent of criticism, if they say 3 months and it takes 6 weeks and one minute they get congratulated. $\endgroup$
    – GdD
    Jan 24, 2022 at 14:08
  • $\begingroup$ @GdD however long it takes I think it's interesting to know what most of that time consists of $\endgroup$
    – Will
    Jan 24, 2022 at 14:20
  • $\begingroup$ So do I @Will, it's a good question and I'm looking forward to seeing the answers. $\endgroup$
    – GdD
    Jan 24, 2022 at 14:23
  • $\begingroup$ One factor is for sure the number of variables. They have 18 mirrors with 6 actuators each. And every mirror reflects an image. There are not shutters, so you get an image that overlays all the images from each mirror. So first, you have to dins out which mirror you have to move (or bend) to correct which error. And yes, 10minutes rtt doesn't help. $\endgroup$
    – TrySCE2AUX
    Jan 24, 2022 at 14:58
  • 1
    $\begingroup$ IMO the "humans in the loop" is what will take up most of the time. JWST is a $10 billion asset. The big names making the big bucks will be damn sure that the big brains (making the not as big bucks) are making the right moves. That means time to think, re-think, test, re-test, press go, and still cross your fingers & pray a little. $\endgroup$ Jan 24, 2022 at 19:09

1 Answer 1


You are missing several pieces of the puzzle. One piece is NASA's operational tendency to underpromise / overdeliver. (This apparently does not apply to the process of building spacecraft.) If it only takes 2.5 months nobody will complain that the ops concept was overly pessimistic. The estimate was designed to be reasonably pessimistic. Some minor glitches will inevitably be encountered along the way.

For example, they might initially see only 16 images of the guide star. (There should be 18, one per segment.) This isn't a flaw; it's a minor glitch. It will take some time to find out which of the segments needs major realignment, and then some more time to perform the realignment. If they do happen to see 18 images on the first try they will have saved multiple days in the alignment process. However, assuming that that will be the case is an example of overpromising.

Another key factor is the JWST does not operate in realtime. It instead records imagery over the course of a day and then downlinks the recorded imagery and telemetry data during daily four hour contacts with the DSN. This contact period is also when commands for the upcoming day are uplinked to the spacecraft. This slows the process down.

Yet another factor is this a \$10 billion dollar (US) spacecraft. One does not send commands willy-nilly to a \$10 billion dollar (US) spacecraft. It's a bit like cooking brisket: Slow and low. The commands to the segment actuators need to be smallish. If large changes are needed, that will take multiple days to accomplish. In addition, some separate team most likely will be verifying and validating the command sequence generated by the segment alignment / wavefront sensing and control team.

  • 1
    $\begingroup$ I didn't really intend to imply that the long time estimate was at all surprising or requiring of explanation; just to get a breakdown of what most of that time will be spent doing. You've partly answered this with your description of the daily contact windows. Do you have any sense, between these periods, whether a day's worth of instructions typically consists of lots of predetermined images and movements or one very slow one? A few minutes of image acquisition followed by hours of slow mirror movement? $\endgroup$
    – Will
    Jan 24, 2022 at 15:17

Your Answer

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

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