Dear JH, Ian and others,
On Wed, 12 Dec 2001, J.H. Lee wrote:
> During the data taking, drift velocities for TPC's varied due to some
> changes in gas flows, atmospheric pressure, temperature and possibly
> some unknown factors. As we know, it's important to calibrate the
> variations since the drift velocity is directly proportional to the y
> positions
> of the tracks. Even though y information is not used for the momentum
> calculations, it's still important to know it accurately because it's
> used for
> - Selecting tracks from the vertex (<y> - y, and BB vertex - y),
> - Selecting matched tracks (front track y : back track y),
> - Time of flight matching (projected y : tof y) and Path length,=
> and also used for
> - Normalization/acceptance corrections (delta y).
> Here are some plots illustrating how much the velocities vary in time.
Beautiful and extremely useful work, JH!
Your plots confirm that we probably have a "lower than usual" drift velocity
also for the 22 GeV part (run 5992 - 6008) which we suspected.
> To reflect the variations on the analysis,
> run-by-run y cuts (vertex, track matching, tof matching) need to be =
> used.
> And, for more accurate tracking/identification/yield calculations
> the dependence of velocities on positions ( x and y) of TPC's need to be
> taken into account in the analysis.
A small illustration of the importance of drift velocities:
Bjørn has been looking at TPM1 dN/dEta for 22 GeV - going from
dv = 0.0017 to 0.0016 brought dN/dEta from ~315 to 340 for the
"0-5% centrality bin" (still very preliminary, and with a highly
dubious centrality normalization.) So this is clearly an important
correction which may influence pbar/p ratios etc ...
and on Thu, 13 Dec 2001, Ian Bearden wrote:
> I have, in the past couple of days, reduced around 1000seq. from various
> settings, but used the 'default' drift velocities for TPM1 and TPM2. If
> someone (Peter???) could put these vdrift in the data base (or teach me how
> to do it) I can quickly redo the reductions that I have done. Also, as far
> as I know, the proper vdrift has been the only thing keeping us from a
> 'final' (ok, maybe semi-final) reduction pass, that is, one in which all the
> local tracking is done.
We should also agree on a set of local tracking parameters for reduction
... you NBI guys have them probably optimized for FS. What we have been
using in Oslo for MRS is the following (wider search parameters, which
give a bit higher tracking efficiency than default parameters) -
what have you been using at BNL?
BrTpcDeconvoluteClusterModule:
clusterdeconv->SetHighPsigCut(1.5);
clusterdeconv->SetHighTsigCut(1.6);
clusterdeconv->SetMinPadCut(2); // to get rid of noise clusters
clusterdeconv->SetMinTimeCut(3);
clusterdeconv->SetMinMaxADCCut(15); (10 for TPM2)
BrTpcTrackFollowModule:
tracker->SetSearchWidthX(0.4);
tracker->SetSearchWidthY(0.6);
> As soon as we have a set of Vdrift in the data base, I will get this effort
> going. One question: how do we do this for runs before the fibers were put
> in? Is it possible to correlate the Vdrift you measure with the results
> from the DVM?
Also what's the most efficient / reliable way to determine it for TPM1
which doesn't have fibres? Is it safe to assume proportionality to
the TPM2 variations (or DVM runs) or should one do more careful studies
of track intersection with beam plane and TPM2 fibres, MRS track match
parameters, bottomcluster positions etc. etc. ....
Have a nice weekend, and best wishes,
Trine
This archive was generated by hypermail 2b30 : Fri Dec 14 2001 - 12:31:49 EST