From: <Tasso.Tzioumis_at_email.protected>
Date: Fri, 16 Jun 2006 18:09:10 +1000

Hi everyone,

It was good to see people getting excited about fixing things. And I
do agree that many things need to be fixed and a concerted effort to
fix them is very welcome.

I also welcome the support offered from UTas, even though I do think
the problem is harder than "a couple of days" of work. I liked the
suggestion from Steven to have Jamie visit and "scope" the work. At a
minimum it will give him a better understanding of our systems.

Working with observatory staff is fine but a lot of the specific
expertise for VLBI at ATCA and especially Mopra is with the VLBI
group. We will be happy to be involved.

In fact I have also been pushing PCFS for a while now. It was rather
intractable or quite difficult until recent developments with
migration to Unix and the TCS system at Parkes and Mopra. As Chris
mentioned, we have already started work in that direction and we have
had Ed Himwich helping us. He is planning to visit later in the year
again.

Having said all that I do believe that unfortunately PCFS will NOT
solve all our problems. I think PCFS will be very good and will do
certain tasks very well. I think the main things are:
1. Antenna control at Mopra and Parkes. Development already underway.
2. Give good ONSOURCE logs at all antennas, especially for phase
reference observations.
3. Control the DAS. There are elements of that there already. I think
this is important as we are moving towards more complex and
dynamically changing schedules. I think some observatories still use
"DOS-in-a-box" which I remember as particularly falky.
4. Control the disk recorders. Work is progressing in that way and it
will be good to integrate it closely with the filed system.
5. LO control will also be good but it seems less suited to our
systems. For many observations is not just "set and forget" once, so
we do need to become clever about it.

HOWEVER, I believe that the PCFS will do little for the Tsys problem.
It is a weakness of the FS system. In fact the EVN have been trying
to sort this for years with only modest success.

It is also important to explain the Tsys problems i.e.
1. At the ATNF antennas we use "noise-adding radiometers" to measure
Tsys continuously and this is recorded in various ways at present.
Our problem has mainly been collecting this data in an appropriate
way and making it available to the user. And this has been
exacerbated recently with the migration to Unix at Mopra and ATCA.
However, this same migration to UNix makes the problem more
tractable. Unix scripts could handle this in an automated way and we
were moving in that direction even before the current discussion.

The suggestion of scripts or processes to transfer this to the FS
logfile has some merit but I think it only moves the data around.
Then you have to use another script to extract it (flogit?). With
Tsys values every 10s, this process will make the FS logfiles fairly
unreadable. My preference will be to write the scripts/processes to
produce the ANTAB files directly.

At the ATCA, although we are moving towards some remote control of
the caobs process, the main push has been towards ATCA functionality.
The VLBI is less than 10% of array operations and I will not expect
it to drive the process. What I have been trying to do with Euan is
improve the current "tied-array" operation and logging, so some of
the recent problems will not recur. I understand that the tied array
will also be used for the NASA 7mm system, so there is another
impetus to fix these things irrespective of the VLBI.

2. There is another Tsys "problem" that has not been discussed yet
but it is potentially more serious, especially for phase referencing
observations. This is a problem when the FS system measures the Tsys
as it is done at Hobart and Ceduna, and it is the problem that has
been bothering the EVN for a very long time.
Measuring Tsys through the FS uses FS procedures to call a routine to
turn on/off a noise diode and record the result. This is a rather
large noise source and affects the system and hence it is usually
done only about every 1/2 hour. In contrast "noise adding
radiometres" use small noise diodes and synchronous detection and add
only about 2-5% to the system noise.
Tsys every 1/2 hour is not that useful to a phase reference
observation that changes sources every 2 minutes. So, it will be good
to look into continuous Tsys systems for all telescopes (and the EVN
is moving in that direction).

So, in summary, I see 3 development strands that will make our
observations more reliable:

A. Use the FS for the tasks it is good at (control of telescopes,
disks, DAS,LO; onsource logging)

B. Develop scripts/processes to automate extraction of Tsys
information from the ATNF antennas, move it to ANTAB format and pass
it to the correlator for distribution to the PI.

C. Develop continuous Tsys systems for the telescopes that do not
have them now.

Thanks everyone for the contributions to this discussion.

Cheers
Tasso

--------------------------------------------------------------------
Tasso Tzioumis, Australia Telescope National Facility, CSIRO
Location: Cnr Pembroke & Vimiera Rds, Marsfield, NSW, 2122, AUSTRALIA
Post: PO Box 76, Epping, NSW 1710, AUSTRALIA
Phone: +61 2 9372 4350 Fax: +61 2 9372 4450 or 4310
Email: Tasso.Tzioumis_at_csiro.<!--nospam-->au
URL: http://www.atnf.csiro.au/~atzioumi/
--------------------------------------------------------------------
Received on 2006-06-16 18:09:33