I believe the logic of (first 12 TOFS AND Cerenkov) OR (last 4
TOFS) is all within the sector MLU, it's just a pattern lookup. So there
should not be timing issues, because the logic operations are not
sequential. The only problem would be if the Cerenkov signal to the MLU
came so late that (TOF AND Cerenkov) came very late, after the TOF Meantime
signal, which is what really provides the event timing. However, that was
checked ages ago and shouldn't be a problem. We'd see it now in the PHYS6
singles events, as a shifted TOF timing peak.
One thing we might want to worry about is whether the last TOF
behind a Cerenkov (#11) should REQUIRE a Cerenkov or not, since there might
be an edge effect. For inbending electrons I wouldn't think there would be
a problem. Also, the way the TOFs go into the sector MLU (paired after the
first 4), you'd actually need to allow TOF10 to also not require a Cerenkov.
I'll keep thinking about other consequences, especially if they
could produce a false asymmetry, or some kind of additive background.
Karen
At 08:22 AM 3/5/2004 -0500, John Calarco wrote:
>Hi Richard,
>
>I may be wrong, which is why I opened my comment with "Be careful." ...
>meaning we need to think carefully to not make any mistakes. Unless I'm
>wrong (which may be the case), PHYS0 contains the elastic as well as
>quasielastic e,e'p. Thus a simple hardware AND of the CC takes the
>last 4 TOFs out of the trigger which means we get no triggers for
>elastic events in the last 4 TOFs, the high Q^2 region. It is possible
>of course to work around this and have a hardware trigger of (any TOF
>AND CC) OR (last 4 TOFs). This keeps the last 4 TOFs in the trigger,
>but now I begin to worry about timing. For every logical operation,
>we accumulate a delay of at least 10 ns in the trigger. If the 2
>Boolean operations I just described have to be done sequentially, then
>we have at least 20 ns delay in the trigger relative to the retiming
>signal. At some point we run into trouble. Perhaps Karen or others
>might wish to comment.
>
>So, to reiterate, "Be careful." Any change in the trigger is possible
>but MAY (repeat, MAY) have possible unforeseen consequences which have
>to be assessed before making the change.
>
>
>On Thu, 4 Mar 2004, Richard Milner wrote:
>
> >
> > I am not sure what you mean John. The idea is that where we have Cerenkov
> > detectors in the acceptance we should seriously consider to use them in
> > the electron trigger. They are now highly efficient and should
> > reduce trackless triggers, e.g. in (e,e'n) as a means to reduce the
> > deadtime.
> >
> > Elastic scattering on both the deuteron and proton has the benefit of
> > being completely kinematically correlated. My understanding is that this
> > trigger type is a relatively small contributor to the deadtime.
> >
> > Also, for the proton target we have only the (e,e'p) assuming we prescale
> > the inclusive.
> >
> > Am I missing something?
> >
> > Richard
> >
> > On Thu, 4 Mar 2004, John Calarco wrote:
> >
> > >
> > > Be careful. Now that we have removed CC3 from behind the 4 rearmost TOFs
> > > and put them behind the BATs, a hardware CC requirement restricts the
> > > high Q^2 end, and that's where the ed elastic T20 overlaps the very
> > > interesting region where the old Bates data overlap the recent JLab
> > > data. I don't think we want to sacrifice that, and I definitely want
> > > the high Q^2 end for the ep elastic.
> > >
> > >
> > > On Thu, 4 Mar 2004, Karen Dow wrote:
> > >
> > > >
> > > > Just spoke with Richard on the phone, he requested that someone
> check how
> > > > much a hardware Cerenkov requirement would cut the trigger rate,
> hoping to
> > > > reduce the rate of PHYS1 significantly (and possibly also
> PHYS0). Tavi and
> > > > Baris will look at crunched data while they're on shift, see what a
> > > > Cerenkov cut does to the number of trig==2 and trig==1, also what
> it does
> > > > to the spectra (z, momentum, theta etc -- presumably we don't lose
> good
> > > > events).
