Subject: Re: [sloan-25mlog 3207] Re: 2.5m obslog 20161117 (SJD 57710)
From: John Downey
Date: Fri, 18 Nov 2016 14:24:41 -0700
Hi Jim,
We are being careful of course, thanks for the heads up.
John
> On Nov 18, 2016, at 2:01 PM, jeg@astro.princeton.edu wrote:
>=20
>=20
> I hope I do not need to say this, but PLEASE be careful fiddling with =
the
> rotator PID coefficients. Keck just damaged badly a
> multimillion dollar instrument when the rotator went unstable, and
> we narrowly missed destroying the camera at least a couple of times
> during `tuning'. It is IMPERATIVE that someone be leaning on a
> stop button whenever this work is going on. You WILL damage the
> dewars/detectors and collimators if the servo goes unstable.
>=20
> --jim
>=20
> On Fri, 18 Nov 2016, John Downey wrote:
>=20
>>>=20
>>>=20
>>> Notes:
>>> -Towards the end of the night we started getting some odd out of =
closed loop
>>> axis behavior coupled with the rotator not going all the way to zero =
at
>>> instrument change. For the last cart of the night this escalated to =
the rotator
>>> not moving at all and going out of closed loop for any movement =
command. We
>>> called in JohnD who determined the cause to be the rotator motor and =
bearing
>>> that was replaced yesterday. We closed and John is working a fix. =
See log for a
>>=20
>> Dan, Alex, and I have been working on this, and we think we now have =
it to a stable state, to be tested all afternoon.  If it misbehaves =
again during testing I will stay around Sunspot at least tonight.
>>=20
>> We think the problem was two fold - the difficulty restarting the =
axis at all was fairly easy to correct by reseating the plug providing =
power to the motor.
>>=20
>> We've also found that the rotator drive is extremely sensitive to the =
tension on the upper rollers/main capstan of the motor.  It has to pinch =
just hard enough but not too hard.  The bolts for the upper rollers were =
inadvertently loosened yesterday while removing the motor, and this is =
most of what we have been tweaking and testing this morning.
>>=20
>> One added wrinkle we've found today is that the inability to easily =
adjust PID values for the servo loop is making the sensitivity of =
tension extreme.  If we could tighten the bolts to a reasonable degree =
and then tune the servo to optimize it for the roller tension.  There is =
an older ticket on this matter that was attempted to be addressed last =
June during the software conference, and while progress may have been =
made we still do not have a means of adjustment without rebuilding the =
MCP software each time a change is desired.
>>=20
>> Dan or I will likely bring this up with the infrastructure list soon.
>>=20
>>=20
>>=20
>>=20
>>> few more details.
>>> -Telescope balance: Going up, voltages weren't seen above 2000. =
Going down the
>>> max we saw was -3000.
>>>=20
>>=20
>> The telescope was slightly back heavy, which is not surprising since =
we had rebalanced at 30 rather than the more typical 50 ALT, and the =
dewars were probably half empty when we did.  While these voltage values =
are not excessive (depending a bit on velocity of course), I should have =
rebalanced at 50 when we were done.
>> ____________________________________________________________
>> Logs of Nightly APO Activities
>> =
https://trac.sdss.org/tracmailman/browser/private/sloan-25mlog/index.html
>>=20