Date: Sat, 18 Feb 2006 12:00:36 -0700 (MST)
From: Walter Brisken <wbrisken@aoc.nrao.edu>
To: brunthal@mpifr-bonn.mpg.de
cc: Data Analysts <analysts@aoc.nrao.edu>, Craig Walker <cwalker@aoc.nrao.edu>
Subject: Re: bb200c (fwd)
Message-ID: <Pine.LNX.4.64.0602181149300.5012@parallax>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed


Hi Andreas,

It appears you are doing more than just changing sources every 30 seconds. 
If you look at the crd files made by sched:

!* --- Scan from 04h43m54s to 04h44m24s   Tue, 2006 Jan 17 --- *!
sname='J0137+312'  ra=01h37m08.733637s  dec= 31d22'35.85551"  qual=  0 
calib='A'
bbsynth=( 1,728.51),( 2,728.51),( 3,736.51),( 4,736.51),( 5,744.51),( 6,744.51)
bbsynth=( 7,752.51),( 8,752.51)
pcalxfreq1=(1,490),(2,490),(3,490),(4,490),(5,6490),(6,6490),(7,6490),(8,6490)
pcalxfreq2=(1,490),(2,490),(3,490),(4,490),(5,6490),(6,6490),(7,6490),(8,6490) 
tape=(1,+RUN)  write=(1,on)
stop=04h44m24s   !NEXT!

!* --- Scan from 04h44m24s to 04h44m54s   Tue, 2006 Jan 17 --- *!
sname='M33/19'  ra=01h33m29.431047s  dec= 30d31'55.40191"  qual=  0 
calib=' '
bbsynth=( 1,730.28),( 2,730.28),( 3,738.28),( 4,738.28),( 5,746.28),( 6,746.28)
bbsynth=( 7,754.28),( 8,754.28)
pcalxfreq1=(1,720),(2,720),(3,720),(4,720),(5,6720),(6,6720),(7,6720),(8,6720)
pcalxfreq2=(1,720),(2,720),(3,720),(4,720),(5,6720),(6,6720),(7,6720),(8,6720)
tape=(1,+RUN)  write=(1,on)
stop=04h44m54s   !NEXT!

The synthesizer frequencies are changing because you are changing the 
doppler reference.  Maybe Craig can add more input here, but I think you 
can prevent the problem (or minimize it at least) by changing:

source='J0137+312' dur=00:00:30 dopsrc='M33/50' gap=00:00:00/
!
group=4 repeat=6
source='M33/19'    dur=00:00:30 dopsrc='M33/50'/
source='J0137+312' dur=00:00:30 dopsrc='M33/50'/
source='B4'        dur=00:00:30 dopsrc='M33/50'/
source='J0137+312' dur=00:00:30 dopsrc='M33/50'/

To Something like:

source='J0137+312' dur=00:01:00 dopsrc='M33/19' gap=00:00:00/
group=2 repeat=6
source='M33/19'    dur=00:00:30 dopsrc='M33/19'/
source='J0137+312' dur=00:00:30 dopsrc='M33/19'/

! Wait 1 minute on source for Mark5 sync after setup change
source='J0137+312' dur=00:01:00 dopsrc='M33/50' gap=00:00:00/
group=2 repeat=6
source='B4'        dur=00:00:30 dopsrc='M33/50'/
source='J0137+312' dur=00:00:30 dopsrc='M33/50'/


In any case, you can always sanity check this yourself by looking at the 
crd files.  If you have pcal on and pcalxfreq1/2 are changing, or if 
recording format/bitrate/fanout changes then the formatter needs to pause. 
Pauses are OK, but they have a recovery period.

Checks like this should be added to Sched.  Sched checks for some cases 
where this may happen (notably the format/bitrate/fanout changes) but the 
doppler change may have slipped by and not been detected.

-Walter


>     Preliminary notes and information:
>     General:
>       Weights plot reveals bad recording at all disks between 04:25 and
>       04:55. Investigation determined that changing setups every 30
>       seconds forced the Mk5 disks to go into "pause" mode, where no
>       recording takes place, further agravated by the correlator need to
>       resync after each change. Data was lost during this period.
> 
> 
> 
> What is meant with "changing setups"? I changed only sources every 30 
> seconds. Is there any way from prventing this in the future?
>  Cheers,
> Andreas
>