K-LINE Telescope History File Topic: Shutter Timing The K-line telescope has 2 shutters. One is attached to the Photometrics camera and is controlled by the Photometrics controller. The other, an Ilex shutter, is located at a pupil image formed by the reimaging lens. The goal is to use the Ilex to control the actual exposure, and use the Photometrics as a dust cover. Both shutters require some time to mechanically open and close after the appropriate command is given. There have been a number of changes in trying to synchronize the shutters to meet the goal. The synchronization is done in the subroutine DOEXP in the file ROOTSUBS. Initial operation - 10/18/88 --------------------------------------- The basic sequence of commands is: expose Photometrics; Stops continuous clear Fires the Photometrics shutter Returns shutt Ilex: Send state (open or closed) Wait for desired time Send close Returns readout Photometrics: Reads out pre-specified image 10/19/88 - 5/91 ---------------------------------------------------- From the initial installation until 10/18/88, this sequence was followed. On 10/19/88, it was found that the Photometrics was not opening fully before the Ilex. The following table gives the measured image intensity vs commanded exposure time. Commanded Intensity Intensity Exposure Before After (ticks) (ADU) (ADU) -------------------------------------------------- 0 81 - 1 81 210 2 81 555 3 200 880 4 500 - 5 670 - 6 1180 1740 9 2130 - 12 2980 3420 16 3880 - ---------------------------------------------------- These show that no light was detected until the commanded exposure time exceeded 3 ticks. The DOEXP code was modified to the sequence: expose Fire the Photometrics shutt(closed, 5) Wait 5 ticks for the Photometrics to open shutt Take exposure readout which resulted in the improved values in the last column of the table. 5/91 - present ---------------------------------------------------- In May 1991 the Photometrics shutter failed. It was locked open and the cable to the controller disconnected. As of that date, the wait was unnecessary. Ghost image 11/88 - 6/92 ------------------------------------------ In 1992 the problem of ghost images was identified. The specified image that the Photometrics reads out is a 384 x 400 pixel subarray of the 384 x 576 CCD. The starting row is actually row 89 of the CCD. The Photometrics controller rapidly shifts the subarray by 88 columns, and then slowly begins to readout the pixels. The Ilex shutter does not fully close before the readout began. Therefore, a weak image, displaced by 88 rows, is seen. With the K-line filter set to 0.3 A passband, the image exposure is 100s of ticks, much longer than the closing time. But at 1.2 A passband, the exposure times can be < 10 ticks, and the ghost image is noticeable. At noon, with the 1.2 A passband, and the reduced target exposure level for the flare observing program (KAB01), typical exposures were 3-4 ticks, or 50-66 msec. The Melles Griot catalog data curve for the Ilex shutter shows a linear ramp in time for the shutter closing, with full closing occurring 3 msec after the close command is received. The expected spurious open time is thus several percent of the image exposure, with no allowance for the time needed to issue commands. The time required for the parallel shift is known from Matt Penn's measurements of the MCCD Photometrics camera, which is identical to the K-line camera. The time for a single row parallel shift without readout is 15 microsec; to shift 88 rows takes 1.32 msec. The readout and digitization of each pixel then takes 3.8 microsec per pixel plus 0.51 msec per row, or 1.97 msec per row. The actual time needed to initiate the command is not known, but should be microseconds. Actual measurements of the intensity of the ghost image were made to determine the effective spurious exposure time, T(spurious) = T(image) * I(ghost) / I(image). The inferred values vary from about 10 msec in the morning to 3 msec in the afternoon. This suggests that the shutter operated more nearly at specification when it was warm, and was slower to close when cold. Tests show that the optimal shift of the ghost from the image is 89 rows; this includes all the times folded together. The DOEXP code was modified on 7/7/92 to the following form: expose Stop the CCD clear shutt Take exposure (no wait first) shutt(closed,1) Wait 1 tick for the Ilex to close readout This solved the ghost image problem and eliminated the unnecessary delay (the Photometrics shutter being absent). To remove the ghost image, its precise amplitude must be measured on each image. This is difficult in the presence of scattered light. The best solution is: a) Average about 30 images together to increase the signal to noise ratio of the ghost image. b) Use least-squares to determine the amplitude of the ghost relative to the average image, with the 89 row shift. c) Shift, scale, and subtract the image to remove the ghost. d) Use the cleaned result to describe the shape of the scattered light. e) For each individual image, subtract the scattered light, then find the ratio of ghost to image. Shift, scale, and subtract the image to remove the ghost. -----------------------------------------------------------------------