Issue:  Shutter didn't fully close at end of session, although it was reported as Closed

Description:  Checking the ObsCam picture on the Observatory Status Webpage at 06:15 in the morning showed that the shutter was not fully shut, it was only 80% closed and daylight was streaming in !

This was despite the fact that ObsViewer, ObsManager, DeviceHub and AstroGuard all reported that the Shutter was 'Closed'. 

Closing time was reported as being 35-40s. This is shorter than the normal closing time (50s) but consistent with the observation that shutter was only partially closed.   Checking the latest ObsCam picture on AllSky/Weather Computer showed that the dome was still partially open and a visual check from the house confirmed that the shutter was indeed partially open.  

Although the weather was dry and the amount of shutter closure more or less protected the observatory equipment inside, the issue is deemed critical since the failure could have occurred with the shutter more fully open than it was and weather conditions could have unexpectedly deteriorated (with rain etc)

In order to make the observatory safe/secure the shutter was remotely commanded to open via DeviceHub at 06:22:54 which then, because of Critical Light Conditions, prompted AstroMain to automatically command an immediate shutter closure at 06:22:55. This closure was successfully executed with the dome finally fully closed at 06:23:06  (closing time 10s).

Analysis:  Examining the Session Report File / Session Events table shows the Session entered the 'Closing' State at 04:25:28 (which is an automated process that checks & closes the Observatory Shutter, parks the Dome, parks the Telescope, turns off the Camera etc)

The Pulsar Trace Log shows the Pulsar Dome receiving a CloseShutter request at 04:25:33, after which the shutter starts closing (shutter state 3) at 04:25:34,  becoming 'closed' (shutter state 1) at 04:26:09,  giving an apparent 'closing' time of 35s.  Subsequently the Dome then moving from Az 76.1 to Park Position (Az 90.0) at 04:26:18 to 04:26:19.  

The ObsCam picture from 05:28 shows the shutter as being not fully closed, dawn sky can clearly be seen through the lower 20% of the shutter slit.

The Shutter Status in ObsCam pictures from 04:26 to 04:27 (Dome Parking),  04:28 and 04:58 is less clear due to the very limited contrast between the dark interior of observatory roof and the dark sky outside. Pushing image processing to the maximum however shows that the shutter was definitely not fully closed at 04:27. This eliminates the possibility that the shutter was fully closed at 04:26  but then became partly open at some point before 05:28.

The Shutter fully closed ok in the following three sessions (S885-S887) as it had done ok in the numerous preceeding sessions (S643-S883)

The only similar previous example of the Shutter not fully closing (other than due to chain issue) occurred in session S642 in Dec 2018 (see Investigation - Observatory Dome Still Open after Session Finished, 2018-12-19 ).  This was not fully explained either though it was felt to most-likely due to a short circut associated with small nicks in insulating layer of the 2 wires to the lower limit switch (apparently dating to original dome manufacture,  now repaired)

Conditions at Time
Freezing, -2.5 degC

The temperature at start of session when the shutter was initially opened  was 2 degC (21:35),  and then fell to 0 degC (22:40)
By the time the shutter was closed at the end of session (04:25) the ambient temperature had fallen to -2.5 degC. 
Conditions were therefore below zero for some 5.5 hours before the incident occurred.

Critical Question
After closing for 35s why did the shutter system deem that the shutter had reached a Closed State without the shutter physically making contact with & depressing the lower limit switch. 

• Could the limit switch have somehow become stuck in the depressed state, by for example ice ?
   
  This seems unlikely.  The temperature were above zero when the shutter was originally opened and the spring on the limit switch was in working order when examined the following day. If the limit switch was somehow frozen in a depressed state, wouldn't this have led to the system thinking that it was closed immediately after starting to close and thereby leave the shutter in a nearly fully open position rather than a 80% closed position ?
   
• Could a transient electrical short have tricked the system into thinking the shutter was closed ?
   
  This might have caused a similar issue in S642 (2018-12-19) but the source of the potential short circuit in that case (small nicks in the 2 wires to the limit switch) was repaired.   If there was a 'short' wouldn't this have led to the system thinking that it was closed immediately after starting to close and thereby leave the shutter in a nearly fully open position (as in S642 incident). Why would the short only happen when the shutter got to the 80% closed position ? (a water drop falling from shutter would have fallen into the observatory interior rather than onto the limit switch / wires ?)
   
• Could a random stray signal or a moment of interference somehow have tricked the base unit into ‘hearing’ a Closed State from the Shutter Unit ?
   
  It is unclear if the Pulsar Shutter /Dome Unit could behave in such a manner.  After any stray signal or interference had ended wouldn't the shutter status have then showed as the error state.(state 4)
   
• Could the Shutter Unit have aborted the closure due to some motor power issue and incorrectly passed on a Closed State rather than an Error State ?
   
  It is unclear if the Pulsar Shutter Unit could behave in such a manner.
   
• Could some person have physically pressed the limit switch as the dome was closing ? 
   
  This seems unlikely. The observatory was unoccupied and running in automated mode. Athough it can't be ruled out, it is felt unlikely that there was an intruder in the garden at 04:25 in the morning (-2.5 degC) who happened to put his hand into the Dome slit and pressed the limit switch at the precise time that shutter was closing.
   
• Could a cat have jumped into the observatory at some point and then pressed its weight on the limit switch when it made a mad dash as it heard the shutter closing ?
  
  This seems unlikely give the height of the shutter entrance/slippery sides but it can not be eliminated. There was no fur or other evidence.

Conclusion
The failure of the shutter to fully close at end of session is currently unexplained.

Action:
Pass on the observations & analysis to Pulsar Observatories for their information and possible comment.

Update 2023-02-15

At least two further occurance of the same issue (Pulsar Dome reporting that shutter is closed when it is only 75-80% closed) occurred in early 2023.  However these were automatically dealt with by introduction of a separate Zigbee based open/closed sensor on the shutter and enhanced autonomous operation by AstroMain's Obs.Manager which upon sensing this condition makes an interventation to properly close the shutter.  It does this by sending an open shutter command to dome and then as soon as shutter is showing as opening to then send a second Close Shutter command, which so far has preven to fix the issue.  The operation also takes an ObsCam picture before the interventation to capture a visual picture of the shutter state.