Date: Sun, 10 Mar 2002 16:37:19 -0700
From: "s. a. snedden"
Subject: Re: 2.5m obslog 20020308 (MJD 52342)
I don't think Watcher will have to be that much smarter (as for the
observers, well ...?). All we really need is for the PGPLOT display to
show both nominal ranges all the time. If we see two ranges on a plot,
we know it's a chip that can be in one of two voltage states and the
measured voltage had better be in one of the plotted ranges.
Steph
Jim Gunn wrote:
>
> At the moment, VS+ is the only voltage we change on the CCDs between
> reading and idle, but Connie and I are strongly considering changing
> others to make the CCDs dissipate less heat and be less liable to
> charged particle damage when idle, so either the observers or Watcher
> need to get smarter. I don't think it will be trivial for Watcher, because
> it will need somehow to be informed when the change is made, and
> especially if it is done in the firmware that synchronization will be
> well-nigh imposible. Simply having two allowable ranges opens the door
> for real errors--maybe Watcher should, before it reports an error, wait
> for another complete status report so it knows what the CURRENT settings
> are.
>
> --jim
>
> On Sun, 10 Mar 2002, Scot Kleinman wrote:
>
> > >Jon Brinkmann wrote:
> > >This is a well known, documented "feature" of all the cameras.  See
> > >
> > >     http://sdsshost.apo.nmsu.edu/folklore/spectro/index.html
> > >
> > >Jon
> > >
> > >> VS+ on the Spectrograph CCDs:
> > >> -----------------------------
> > >> On fMJD 52342, around 0005Z, watcher reported a vs+ error on CCD04.  The
> > >> real-time display showed vs+ on CCD01 as +0.04V, and CCD02/3/4 as +0.06V.
> > >> After clearing the error, watcher reported "vs+ is within limits for all
> > >> CCDs".  That's fine, but a _plot_ of vs+ for the spectro CCDs showed the
> > >> nominal range as about 7+-0.5V, and the CCD04 value as about zero since at
> > >> least 10am MST (1700Z), so we were disconcerted.
> > >>
> > >> We checked with Alan Uomoto, then Connie and learned that vs+ should
> > >> indeed be registers are being read.  So, the problem seems to be that
> > >> watcher is not smart enough to plot two nominal ranges for a voltage. A
> > >> PR was filed.
> >
> > This may indeed be true, but since we continue to get "bitten" by this
> > feature, it would be nice if something were done to avoid the confusion.
> > Unoubdtedly, somebody in the future will forget about this piece of
> > folklore again and will be confued by this.
> >
> > scot
> >