> > > >
> > > > Karen
> > > >
> > > >
> > > > At 01:42 PM 3/4/2004 -0500, Richard Milner wrote:
> > > > >Following Tancredi's mail, I think we should significantly
> prescale the
> > > > >inclusive and put more lead shielding in front of the forward
> LADS. Ernie
> > > > >is working up a modification of the collimator which has the
> potential to
> > > > >improve the deadtime situation for the inclusive trigger. Until
> we can
> > > > >implement that, we should optimize running conditions for (e,e'd),
> (e,e'p)
> > > > >and (e,e'n) both vector and tensor.
> > > > >Richard
> > > > >
> > > > >
> > > > >
> > > > >On Thu, 4 Mar 2004, Tancredi Botto wrote:
> > > > >
> > > > > >
> > > > > > A brief summary of the present understanding of deadtime
> sources from the
> > > > > > analysis of recent data:
> > > > > >
> > > > > > Deadtime is limiting us in the use of higher beam currents.
> There are many
> > > > > > components to this: the most significant is "trackless"
> triggers that pass
> > > > > > the 2nd level trigger thanks to random hits in the wch. The
> ratio of
> > > > > > these fake 2ndl level triggers (abot 2/3 of total data) is
> consistent with
> > > > > > the Wch S/N ratios. The ratio of trackless triggers is nearly
> independent
> > > > > > of trigger type.
> > > > > >
> > > > > > Trackless triggers have no known vertex or momentum
> distribution of course
> > > > > > but they contribute fully to DAQ deadtime. They are very
> sensitive to Wch
> > > > > > multiplicity and S/N. Possibly this is related also to the
> collimator
> > > > > design.
> > > > > > Trackless events really have too few wch hits (often < 3 hits
> in the tdc
> > > > > > range used in the reconstruction of the wch events). We can't use a
> > > > > > momentum cut to truly speak about deadtime..
> > > > > >
> > > > > > A second contribution is coming from low-momentum particles
> that originate
> > > > > > mostly upstream of the target. These events constitute the vast
> majority
> > > > > > of "tracked" triggers, but a smaller fraction of the overall
> yield. They
> > > > > > are well characterized in momentum (100-200 MeV/c), charge
> (positrons for
> > > > > > inbending field, electrons for outbedending - both fire the
> Cerenkovs) and
> > > > > > location in the detector (tof #'s 10-14).
> > > > > >
> > > > > > These events must originate from 300 MeV photons in a EM
> shower. The
> > > > > shower
> > > > > > having photons (which are not "bent") may contribute again to
> the Wch S/N.
> > > > > > Note that trackless triggers are instead *uniformly*
> distributed in the
> > > > > TOF's
> > > > > >
> > > > > > We have never experienced such a harsh environment before
> because we were
> > > > > > not running with an inclusive trigger (requires a cerenkov)
> prescaled by 6
> > > > > > and because we did not add the LADS to the e,e'n trigger.
> Having done so
> > > > > > it offers many more opportunities for trackless and
> low-energy-background
> > > > > > triggers. Indeed trig==2 and trig==7 are the dominant
> distribution of
> > > > > > trigger types.
> > > > > >
> > > > > > To make matters worse, any of these trigger rates will show a
> dependence
> > > > > > on beam current and as mentioned in the prev email it is
> important to
> > > > > > operate in a linear region. Regards,
> > > > > >
> > > > > >
> > > > > > -- tancredi
> > > > > >
> > > > >
> ________________________________________________________________________________
> > > > > > Tancredi Botto, phone: +1-617-253-9204 mobile:
> > > > > +1-978-490-4124
> > > > > > research scientist MIT/Bates, 21 Manning
> Av Middleton MA,
> > > > > 01949
> > > > > >
> > > > >
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> > > > > >
> > > > > >
> > > > >
> > > > >--
> > > >
> > > >
> > >
> > >
> >
> >
>
>--
>John R. Calarco
>Dept. of Physics
>Univ. of New Hampshire
>Durham, NH 03824
>phone: (603)862-2088
>FAX: (603)862-2998
>email: calarco@unh.edu
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:30 EST