Dariusz Miskowiec - CERES analysis page
2000 data - general
2000 data - trigger
CRTrig* trig=event->getTrig();
if (trig.getPattern()&4 &&
// central trigger
trig.getNevadc()==1 &&
// only one ADC gate in this event
trig.getNevtdc()==1 &&
// only one TDC stop in this event
fabs(trig.getBC1Adc())<3 &&
// BC1 ADC in peak within 3 sigma
fabs(trig.getBC2Adc())<3) {
// BC2 ADC in peak within 3 sigma
// before/after protection
int nhit = 0;
for (i=0; i<trig.getNBC1(); i++) { // loop over
BC1 hits
if (trig.getTimeBC1(i)>-2000 &&
trig.getTimeBC1(i)<2000) nhit++;
}
if (nhit>1) {
cout<<"This event had additional
beam particle(s)";
} else {
// do here your cleanest, most advanced,
amazing analysis
}
}
As of September 2004, things evolved a little. A circular cut is
better for BC1 and BC2, and Oliver discovered that the before/after
protection in the form shown above under some circumstances
removes good events, and sometimes accepts bad ones. (This is because
the TDC can store only 8 hits, and if 8 hits occur before the good one,
i.e. the one that made the reaction, then the good one is absent in the
data and the idea of having exactly one in the window does not work.)
The correct way is as given below, sorry!
CRTrig* trig=event->getTrig();
if (trig.getPattern()&4 &&
// central trigger
trig.getNevadc()==1 &&
// only one ADC gate in this event
trig.getNevtdc()==1 &&
// only one TDC stop in this event
trig.getBC1Adc()*trig.getBC1Adc()+trig.getBC2Adc()*trig.getBC2Adc()<16 &&
// BC1 and BC2 ADC values in peak within 4 sigma
// before/after protection
int nhit = 0;
for (i=0; i<trig.getNBC1(); i++) { // loop over
BC1 hits
if (trig.getTimeBC1(i)>-2000 && trig.getTimeBC1(i)<-20) nhit++;
if (trig.getTimeBC1(i)>20 && trig.getTimeBC1(i)<2000) nhit++;
}
if (nhit>0) {
cout<<"This event had additional
beam particle(s)";
} else {
// do here your cleanest, most advanced,
amazing analysis
}
}
The variable names refer to step2-output format. To analyze
step3c output they need to be substituted with the corresponding
CSEvent data member names. It should be straightforward to anybody.
Not included above but recommended for RICH analysis are the
RICH recovery cuts.
2000 data - laser
TPC electric field
The 3-d electric field was first calculated
by ShinIchi
using Mafia and Maxwell. The field could not explain the observed curvature
(in radius vs z) of the laser tracks (laser smile). I started calculating
the TPC electric field in the hope that a careful implementation of geometry
and boundary conditions would help - but it did not. A break-through was
the discovery that some of the field cage resistors were changing resistance
under applied voltage. The effect was approximately R=R0(1-0.3 U) with
U being in kV. A calculation with the resistances modified according to
this formula nearly straightened the laser tracks.
Maxwell
Since I thought that a precise implementation
of geometry and the boundary conditions was essential I first decided to
use Maxwell.
A
screen shot of a half-a-chamber calculation (geometry + equipotential lines)
is here.
mymax
After we solved the problem of the smile,
however, we noticed that the accuracy of the maxwell calculation
was not good enough, and that it cannot be easily improved. I then tried
to smooth the bumps in the Maxwell potential. Of course, linear smoothing
would be not acceptable because the potential is a non-linear function
of the coordinates. Thus, I implemented a smoothing which obeys the Poisson
equation. Also, I thought I should make sure the smoothing fulfills the
boundary conditions. When I implemented this I noticed that the code worked
so well that I did not need Maxwell field at all. I could start from ANY
potential and the smoothing would ultimately bring me to the proper solution.
I called the code mymax. One mymax run gives a potential of half-a-chamber,
i.e. 1/32 of the TPC. It is orders of magnitude faster than Maxwell and
it has no bumps, see the comparison.
data format
The results of the Maxwell and the mymax
calculations are saved as potential on grid in the following format. The
grid is
grid3: (r 486..1350 by 1 mm, phi 11.25..22.50
by 0.25 deg, z -1100..1100 by 20 mm)
grid4: (r 486..1350 by 2 mm, phi 11.25..22.50
by 0.50 deg, z -1100..1100 by 50 mm)
R=486 is the HV cylinder. Phi=11.25 is
the center of chamber 0. Z=0 is the middle of the TPC. Only potential values,
not the grid, are saved on file. One number per line. The z-index is changing
fastest, i.e. the first number is potential at (r,phi,z) = (486,11.25,-1100),
the second at (486,11.25,-1080), etc. The files are stored locally on lxial15.gsi.de
in /work/misko/ceres/field.
Different files correspond to different
geometries, boundary conditions, etc.
The most prominent ones are:
-
ceres31 - Maxwell, badness as needed to straighten
smile, upstream 1/4 chamber and downstream 1/4 chamber calculated
separately, then combined.
-
ceres54 - Maxwell, 1/2 chamber. Semi-manual
seeding, thus more finite elements in drift volume, thus less bumps.
-
ceres64 - mymax, 1/2 chamber.
Using the seven laser rays in the special
sector of the TPC, located at the same phi and different radii, and the
electrons knocked out of the inner electrode cylinder at R=486 mm, the
electron drift velocity can be determined as a fuction of the electric
field. This
figure shows
1) the drift time measured for the seven
rays and the cylinder as a function of z
2) the mobility, i.e. ratio between drift
velocity and electric field, versus electric field; points are from Magboltz;
solid black line is from the fit to laser
data
3) radius reconstructed using TPC data
vs z, compared to the known position of the laser rays (lines)
4) one ray zoomed
mymax2d
Now that the smile and the bumps were
gone, we tried to take into account the small corrections of the readout
chamber positions. The calculation was done in two dimensions only. The
pages of this file
demonstrate 1) calculated potential, 2) agreement with the 3-dimensional
mymax calculation, 3) potential with tilted chambers, and 4) potential
with a displaced inner cylinder.
This internal note from 2006
describes my calculation of the electric field in some detail.
go back
to my homepage