Study of SVT track/hit finding efficiency


This is intended to be a short summary of the study of track/hit finding efficiencies. Detailed descriptions of the methods and definitions used can be found here.


We have looked and the phi hit finding efficiency vs. phi' for the three inner layers, for the following two conditions: The minumum transverse momentum Pt was taken to be 0.3 GeV/c.

The results can be found here (for forward and backward half-modules). The vertical lines on the plots correspond to the phi' that straight tracks going through the edges of the wafer would have (so anything beyond those lines is due to very soft tracks). Note that while the hit finding efficiencies for the two different layouts do not appear to be dramatically different for phi' less than about 0.2, for large phi' the larger bonding pitch improves the efficiency quite significantly.

We have also looked at the efficiencies for noise 60% higher than what has been measured and with 3% random channel inefficiency. The results are available here.

To ease the conversion from phi' to chip number, plots of chip number vs. phi' are available for layer 1, layer 2, and layer 3, also for both forward and backward half-modules. One might find it puzzling that the chip number appears to be not quantized: this is explained by the fact that the chip number here is really an average chip number for the digis which constitute a given cluster (or hit). The plots also have some amount of scattering due to soft tracks.

One can also compare the track finding efficiency for these conditions (in these studies, we have changed the pulse height cut for space points to a value less than 1; see, for example, this hypernews posting and the subsequent messages). Plot (a) shows the track finding efficiency as a function of transverse momentum Pt, and plot (b), as a function of cos(Theta).


We have also looked at the "impact parameter" resolution for the two different setups described above, and for Pt > 0.5 GeV/c. Here are the plots of the resolution for four different phi' bins, and below is a table summarizing the results:

Phi' Range Old setup resolution New setup resolution
-0.5 < phi' < -0.2 112 microns 122 microns
-0.2 < phi' < 0.1 111 microns 119 microns
0.1 < phi' < 0.4 129 microns 128 microns
0.4 < phi' < 0.7 158 microns 140 microns

Note, however, that these results are based on the SvtTrackFinder alone, and so track parameters do not include Kalman fitting (accidentally, this is also the most likely reason for the zero offset one can see in the resolution plots). We are currently working on including Kalman filtering in our set-up. However, one can already tell that at any rate the "new layout resolution" does not worsen significantly for low values of phi' and in fact it appears to get better by ~10% for large angles (which is indeed what one might expect).

Note on number of channels per chip

While doing these studies, we have noticed that the current simulation does not assign the correct number of channels to the first chip (chip 0) for layer 1 modules. It assumes that chip 0 is fully bonded (has 128 channels) and chip 6 is partially bonded, whereas in reality the situation is reversed: chip 0 is partially bonded, and chip 6 is fully bonded (some plots are available here). As far as we could tell, the number of channels per chip for layer 2 and 3 modules is correct.

Also, simulating the 100 micron phi bonding pitch the way we have been doing it (by changing the layout geometry datafile, svtgbbsim_273.dat in SvtGeom, as described, for example, here), does not appear to work as far as assigning the right number of channels to chips is concerned. These plots of chip number versus phi strip number show that the simulation still assumes 128 channels per chip (as opposed to 64). Note also that for layer 1 the first chip is again assumed to be fully bonded.


Taking into account all the above mentioned considerations, it appears that the best efficiency and resolution will be achieved if:
This bonding scheme has been accepted as final in October '98.

Questions or comments? Please send e-mail to Natalia Kuznetsova

Page Last Updated: November 20, 1998