This page is meant to use the magic of an animated GIF to demonstrate how images will be trailed during shutterless use of the PXL guide cameras. This is a "one-shot" animation, which means to watch the animation from beginning to end you may need to hit "Reload" on your browser each time.
See also examples of shutterless/eraseless guiding from NIRSPEC.
The beginning frame of the animation shows three areas. At middle is the real CCD, a 14 row, 15 column grid. The blue pixels near center represent a subarray, but the animation will follow all of the pixels through the process of reading out a frame. This will allow it to serve double duty as an example of full-frame readout and subarray readout. There aredifferences, however!
At the top is a gray grid of cells that represent the final image as saved to disk. At the bottom is another gray grid of cells representing "virtual" rows. These provide a mechanism to demonstrate what happens to the second frame when it is read out. The row numbers down the left side of the animation will represent row numbers in the final image (not physical rows on the CCD). This will be clear in the animation, when the rows are seen clocking up into the final image, and the row numbers follow them.
Three stars are also shown on the CCD: one above, one in, and one below the subarray region. These will show the effects of smearing stars in different positions on the detector. The smears will be shown by lighter shades of yellow. The pixels which are exposed during integration (the beginning of the animation) are a deeper yellow, and represent the desired image. (Unfortunately, different types of computers and different monitors will display colors differently, so apologies if your combination do not show some of the subtleties described below.)
Watch the animation once or twice, then read the discussion below it for further insights and information. The "rest position" of the animation is actually its last frame, which shows the image fully read out.
Before this first frame is read out the CCD was cleared. This is simply to give a familiar starting point to the discussion; there is no history of previous exposures on the first frame. This will not necessarily be true of subsequent frames, but this lets us discuss the differences between using an erase cycle and not using an erase cycle.
The first frame of the animation is meant to pause for a second to indicate the intended exposure. The stars are providing photons to the intended pixels in rows 3, 7, and 12.
After the integration time the rows begin to clock out of the CCD and into the frame which will eventually be saved. Note that in reality the first pixel would be read straight into the first memory position on the computer; it would not be "clocked" up through the array. The pixels are shown schematically clocking into the image because the symmetry of the pixel movement shown in that manner is less obtrusive. (Plus, the animation would have been much more tedious to create if it showed the shift register and pixels reading out one by one.)
Note that after a row is shifted and is reading out (with a read time much exaggerated here so that you can see what is happening), the stars fall on different pixels. These pixels will end up being at higher row numbers in the final image, hence the final image will show trailing downward from the stars.
As the CCD frame is read out the bottom rows of the CCD are replaced by "virtual pixels" from the endless supply below. This is a schematic way of showing that during a parallel shift, the last row of the detector will be replaced by legitimate pixels which, however, are empty. They will begin to accumulate photons from the sky (as well as dark current) as soon as they are created. In fact, these pixels will become the rows of the second frame!
Note that while the rows containing the subarray are reading out it takes a longer period of time per row than for the other rows. This is shown in the animation by a longer pause between frames of the animation at those stages. It also implies that some pixels will sit underneath a star for longer periods of time. This is represented by a deeper yellow in the tails.
For full-frame images, of course, all of the pixels are read out so there will be no asymmetry along the lengths of the tails.
Finally, note that even though the star above the subarray would not be recorded in a normal, shuttered image, its tail is indeed recorded in shutterless mode.
At the end of the animation the first frame has been read out. What happens to subsequent frames? The pixels which will make up this second frame have already been exposed to sky, and indeed to different parts of the image. There are already tails from the stars on those pixels.
If the second frame starts with an erase cycle, then the photons already recorded as the first frame reads out will be cleared, and you will then have the same situation as we discussed for the first frame. The stars will show asymmetric tails as they smear down the image.
If, however, no erase is performed, then those tails and the photons from the sky in each pixel will remain. Note that they lie above the stars in the second frame. The integration time will add more photons to the intended pixels of the image. Then the readout of the second frame will again smear tails down the image.
Hence, without an erase the second frame will show symmetric tails, both at higher and lower rows than the main star images. The tails above the stars were produced during the preceding readout, while the tails below are produced during readout of this second frame.
Not shown in this animation is the effect of sky. We will leave this as an exercise for the reader, but the first frame will have a gradient in the sky background. The second frame will of course look the same as the first frame if an erase is used, but will have a constant background if there is no erase cycle.
As a hint, note that in the first frame the initial erase starts all pixels at "zero." After the intended exposure time the first row is immediately clocked off the CCD, and no longer receives photons from the sky. All other rows, however, are still recording sky photons. The second row will then be clocked off the detector, and stop receiving photons, etc. The last row will receive the longest exposure time of all.
For the second frame, as the "virtual rows" are turned into real rows on the CCD, they immediately begin to record sky photons. They continue to record sky photons as the first frame is read out, then received the intended exposure time, then receive more exposure as the second frame is clocked out. The first row of the second frame receives the most exposure as the first frame is read, but receives the least exposure during readout of the second frame. At the other extreme, the last row of the second frame receives almost no exposure while the first frame is read (it has only recently been "created"), but receives the most exposure while the second frame is read out.
The result will be that all rows will receive the same sky exposure. In fact more in depth analysis shows that this will be true regardless of whether a subarray is being used, as long as the subarray in frame 1 is the same as in frame 2.