| Date of change | System | What changed? |
| June 4 2008 | Software and Headers |
The four on-chip guiding windows that are saved used to have problems: many
frames were either skipped and lost or duplicated. Doug Teeple tracked the
problems and made sure that neither case occur anymore. A new header keyword
has been created to monitor if frames are skipped: WCSKIP. Please note however
that guide cubes are generated to files, and no program writing to the
filesystem can "absolutely" guarantee that slowdowns might not occur and cause
skipped frames.
|
| June 3 2008 | Software |
A new command, gset, has been added to enable deliberately choosing specific
guide stars so that guide cubes can be used for science projects. Thanks to Tom
V.
|
| Mid March to April 7 2008 | Hardware |
The edge crosstalk has been removed in hardware and the noise level has been
slightly reduced. More very soon on this.
|
| February 29 2008 | Image Headers |
An error in the values of WCGDDEC1,2,3,4 was fixed. Previously, the DEC values
were divided by 15 before being converted to sexadecimal values.
|
| October 26 2007 | Image Headers |
The SLDATE?? keywords have been added. They represent the exact time when each
image slice was obtained, starting with 01 and up to 28. Per Doug Teeple's
email: "The DATE and HSTTIME keywords are now accurate to within a few
milliseconds of the start of exposure of the first corner of a cube. New
keywords SLDATExx (where xx is the slice number starting from 01) have been
added to identify the start of each slice exposure time to within a few
milliseconds. When the cube is built from the slices, the SLDATExx keywords
appear in the primary header and all extension headers consecutively after the
DATE keyword." They correspond to the time at the start of the reset.
|
| October 2 2007 | Image Headers |
Per Doug Teeple's email: "The DATE and HSTTIME headers now reflect the time at
the beginning of exposure (start of the reset) of the first slice (fixing a bug
where the time was the end of exposure)"
|
| July 3 2007 | Hardware |
One video board (the lower one y=1-512, on chip #54 - ext 3) has been swapped
back to the one that was before. Noise level was higher and it experienced
transient disfunctionning. This video board is special, it has a load resistor
of 15k instead of 10k for all others.
|
| June 29 2007 | Hardware |
Cabling to the video board has been changed, i.e. the reference voltage bias
(VREF) is no longer shared with the VSOURCE bias on the first video board of
each array. This should totally remove the negative crosstalk that has plaggued
WIRCam since day 1. The DSP code was adjusted (voltage changes) to minimize
first horizontal line level offset of each amplifier. The system noise has been
reduced by connecting all video boards ground wires. This should improve the
noise level by a couple of ADUs. Five video boards have been replaced or
swapped in the process.
|
| June 19 2007 | Hardware | Video
board #3 of extension 4 (third from the bottom - pixels y=1025-1536) was
swapped for one having a load resistor of 10k-ohm like all other boards instead of
previous 5k-ohm. It was shown that this board suffered from more positive
crosstalk than all other boards. That should therefore reduce the crosstalk on
that board. Also, many loose capacitors were resoldered on all boards but this
should have no impact on images if not for lower noise levels. All the video
boards were removed and reinserted in the process as well as one timing board.
|
| June 6 2007 | Image headers | The
default World Coordinate System (WCS) solution has been changed and now uses
the values determined by Terapix. The 6 keywords to have changed are: CRPIX1,
CRPIX2, CD1_1, CD1_2, CD2_1, CD2_2 for all 4 extensions. |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
