Lev 0 delayed coincidence

TG, RS, DC, TRKTIM setup

Gross p cut

1 matched RS-DC track

Dip angle < 10 degrees

The dip angle cut was to avoid having track-reconstruction foibles cause bias in the stopping distribution. For events passing the cuts, (x,y) of the Kstop from TARBNK and z of the DC track projected to the target edge were recorded in the beam file. (Note the dip angle cut mitigated the error in projecting the DC track only to the target edge.)

The K is pulled back and released with 50 MeV so that a findable Kaon blob will be left in the target. The K's are given directions such that they will typically travel .5 cm transversely.

A complication of the "real beam" methods is that using triggered data will result in a bias of the stopping distribution if the trigger acceptance is not independent of stopping position. We measure this bias using Monte Carlo. For the old beam files, muons with the Kmu2 momentum were generated at zero dip angle in 0.5 cm steps from r=0 to r=5 cm. The acceptance of the Kmu2 trigger was measured, and a table made. FUNCTION WGHT(RTGT) at the end of ustart.cdf uses this table to remove the trigger bias from the stopping distribution.

Problems with the old method:

-- The K's, when released with 50 MeV, did not go back to where they were supposed to. They stopped about 3 mm short.

-- Despite elaborate code to handle the case where the K stopped less than 50 MeV into the target, there were still anomalies in that case.

-- Since the K's were given some transverse momentum, they could actually leave the target before stopping. Since the common mode for running UMC is to demand a stop in the target, these events were cut. Since they come preferentially from near the edge of the target, this depleted the stopping distribution at large radius in a way that was NOT corrected by the bias correction.

-- The Kaon blobs were not realistic in the target.

-- Accidental hits in the target, important in the acceptance of the NTG part of refined range, were put in by incrementing NTG in the trigger calculation only. They therefore had no correlation to hits in the target in the underlying event.

Details_on_new_methods

Element is a one-dimensional version of (fib,row)

E is energy in 0.1 MeV steps

t is calibrated time in 1/8 ns steps starting at -50 ns.

______________________________________________ | | t (11 bits) | E (10 bits) | elem (10 bits) | ^--------------------------------------------^ msb lsb

The files are generated from Kmu2(1)'s in a similar manner to the old files. Differences:

-- SETUP_KINE is used for the setup cuts

-- Offline delco, as well as online, is applied.

-- An additional cut is applied: EIC < 4 MeV, where EIC is I-counter energy within 3 ns of TIMEK.

-- SWATH is used for the stopping (x,y). Rather than using the point returned by SWATH (the point midway between the centers of the two K fibers at the extremes of the swath), pick a point randomly between the two endpoints, after extending them to the edges of their respective fibers. The reason for this: the stopping points returned by SWATH can be no closer to the target edge than half a K-blob width (and certainly no closer than half a fiber width). This biases the distribution near the target edge.

-- z is found by projecting the UTC track to the point of closest approach to the (x,y) of the K stop.

Modes of operation are still controlled by IBMMOD. All new modes (IBMMOD = 3,4,5) read the same file; they just use different information from it. NEW BEAM FILES WILL ONLY WORK WITH NEW TARGET (ITGVER=1).

The real K hits appear in all the appropriate locations: track bank, energy arrays, TD arrays, trigger, output tape. The value of TIMEK stored on the beam file is subtracted from all times.

If requested (IBMMOD=5), the accidentals will also appear in all appropriate places, notably, in the trigger. Using this mode and a beam file from 1.2M KB/spill running, the average number of target hits found by the refined range code for UMC Kmu2's is 18.2, to be compared to 13.9 with old methods and 19.0 in data.

A new set of weights for the trigger bias correction was calculated by using a new beam file, generating Kmu2's with IBMMOD=-5, and measuring the acceptance. The profile matches the old values. I looked for a z-component to the bias (at fixed r), but found only small effects.

Note that the new method solves all the problems listed above.

Possible future improvements:

-- allow accidentals to come from different events than K blobs.

-- Add B4 hodoscope info to beam files.