some good data, more checks, considerations about background/trigger,
vacuum trip and software update.
Good elastic runs: 790 and 792 (the latter w/ modified trigger, requiring
================= or excluding cerenkov where appropriate). Run 790 (1 sccm)
is very long (ca 70k events).
Background and rates: When I take start coutner out of the trigger (still
===================== requiring a L/R coinc) the trigger rate goes up by
a factor 20/30 - ca. 20 Hz/mA which we can't handle.
Conclusion: we are actually *swmaped* by background but have never
looked at it yet. This coincident stuff is clearly coming from upstream of
the target as the start counter integrates everything coming out of the IT
point. Later in the day I play with the slits. The values at which slits
start affecting the beam are slightly different that what compton people
found. After injection, I am able to reduce the normalized rate by a
factor of 10, without many changes to the lifetime (anyways short today,
5.5 mins at 1 sccm). I can inject 60 mA with slits at some intermediate
position, reducing the normalized rate a factor of 3. Note slits movement
is slow (can be changed) so I decide not to ask for a dynamic setting yet
(inject, close, measure, open, re-inejct..). CLearly beam position should
be as stable as possible, but experience tells you may end up doing the
slits game over and over again. Sure it'll be more fun with a collimator
and field on.
KEEP start counter in trigger or coda will immediately crash.
Trigger studies: given the above, can not take out start counter
=============== to study the missing e-p coinc with protons on left.
Still missing. I play with cerenkovs (see logbooks) and
enofrce them in the trigger for selected paddles, can improve S/N ratio
of elastics a factor about 3 (!!) in L11-L12. This should also come out
of analysis (see later).
Vacuum problem: after overnight compton we are about to start. Set target
============== flow to 1 sccm. We then realizing injection valve was
closed and open it right away. Results in a vacuum
trip of the IT are. The transient flow (according to the epics screen)
went as high as 13 sccm (we know that don't trip at 10 sccm) at there
was beam on target. LIGIT switches off and turbos protected. Bad bad bad!
We understand that this could have happened at the flow controller, since
the gas pressure is not buffered. Found no comments about this
possible scenario in manual. It resulted in losing about 3 hrs before a
tech came in, brought roughin pump downstairs (no bypass either).
Workaround: it is very important to prevent operator dumbness.
========== I suggest that the control screens must inhibit setting
a flow if injection valve is not open.
During down time for vacuum crash we confirm that we are definitely over
any mis-cabling issue regarding tof's and cerenkov. Everything is in the
right place in the data and the analysis now. Finally!
Software: Little to add to jason's email. A number of people are taking
========= over this task and soon will have public utility routine. The
first goal is reproduce the 16x16 matrix plot, including
now making use of pid with the C data. Have fun!
Plan: as discussed with John. Tavi will progress in our studies of the
retiming issues (seem solved) and trigger electronics. In general,
cerenkov are still very hot, and some suspiciously low efficient put
go to back burner tonight.
Regards,
-- tancredi
________________________________________________________________________________
Tancredi Botto, phone: +1-617-253-9204 fax: +1-617-253-9599
research scientist MIT/Bates, 21 Manning Av Middleton MA, 01949
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:28 EST