13
$\begingroup$

CNET's April 23, 2024 World's Largest Camera, the 3.2-Gigapixel LSST, Is Complete says after about 02:19

That shutter is actually so big it had to be broken down into two pieces just to fit inside the camera. Now it takes just 0.9 seconds for the shutter to either open or close and it's going to do that a thousand times every night.

Question: Why does Very C. Rubin's [LSST]'s ginormous camera have a shutter, why does it open and close a thousand times a night, and is this typical for large-format survey telescopes?

cued at 02:19

$\endgroup$
0

4 Answers 4

24
$\begingroup$

From Technology of the LSST focal plane:

LSST’s survey efficiency depends on the telescope’s ability to rapidly move from field to field on the sky during a night’s observing. The mount has been engineered to slew and settle within 5 s. During this time, the shutter is closed and the camera must read out the complete 3.2 Gpixel image acquired in the previous field.

The CCD's are full frame imagers: they have no storage pixels to hold the image while it is being read out. The array is sensitive to light during the readout process: even if the telescope didn't move, readout would cause streaks in the image, if light could reach the focal plane. Moving the telescope during the readout would cause even worse streaks. So, it is important to prevent light from reaching the focal plane during the readout.

$\endgroup$
0
16
$\begingroup$

The "sequence" of a CCD exposure might be to flush the CCD to get rid of all charge on the silicon, with the shutter closed; open the shutter; expose the CCD to astronomical light; close the shutter; readout the charge on the CCD.

The shutter is there because the active area of a CCD is active all the time, and because a typical readout time is much longer than the time it takes to open or close a shutter. i.e. Just relying on flushing the CCD and then reading out at the end of the exposure would result in an effectively non-uniform exposure across the (multiple) CCDs and multiple images of bright stars that would appear in every pixel in the direction of readout.

As for why there are so many shutter operations. The baseline Rubin plan is to have 825 visits to each 9.6 sq degree field over a 10 year survey period, covering 18,000 square degrees (Bianco et al. 2022). If we assume every night is useable and each visit consists of 1 exposure, then that leads to the shutter being used 423 times per night (plus calibration exposures). However, a remark on the Rubin Survey Strategy web pages suggests that 5 million exposures will be taken over the ten years, which equals 1369 per night, so it must be that many of the "visits" will consist of multiple exposures.

Is this typical of large survey telescopes? No, not really. Most previous survey telescopes are producing deep, panoramic views of the sky, using much longer exposures. That means the shutter might be operated only a few tens of times per night. The Rubin telescope is designed to quickly survey large fractions of the sky every night, using many more, relatively short ($\sim 30$s) exposures.

There are some survey telescopes (the TESS satellite is an example) that have frame-transfer CCDs. These can rapidly transfer (but not read out) the stored charge into an adjacent, unilluminated, frame store very rapidly. In these cases (for example, TESS) you do not need a shutter at all because the frame transfer process is very much shorter than the astronomical exposures. But that is not the case for Rubin/LSST.

$\endgroup$
8
  • 1
    $\begingroup$ You generally only have to flush between exposures if you've exposed the CCD to unwanted light. Just reading a CCD out clears it of charge. $\endgroup$
    – John Doty
    Commented Jun 2 at 13:57
  • $\begingroup$ I note also that the terminology here is a bit confused. The charge you collect and sense is from the channels, not the substrate. The substrate is held at a constant potential, often just connected to "ground". $\endgroup$
    – John Doty
    Commented Jun 2 at 14:08
  • $\begingroup$ @JohnDoty second point agreed. First point not. Astronomical CCDs do a flush (not a readout, because that takes time) before starting an exposure. $\endgroup$
    – ProfRob
    Commented Jun 2 at 15:14
  • $\begingroup$ If you've just read the CCD out, and haven't exposed it to unwanted light, you need no flush. TESS, for example, has no flush function: the first readout or two of a >100,000 readout observing cycle is garbage, but so what? $\endgroup$
    – John Doty
    Commented Jun 2 at 15:23
  • 3
    $\begingroup$ Why would you assume that it has just been read out? @JohnDoty Most astronomical CCDs you do not have to discard the first readout or two; because they are flushed. The TESS CCDs are frame-transfer devices and can readout to the frame store in 4ms and so do not need and do not have a shutter at all. $\endgroup$
    – ProfRob
    Commented Jun 2 at 15:31
2
$\begingroup$

why does it open and close a thousand times a night?

