Ghost-Busting in the Dark - Better Mimir Data
Ghost-Busting in the Dark - Better Mimir Data |
||
|
|
12/15/2004 - Flagstaff, ArizonaThroughout the lab testing phase and the first two engineering runs of Mimir on the Perkins telescope, the images returned by Mimir have been plagued by "ghosts." These echo images show up as weak to strong (5-30%) repeats of bright parts of the image shifted by 8 pixels. An example of such a ghost is shown in the image of a slit, below. There the bright end of the slit is accompanied by a weaker (~20%) ghost slit offset from the first by 8 columns.
Last night, a careful series of tests revealed the cause of the ghosting and indicated how to eradicate it. One clue is the 8 pixel shift. Each of the four quadrants of the Mimir Aladdin III InSb array detector has 8 signal readout lines, so the data are read out in parallel, 8 pixels in a row in each quadrant (32 total) read out together. Then, the very next pixel any particular readout line sees is 8 pixels farther along in a given row. So, the ghost offset of 8 columns or pixels means that some of the signal from one pixel is bled into the very next pixel being sampled by that readout chain. The second clue was found when we began varying the time delay between when a pixel's information is requested by the analog-to-digital converter (ADC) electronics and when the ADC conversion begins. Although the Raytheon manual for the Aladdin III does state that this delay time should be between 1.5 and 5 microseconds, we needed to "rediscover" this fact. Once we increased this delay time to about 3-4 microseconds, the ghosting went away, as the images below show.
There is a time penalty for slowing the pixel reading, and we will implement several speed/ghost suppression options in the Mimir array control software over the next few days.
|
