Shutter ballistics and exposure time

After two years of operation (Feb. 2005) with nearly 50,000 activations, it became important to check the ballistics of the shutter built by CEA, clearly the piece of equipment in MegaCam which gets the most mechanical stress during its operation. CEA delivered the shutter with the following specifications: an accuracy of 0.003 second, and an uniformity of the exposure level over the mosaic of 99% for a 1 sec exposure. The following study based on data gathered during the 05AQ01 run (Feb. 2005) confirms these numbers are still valid.

Shutter mechanism description

The shutter is a half disk which motion exposes and then covers the CCD mosaic at a constant speed for both opening and closing, ensuring that all points of the mosaic get exposed for the same time. Figure 1 shows a view of the shutter from above and the relative configuration of the MegaCam CCD mosaic:

Figure 1. Shutter and mosaic geometric configuration.

The illumination "edge", that is how the shutter unveils the CCD mosaic to the light is illustrated in Figure 2. Since the axis of rotation of the shutter is at a distance from the mosaic, the illumination edge describes a long front during most of its cruise:

Figure 2. Illumination pattern across the CCD mosaic.

Illumination uniformity and exposure time

To investigate the uniformity of the illumination, a set of flat-fields of increasing exposure times is used: 1, 2, 3 and 5 seconds. The ratio of the 3 and 5 seconds frames shows no structure at all. However the ratio of the 2 and 3 seconds frames shows a slight gradient of similar shape as the illumination pattern described on figure 2: the largest difference is of course between the top right corner and the bottom right corner and amounts to 0.3 percent for the 2-3 seconds set. This structure is shown on figure 3:

Figure 3. Non uniform illumination (<0.3%) [mosaic as shown on Fig. 1].

For the 1 versus 3 seconds frames (the 3 second exposure time is considered as the threshold where the exposure is 100% uniform), that difference amounts to 1%, right within the specifications of the builder. The bottom corner CCD (CCD07) is getting more exposed than the top right corner (CCD35) for the 1 second exposure at a level of 1%, but the vast majority of the mosaic is more uniformly illuminated, specially the center where most photometric standards are observed.

Repeatability of the shutter timing and exposure time

To check the exact timing of the shutter, a reliable stable light source ought to be used: a dense star field (the MegaPrime winter PSF field) observed under stable conditions (seeing, transparency, airmass) is the most adequate candidate. Two sequences of 1,2,4,8 and 16 seconds exposures were obtained under such condition during the 05AQ01 run:

 Set1:
 -----
   781021o|Feb 14 20:44:28 2005| 1.01|OBJECT|Field PSF        |   1.1|r.MP9601|
   781022o|Feb 14 20:45:22 2005| 1.01|OBJECT|Field PSF        |   2.1|r.MP9601|
   781023o|Feb 14 20:46:22 2005| 1.01|OBJECT|Field PSF        |   4.1|r.MP9601|
   781024o|Feb 14 20:47:24 2005| 1.01|OBJECT|Field PSF        |   8.0|r.MP9601|
   781025o|Feb 14 20:48:31 2005| 1.01|OBJECT|Field PSF        |  16.0|r.MP9601|
 
 Set2:
 -----
   781026o|Feb 14 20:49:26 2005| 1.01|OBJECT|Field PSF        |   1.1|r.MP9601|
   781027o|Feb 14 20:50:21 2005| 1.01|OBJECT|Field PSF        |   2.1|r.MP9601|
   781028o|Feb 14 20:51:16 2005| 1.01|OBJECT|Field PSF        |   4.0|r.MP9601|
   781029o|Feb 14 20:52:16 2005| 1.01|OBJECT|Field PSF        |   8.0|r.MP9601|
   781030o|Feb 14 20:53:26 2005| 1.01|OBJECT|Field PSF        |  16.1|r.MP9601|

There are sensors positioned in the shutter structure measuring exactly the time the shutter was open for: this is loaded in the EXPTIME keyword of all exposures (REQTIME only reflects what was requested by the acquisition system, the actual exposure time always turns out to be a bit longer, hence REQTIME should NEVER be used for any serious photometry work on MegaPrime data).

The exposure times recorded for this set are:

 Set1:
 -----
 EXPTIME =                1.087 / Measured integration time (seconds)
 EXPTIME =                2.067 / Measured integration time (seconds)
 EXPTIME =                4.062 / Measured integration time (seconds)
 EXPTIME =                8.047 / Measured integration time (seconds)
 EXPTIME =               16.047 / Measured integration time (seconds)
 
 Set2:
 -----
 EXPTIME =                1.053 / Measured integration time (seconds)
 EXPTIME =                2.055 / Measured integration time (seconds)
 EXPTIME =                4.049 / Measured integration time (seconds)
 EXPTIME =                8.034 / Measured integration time (seconds)
 EXPTIME =               16.054 / Measured integration time (seconds)

Stars were extracted from the images and the difference in flux was measured, for an identical level of exposure, the ratio ought to be equal to 1.000. Since the exposure times, as reported by EXPTIME, differs slightly, the flux ratio between exposure has to be compared to the ratio of the measured exposure times. The following table gives the flux ratio between the exposures of similar length for the two sets, the ratio of the measured exposure time, and the ratio between these two ratios:

                 Flux measurements       EXPTIME measurements    Ratio
         1-1:    0.9722                  0.9687                  1.003
         2-2:    0.9934                  0.9941                  0.999
         4-4:    1.0014                  0.9968                  1.003
         8-8:    1.0001                  0.9984                  1.001
         16-16:  1.0004                  1.0004                  1.000

The precision of the level of exposure and the reported timing of the exposure is extremely precise: to within 0.3%, again still right on the delivered specifications by the builder. When comparing the scaling of flux versus the scaling of different exposure times (here comparing the exposure times that differ by a factor of 2), one can see that indeed shorter exposures are a bit longer than they should (but then again EXPTIME keeps nicely track of that), and the ratio between the flux ratio and the exposure time ratio is within 1 percent, but there are most likely systematics caused by the photometry extraction due to comparing objects which flux has doubled between the two exposures. The error is most likely not due to a shutter timing as the previous table shows a perfect timing at the 0.3% level.

                         Set1-Flux Set1-EXPTIME  Set2-Flux Set2-EXPTIME
         1sec vs 2sec    1.9190    1.901         1.9589    1.951
         2sec vs 4sec    1.9590    1.965         1.9841    1.970
         4sec vs 8sec    1.9959    1.981         1.9922    1.984
         8sec vs 16sec   2.0025    1.994         2.0008    1.998