New additions and improvements (July 2004):

• Frame read time is now faster; 1416ms instead of 1817ms


• All 4 PHARO wheels (slit, lyot, filter, grism) now move one direction only. Position accuracy is vastly improved.


• All 4 PHARO wheels and the Carousel now incorporate HOME sensing logic. These wheels should not get lost as in the past.


• The xpharo Radial Display plot now reports strehl ratio. This ratio is calculated using the current status of the PHARO wheels; ie it is intended for use with real time images, not with images loaded at a later time into the display tool.


• Two new scripts are available to help reduce FOI reset errors:
  • sudo lockxpharo renices xpharo to -19 (raises xpharo to highest priority)
  • sudo killxpharo kills xpharo, esp above lockxpharo if -19 causes a hang

New additions and improvements (February 2003):

• 2 new macro commands, plus macro improvements:
  • set_tint sets integration time in milliseconds
  • set_ncycles sets number of integer cycles
    
    
  • take_bgd now writes data to the background buffer
  • tab characters are now permitted (ie, in commented lines)
  • TCS time is now updated properly during Pause statements, no more popup errors
  
  
• TCS FITS keywords are now updated properly in multiple cycle exposures
  
• Beeps now sound off if exposure time counter reaches -5
  
• Debug messages are now automatically time-stamped and written to pharo.debug in the users working data directory. If you experience any trouble with Pharo, please note the current time of the problem on the PINK trouble report sheet.
  
• We believe that the wheel and Carousel failures (items #3 and #4 in the following table) have been fixed. The exposure failures and the wheel / Carousel failures related to the FOI card (items #1 and #2 in the following table) have been sharply reduced. Please see Tom Hayward's detailed report for more information.

Bug	Rate	Description	Temporary Solution	Longterm Solution	Date Bug Fixed
Exposure Failure	~ 1 - 5 times a night, esp in CYCLEs	Exposure starts normally, exposure counter reaches zero but continues to "negative infinity" and beeps. This countdown will continue until observer intervenes.	Press the STOP button	Lock memory buffers, investigate spurious signals from FPGA	02-06-2003 *
Filter / Carousel Failure	~ 1 - 5 times a night	After requesting a motor move, the GUI selector changes to MOVING but the motor never moves. Selector freezes on MOVING until observer intervenes.	Reset the fiber optic interface; Menubar item FOI / Reset FOI	Lock memory buffers, investigate spurious signals from FPGA	02-06-2003 *
Filter Wheel Failure	FIXED	Slit wheel, Lyot wheel, Filter wheel, or Grism wheel HOME request ends up in wrong location. If the resulting occultation is ~ 30% or greater, images will appear oblong. Occurrence depends on use of wheels.	Re-home the offending wheel. Check the Pupil image against background to look for any occultation. Don't HOME a wheel unless necessary.	Fix noisy Lyot HOME sensor. Develop better mask algorithm around HOME locations.	02-06-2003
Carousel Failure	FIXED	Carousel moves only a fraction of requested amount, but GUI incorrectly updates to new position. Usually occurs after a HOME of the Lyot Wheel. You may or may not notice a difference in images, depending on the position where the error occurred. Occurrence depends on use of Carousel. If you notice the Carousel GUI selector updates to the requested position too quickly (no MOVING flag), then the Carousel has just failed.	Select the Pupil Carousel position. TWEAK the Carousel in NONE mode using 400 steps UP until the pupil limit is tripped.	Fix noisy Lyot HOME sensor. Develop algorithm to account for true Carousel position.	02-06-2003
CYCLE Image Timing	All images taken in a CYCLE	The signal in the 2nd to Nth images taken in a CYCLE is 2-3% brighter than the 1st image. The signal in the 1st image is similar to single images taken with the expose buttons. The readnoise of the 1st two to three images in a CYCLE is slightly higher than normal. The signal in the 2nd to the Nth images in a CYCLE appears stable, although the first two to three RESET frames exhibit odd behavior.	Consider writing a macro script that simply takes as many single images as you need. The added overhead to do this amounts to 1.84 seconds per image.	Investigate bug in FPGA electronics	not fixed
CYCLE TCS Header Information	FIXED	There is no TCS information in CYCLE frames #2 - N. Note that the relative timing of CYCLE images is good to within one millisecond	None (take good logs)	Unknown	02-06-2003
FSM / DM Header Information	Unknown	Image header keywords DM_ON and FSM_ON (true or false) are sometimes incorrect	None (take good logs)	Unknown
MACRO Take_Bgd Command	FIXED	The MACRO command take_bgd puts the resulting image in the Src buffer	None	Fix the bug in the macro.c code	02-06-2003
MACRO Pause Command	FIXED	The MACRO command pause value (say 5 seconds) is added to the system time, the result of which becomes the TIME_TCS header keyword: TIME_TCS = TIME_TCS[true] + PauseIf this Pause value is 5 seconds or greater, a popup window appears with each exposure warning that the difference between computer time and TCS time > 5 sec.	Move the annoying popup window out of the way and leave it there; the true tcs and unix time are OK	Fix the bug between macro.c and timing code of xpharo.c	02-06-2003
Pharo Crashes	Unknown	Pharo disappears with a core dump	Restart xpharo from the home directory (/export/home/pharo/xpharo &)	Unknown

^* These failures have been reduced, but still occur intermittently especially in multiple cycles.

---

If you would like to report a Pharo bug not listed here, please contact: Rick Burruss

---

Last updated Oct 24, 2004