The other answers, while correct and fascinating, are completely irrelevant. The CCD readout time is important to the designers, but irrelevant to why the shutter opens and closes so much. Slew time is also irrelevant. The plumage don't enter into it.

The shutter opens and closes a 1,000 times a night because it takes a 1,000 pictures a night. It doesn't take just one single frame of the entire sky.

Assuming a dark night sky that lasts six hours, they could take 1,000 pictures even if it took 21 seconds for each frame to slew, stare at the sky, read out, plus 2 more secs for the shutter to open/close.

That's why the shutter opens and closes a thousand times a night.

$\endgroup$
13
  • 1
    $\begingroup$ Nope. Shutterless operation is quite common with astronomical CCD cameras. Many don't even have shutters. TESS has no shutter and takes a picture every two seconds. $\endgroup$
    – John Doty
    Commented Jun 5 at 12:56
  • $\begingroup$ @john_Doty: " Shutterless operation is quite common with astronomical CCD camera" ==[ The OP didn't ask about other CCD cameras; he asked about this one. And this one opens once per frame so it can take 1,000 frames per night. You'll be able to trace why I got banned back to this, and everyone will pretend it's fair. $\endgroup$ Commented Jun 5 at 14:51
  • $\begingroup$ You can take lots of images without needing a shutter. So why does this particular camera need a shutter at all? $\endgroup$
    – John Doty
    Commented Jun 5 at 15:01
  • 1
    $\begingroup$ About the "==[" message, all I can think of is that "simple, obvious and true" are different for each person. I know it doesn't feel that way (I'm generally convinced everything I think and say is "right" except for an occasional mistake) but the deeper truth is that each person's world is unique, and when we communicate with others, we make these little bridges between worlds and exchange bits of our perspectives with others. It's OK to believe "everyone else is wrong and only I'm right" but its unproductive to say it out loud so explicitly. $\endgroup$
    – uhoh
    Commented Jun 6 at 1:52
  • 1
    $\begingroup$ Instead, let the rest of your post speak for itself. Instead of "The other answers... are completely irrelevant." which while it may be true to you, is not true for everyone ("relevance" is really subjective, we each decide what we find relevant in our view) try instead something like "While the other answers address X, I prefer to answer this reductively. Why does it close n times a night? Because it takes n pictures a night! The other answers address the shutter's raison d'etre but this is the actual answer to the question, as asked". This doesn't berate or cast shade on other answers. $\endgroup$
    – uhoh
    Commented Jun 6 at 1:58
2
$\begingroup$

Supplemental, re-explaining answer post for those like me who are not so familiar with CCDs.

I find the video below very helpful. Of the five readout systems illustrated, the two relevant here are:

  • Top-Center: "CCD Frame Transfer" (with a 2nd CCD area for storage)
  • Top-Left: "CCD Full Frame" (direct readout)

In both cases, a physical shutter is included - a gray square appears over the imaging area whenever the exposure is not happening.

The shutter is closed for only a brief time for the Frame Transfer because internal transfer only can be pretty fast (answers + comments here suggest milliseconds to low tens of milliseconds)

For astronomical applications (ground or space-based) Frame Transfer is often used, but without the shutter. That means there will be a small amount of streaking as the light hitting the imaging area doesn't stop during the fast, but not infinitely fast transfer. For some applications, especially when the exposure times are orders of magnitude longer than the Frame Transfer time, a very weak streaking is acceptable, and can potentially be mostly subtracted. cf. William M. Owen Jr. et al 2020 PASP 132 014504 A Quick Method for Removing a CCD Electronic Shutter's Two-sided Smear

But when you decide to tile a large focal plane with a 2D array of CCD imagers without big gaps between imagers, there is no physical space for the storage array. And it appears that the Interline method (Bottom Left) is not so popular for astronomical applications.

In this case, with no room for local storage, we must move to the Full Frame method, where the readout is much slower. The video illustrates that the presence of the gray square "shutter" is much longer.

This then is the case for the camera described in the question.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .