Документ взят из кэша поисковой машины. Адрес
оригинального документа
: http://www.apo.nmsu.edu/35m_operations/agile/Procedures.html
Дата изменения: Fri Mar 11 00:00:46 2011 Дата индексирования: Sat Apr 9 23:21:03 2016 Кодировка: Поисковые слова: m 101 |
To reduce the danger of corrupt images and failed exposure sequences I recommend the following procedure for every night that Agile is to be used. Do this least an hour before the camera is to be used (to give the camera time to cool):
runAgileICC start
hub startNubs agile
hub startNubs agileExpose
telnet nimble 1025
If runAgileICC start
fails, here are some things to check from your login on nimble:
Xvfb :1 -fp /usr/X11R6/lib/X11/fonts/misc &
The best way to shut down the ICC is to issue ICC command shutdown
.
If you prefer you can stop the ICC from a nimble login by issuing the command runAgileICC stop
.
To disconnect from the camera (e.g. to reset the camera or allow aacq to use it), issue ICC command disconnDev camera
. To connect to the camera again, issue ICC command connDev camera
.
Note: AgileICC turns on the camera's thermoelectric cooler when it connect to the camera. The thermoelectric cooler then stays on until the camera is reset. or power cycled. (In theory, software can turn off cooling, but as of 2009-02-02 none of our software does this.)
If Agile starts writing corrupt images then you must reset the camera (and you may end up having to reboot nimble as well). To do this:
disconnDev camera
.
resetAgile.py
and enter the password for Agile's power controller when requested.
powerCycleAgile.py
and then powerCycleNimble.py
(in that order).
connDev
.
Agile ICC runs on the computer "nimble". Sometimes it needs a reboot (the camera's usb driver seems to corrupt the kernel).
From the ccd account issue the command: reboot
If nimble hangs (as it sometimes does when the kernel is corrupted and you try to reboot it) then you must power cycle it. Log into agile@newton and run powerCycleNimble.py. You will need the password for crps01, the power controller nimble is plugged into. Or you can telnet to the power controller directly; nimble is on port 7.
Check out the source code from its svn repository: svn://svn.apo.nmsu.edu/agile/ICC (subdirectory tags/tagname or trunk).
Read the ReadMe.txt file in the root level of the source code.
We have had serious problems with images being corrupted, typically by having a band of 0s along the top. The problem appears to be a bug in the camera control software from Princeton Instruments (one of many such bugs); see PVCam Gotchas for more information. We have tried to get Princeton Instruments to fix the problem but so far they have ignored all such requests. If you wish to have a go at it feel free to contact Princeton Instruments 609-587-9797. Rob Allen RAllen@piacton.com knows our system best but is not reliable about returning phone calls or responding to emails. The problem is described in detail in the Serious Bugs section of PVCam Gotchas.
Workaround: the ICC automatically disconnects from the camera and reconnects after every exposure sequence. This takes roughly 20 seconds and the ICC will be unresponsive for that time. This seems to dramatically reduce the frequency of the problem, but we're not yet sure if it is a complete cure.
Also, if an image is taken whose upper left pixel has value 0 the ICC will output keyword ImageIsBad=path
and the FITs file will contain the keyword BADIMAGE. This is a crude test that could fail for several reasons: an image might actually end with 0 (false positive) or it might be corrupted in some other way (false negative).
We have seen two forms of corruption that this test would miss:
Note: the ImageIsBad keyword is a temporary hack while we wait for a fix from Princeton Instruments. Please do not rely on it. It is intentionally not documented in the standard set of keywords below.
There are two things to be aware of regarding overscan:
In addition to Y overscan corrupting data, we're not sure what the pixels in the Y overscan region actually mean. The Y overscan regions we've seen do not look like bias regions. Perhaps this is because we've not read out enough rows to get to the bias region (like X overscan, that requires 18 unbinned columns to get to the bias region). Perhaps the fact that it is a frame transfer camera affects the Y overscan region.
Agile ICC occasionally aborts an exposure sequence with a complaint about timing (sometimes when you first start the sequence, sometimes well into it). Our best guess at this point is that the kernel is getting corrupted and confusing the camera or the timing card.
The best bet is to reboot nimble.
To send an ICC Command to Agile ICC you need a connection, which you can get by logging into newton or nimble and typing:
telnet nimble 1025
If this fails then the ICC is probably not running; read Startup and Shutdown for instructions on starting the ICC.
If AgileICC is running and connected to the hub, then a simple way to send an ICC command is to type:
agile desired command
for example:
agile disconnDev camera
into the command bar along the bottom of TUI's Log window.