Status of TOF analysis/tools

Status of TOF analysis/tools.
Outline.


	Summary of TOF data structure changes.

	Calibration and monitoring procedures: how well do we deal with the variations in time delay?

	What do we need to know in order to get a good time-of-flight?

	Tracking to the TOF: wing update.

TOF/T0 Analysis Software

How to get a good time-of-flight?


	Many corrections need to be applied to go from the raw TDC channel to a good time-of-flight:


	where
		TOFraw = raw FB TDC value, converted to ns, based on FB time calibrations (TPhFBGainCalcMod). Calibrations carried out during the run (once per day), and corrections made during DV.
		Toffset = offset correction required to account for the variations in cable delay (TPhTOFTimeDelayMatchingMod). Calibrations carried out using the DV trees for each run, and corrections made during processing of PhoHit files.
		Tcollision = correction for jitter in the start of the TDC by using the collision time obtained from the T0 counters. Requires calibrated T0 data, and this correction is made during processing of PhoHit files.
		Slew = slewing correction. This correction is based on the know threshold and the measured PMT amplitude. Slew corrections are made during processing of PhoHit files.

TOF/T0 Calibrations.

T0 Gain Matching
TPhT0PHGainMatchingMod

T0 Time-Delay Matching
TPhT0TimeDelayMatchingMod

TOF Time-Delay Matching
TPhTOFTimeDelayMatchingMod

Time-delay Matching


	Relies on our capability to monitor the centroid of a distribution that is 750 ps wide or wider.
	The raw timing distribution are vertex dependent and thus need to be corrected for this dependence (done during DV).
	This procedure works well for the T0 counters were we have a no problem with statistics, and path length corrections (due to vertex variations) are simple to apply.
	The situation with the TOF is more complicated for a number of reasons:
		The path length variations due to vertex variations are hard to correct for using the raw data.
		For a narrow vertex region, the path length will depend on the emission angle and momentum of the particle (which will widen the distributions).
		Applying a narrow vertex cut reduces the available statistics.

Time-delay Matching


	In order to improve the quality of the time-delay matching, we need to use a calibration signal that monitors the cable delay continiously (e.g. use the time calibrator signals with a rate of 1 Hz and with period = range).
	We make the assumption here that the delay variations are due to changes in the delay of the delay cable. There is no reason to believe that the variations are due to the PMTs and in fact this can be demonstrated by comparing the timing of the ORs (which do not show the drastic variations, and travel over BNC cables to the CH) with the time of the individual detector signal which travel of ribbon cables.
	We need to develop the tools to use these time signals to monitor the time delay.

Quality of time-delay corrections.


	The quality of the time-delay corrections can be examined by comparing the T0 vertex with the silicon vertex.
	Any drifts in the delay time will change the position of the T0 vertex.
	Results over a 2.5 month period:
		Run 8045: dZ = +0.417 cm
		Run 8963: dZ = -0.035 cm
		Run 9174: dZ = -0.033 cm
	dZ = T0 Vertex - Silicon Vertex
	These results suggest that the calibration procedure is stable to better than 62 ps.

T0 Issues
Time jump for large T0 signals


	The T0 counters exhibit a saturation effect when many MIPS are detected simultaneously.
	This saturation effect shifts the time to longer times, and thus does not effect the T0 OR as long as not all PMTs saturate.
	We are currently investigating whether this is a discriminator problem or a PMT problem.

Slew corrections: T0 counters.


	Separate slew parameters were extracted for each T0 counter:





	These parameters are assumed to be constant during PR01.
	Applying the slewing corrections improves the timing resolution by a factor of 2.

Collision time and vertex resolution.


	Obtaining the T0 time:
		Use the fastest signal on either side. This will produce a multiplicity-dependent systematic shift towards shorter times.
		Use the average time of all T0 signals on either side. PMTs with large signals will induce a systematic shift towards longer times.
		Use the average time of all T0 counters that fire within 250 - 500 ps of the fastest one.
		Use the average of the largest number of T0 counters that fire within 250 - 500 ps of the each other.

Collision time and vertex resolution.

Tracking to TOF.
Central tracks.


	Tracking to TOF is now available in module format.
	Procedure works very well for tracks with p > 1 GeV/c.

Tracking to TOF.
Central and wing tracks.

Tracking to TOF.
Central and wing tracks.


	We have studied tracking to wall TC in detail.
	Track propagation is done using the procedure developed by Christof.
	The difference between the predicted hit position and the actual hit position is a function of the momentum of the track and is in the order of 1 - 2 cm. Tracking matching relies on an empirical correction we apply to the propagated track.

Tracking to TOF.
Central and wing tracks.


	Based on our studies of wing tracks we conclude that track matching is no problem for tracks with p > 1 GeV/c.
	Our studies show however, that there are more high-momentum tracks in TC coming from the central region of the spectrometer than there are coming from the wings.
	Maybe this can be understood in terms of geometry and vertex distribution, but this is not obvious.

TOF Multiplicity and TOF tracks.


	The TOF multiplicity is strongly correlated with the centrality of the collision.
	For B = 0 data, we have shown the measured multiplicity, after correcting for the secondary ratio, is consistent with the charged-particle yields extracted from the silicon data. The ratio of primaries to secondaries is 0.45.
	For field-on data, the number of secondaries is reduced and the ratio of primaries to secondaries is 0.53 (B-) and 0.59 (B+).

TOF Multiplicity and TOF tracks.


	Based on the number of TOF hits, we can estimate the number of primary tracks that should be directed towards the TOF.
	Many of the tracks in the track containers do not point towards the TOF, and the number of tracks shown in the figure is thus larger than the number of tracks directed towards the TOF.
	The observed correlation between TOF multiplicity and tracks suggest that there is an inefficiency in finding tracks directed towards the TOF.

Status


	All tools are in place for PR03.
	The calibration process works reliably. However, software issues related to DB Keys need to be addressed.
	The timing stability of the T0 detectors can be monitored using the data itself, but to monitor the TOF stability requires the continuous operation of the time calibrator during data collection.
	Although the fringe fields have an impact on trajectory matching the empirical corrections we apply are simple and sufficient. There is no indication that a more realistic field in the region between the spectrometer and the TOF will improve the agreement.