Preliminary efficiency results for the February cosmic runs
The data:
:
I have looked at the following runs in detail:
| Run# |
Occupancy |
Configuration key |
| 990225-0831 |
0.5% |
7000 |
| 990302-1211 |
0.5% |
7008 |
| 990227-0710 |
2.0% |
7006 |
| 990227-0838 |
5.0% |
7007 |
| 990301-0624 |
10.0% |
7003 |
| 990301-0852 |
20.0% |
7004 |
Tools, parameters, etc.:
:
- Track finder:
I have used SvtTrackFinder with minimal modifications needed to match the cosmic running conditions.
- Alignment:
Survey alignment information provided by Eric Charles was loaded into the SVT conditions database and used in the usual way.
- Analysis:
- The efficiency analysis was similar to the one done for the January cosmic runs (see, for example, this and this for description).
- Hits expected in the inactive regions of the wafers were not taken into account
- Hits in the known bad modules (see Eric's postings on this, this one and this) were also ignored.
- In order to be able to look at the runs with very high occupancies (10-20%), I had to pre-select digis with the "right" corrected time (falling in the window between 6 and 11 s, see this plot of the space point corrected time for one of the low occupancy runs; as far as I've been able to find out, the timing was the same for all the runs).
Efficiencies:
:
The following table summarizes hit & space point finding efficiencies (recall that ephi = phi hit finding efficiency, ez = z hit finding efficiency, ecomb = phi and z hit combining efficiency, esp.pt. = space point finding efficiency (ephi*ez*ecomb):
| Run# |
Occupancy |
ephi |
ez |
ecomb |
esp.pt |
| 990302-1211 |
0.5% |
99% |
99% |
99% |
97% |
| 990227-0710 |
2% |
96% |
98% |
96% |
90% |
| 990227-0838 |
5% |
97% |
99% |
91% |
87% |
| 990301-0624 |
10% |
96% |
98% |
90% |
85% |
| 990301-0852 |
20% |
99% |
99% |
85% |
83% |
This plot shows space point finding efficiencies as a function of the occupancy.
Some conclusions and concerns:
:
- It is remarkable that in all of the runs, regardless of the occupancy, we get very high single hit efficiencies, so our clustering procedure seems to work very well.
- The main source of efficiency loss appears to be combining phi and z hits into space points. For the high occupancy runs I had to set a fairly high cut on the pulse height (1.4). For lower occupancy runs, a lower cut made a signifant difference: e.g., for run 990302-1211 (0.5% occupancy) going from 1.4 to 1.1 led to a 3% increase in the space point finding efficiency. Obviously, this needs more work and optimization.
Cosmic ray gallery:
:
Just for fun, check out these nice pictures of the cosmic tracks:
Questions or comments should be directed to Natalia Kuznetsova
... last updated 3/12/99
