Re: Things that customers do that drive me nuts on CN
Because AP has a strong commitment to their products and customers that they naturally are dismayed when a customers experience is not satisfactory. I feel bad for Rolando and wish their was something I could do beside writing this supportive email.
From: ap-gto@... on behalf of chris1011@... [ap-gto]
Sent: Monday, July 9, 2018 7:54 PM
Subject: [ap-gto] Things that customers do that drive me nuts on CN
So here is an entry on Cloudy Nights that a customer posted under the following thread:
"I'm not the OP but I can tell you that my PI measured eccentricity goes from < 0.4 on one side of meridian to > 0.6 on the other side, due to the DEC oscillations I'm seeing, and the stars are obviously not round."
However, this same person posted this in the PHD Google groups:
"12:40: Disabled Backlash compensation. Had no obvious effect - continued to see oscillations
12:46: Reduced MinMo from 0.20 to 0.10. The more aggressive MinMo resulted in more frequent oscillations
12:55: Increased MinMo to 0.40. At this point I also removed the motor cover on DEC so I could watch what was happening! (hence the large excursions in DEC and RA at this point :). With MinMo set to 0.40 however, guiding stays within the seeing limits and DEC oscillations disappear.
01:01: Reduced MinMo back down to 0.20.
01:02: Reduced MinMo back down to 0.10. Immediately start to see oscillations again
01:04: Increased MinMo back up to 0.30. Oscillations stop
01:06: MinMo back down to 0.10. Oscillations start again
01:08: MinMo back up to 0.30. Oscillations stop etc"
So, he already knows how to stop oscillations or what I call chasing the seeing. SET the MIN MOVE to a level that corresponds to the seeing. There is never a time when the seeing is 0.1 arc sec, most likely it will be 0.5 to over 1 arc sec., and this can be seen by simply taking a few minutes of Dec data with guiding turned off. You will see the star bounce around +- X arc seconds. Using that data, set your Min Move to between 50% and 80% of that value. There is no use trying to get a guide loop to hunt down atmospheric motions of less than that because the error measured comes well after the fact and by the time a correction is sent to the mount the star probably is elsewhere. A lot of times the error gets compounded instead of reduced, and it begins a back and forth oscillatory motion.
The other issue is the amount of correction by adjusting the aggressiveness to less than 100%. Near 100% you may exceed the 100% loop gain and instruct the mount to move more than the error size, and this will indeed set off oscillations. Close loop gain must always be less than 100% for any servo loop to be stable.
So would somebody that frequents CN please advise him of these simple facts? I am not on CN, but have covered this topic here extensively.