As we have already seen in E787 data, the B-field keeps decreasing, resulting in a higher momentum observed in the monitor samples. There is  an 0.4% B-field variation in average for the E949 2002 data . An investigation of how well our two hall probes installed inside the detector  (where the black dots are for the Kp2 momentum using the recorded B-field, while the open triangles are for the one using a constant 10 kG B-field. To avoid dip angle momentum dependence, a cut on |cos3d|<0.1 is required.) is done, showing an about 0.3% higher of mean Kp2 momentum in 2002 data, in which the calibration constant for B-filed is still adopted the one for 1998 data. Nevertheless, the variation of B-field are correctly recorded and has already been applied in our standard data processing.  As for the calibration constant  for the absolute B-field value can be easily put in  'my_end_spill.F'.
  In 2002, the center of UTC is shifted by1 cm to upstream in order to match the center of T counter, which was claimed to be the center of the detector. Under this assumption, the Z position at I-counter, T-counter and the edge of target close to the B4 are checked ( see these plots from a study of cosmic rays taken in 2002 ). The edge of I-counter starts from Z = -11 cm and ends at Z = 13 cm. The length is 24 cm as what we expeted. But the center is not at zero. As a result, the center of I-counter in UMC  is shifted. The target center is also checked indirectly by looking for the Z position for x=y=0 with and without a B4 hit.  The edges of target  are found to be -11.5 cm and 13.5 cm at upstream and downstream side separately. Again, the length of target fiber is found to be 24 cm, agreeing with what we expected. But the center is at about 0.5 cm. The center of T counter is observed to exactly at zero with a 52 cm length. A positive overshoot on the UTC analog pulses were observed in E949 when the kaon incident rate went up. It caused a serious problem of UTC acceptance loss because the ADC of the inner two UTC foil layer ceased functioning. During the run stop for the NASA run, an UTC  electronics modification was done to accmmodate the ADC baseline to the rate, mostly resolving the problem of UTC acceptance loss.  At the mean time, a rescuing effort was done by exploiting the TDC information recorded  to reduce the magnitude of the problem before the NASA run. Apart from this, the loose z resolutions were used in the UTC track fitting, substituting the previous value of 0.2cm used for all the foil layers (users can always  use kofia command utczres(1-6) to input the proper values). At present, the utczres values are set to be  0.51, 0.43, 0.28, 0.31, 0.31 and 0.37 cm, which are taken from the z residuals study. Also note that these utczres values can only increase the z points found along the track, but they cannot help for any attempt on improving  the momentum resolution. A plot of UTC acceptance vs. run shows the results with or without these efforts added in software, where the open triangles are for the efficiency without these efforts and the black dots are for the efficiency after all the efforts were put in (here what I have to mention is that the efficiency in 1998 data is 97%). Unlike many high energy physics experiments, in E787/E949, charged particle momentum is measured by the target and the UTC, and this is the reason for why it is sometime refereed as PTOT. The target provides the determination of the kaon decay position, and then the track extrapolation from UTC will extend to this position, giving the path length from the decay position to the outer surface of the I-counter. This path length is transformed into the range in the sincilator, which will be added to the range predicted by the UTC momentum measurement. The total range value is then converted into the momentum. The contribution to the momentum measurement from the target and the I-counter is less than about10% of the total.That is how a charged particle momentum is measured in E787/E949.  To improve the momentum resolution, a number of modifications either in the software or in the calibration files have been done, such as the TDC information used in the strip hit finding, the UTC geometry correction, the target rotation correction with respect to the UTC and  the initial timings (t0)  and ADC calibration of the UTC anode and cathode channels. To check the improvement,   a comparison of momentum measurement between the ntuples created by the online autorun (from March to June, 2002, neither the software modification nor the new calibration files was applied.) and those created by the latest pass2 (in October, 2002, with all the modifications up to this month) shows some interesting results.
 
 
Ntuple samples  Comments
 Latest pass2 ntuples (2002) With all the software modifications and the new calibrations, the momentum resolution was improved. Interestingly, the pion momentum resolution doesn't show much difference between those before and after the UTC electronics modification, while the muon momentum  does. My model to this is that the pion ionization in the UTC gives rise to a relative higher pulse than that of muon for momentum around 200 MeV/c, resulting in a more efficient strip hit recorded (as shown in the histograms, where the solid one is for the kp21 monitor, while the dashed one is for the km21 monitor in the same run. It is noted that the number of Z hits significantly increases after the UTC electronic modification ( see the histograms. The solid one is for 1998, the dashed one is for 2002 before the modification and the dotted one is for 2002 after the modification. ). Since the number of Z hits can significantly affect the momentum resolution ( see the Km21 momentum plots with NZ=4 (histogram), NZ=5 (black dots) and NZ=6 (open triangles) ), increasing the foil efficiency (especially the inner foil efficiency) is very important for improving the momentum resolution. At the meantime, we can also observed that the momentum resolution with the same Z hits still look slightly difference between 1998 and 2002 data (see the plots for  NZ=4 ,  NZ=5  and  NZ=6 , where histograms are for 1998, open circles are for 2002 without electronic modification and black dots are for 2002 with electronic modification.). This second order effect has not been investigated yet.
 Autorun pass2 ntuples (2002) The mean of pion momentum looks a bit greater than that shown in the later pass2 ntuples. This is because the old target rotation file was used, driving the curve outward when undergoing the x-y track fitting with the target pion fiber constraint.  Also noted is  the  effect on the momentum resolution due to the UTC electronic modification. 
 Final pass2 ntuple (1998) The trend of momentum resolution and rate correlation is seen already in 1998 data, since the rate kept increasing with the run number. Given the fact that the rate in 2002 is about a factor of two larger, we should not expect a miracle of momentum measurement in E949. 

Work on improvement of the mometum resolution
1) First order correction
     a) TDC information  (done)
     b) Z resolution input in the fit (done)
2) Second order correction
     a) Z(E_K)  (done, but tiny improvement was obainted)
     b) Z(RSSC) (done, no improvement was seen)
     c) moving ADC pedestals (not done yet)
     d) Z(RS) in the inner 2-3 RS layer (not done yet)
     e) Cos3d and phi0 correction (done and ready for use, but the mechanism for phi0 dependence is still unknown)
     f) Understanding the low efficiency in the inner two UTC foil layers even after the electronic modification (partically done)
     g) Z position resolution
     ...

The range measurement is done by the target, the UTC, the straw chamber and the Range Stack. Like the way we work for the momentum, very often is the range value in E787/E949 refereed to RTOT. How the range is determined by the target and the I-counter has already been described in the section for the momentum measurement. Major contribution is from the Range Stack. The path length of charged particle is from a track in x-y plane that fits best to the sector crossing,  the Range Stack straw chamber hits, the position of each of each hit layer, and the energy in the stopping layer. Scaled by the dip angle measured by the UTC, the range can therefore be obtained from this path length.
 
 
Ntuple samples Comment
  Latest pass2 ntuples (2002) The range doesn't show too much dependence on the UTC performance, since no clear difference is observed between those before and after the UTC  electronic modification. It was noted that the range drifts upwards by 0.5% at maximum after run 48500. This matches the +5 MeV energy offset around this period , but it is still hard to quantitatively give a value to check if this amount is consistent with the offset seen in the range measurement, since only the energy in the stopping layer is used in the range measurement. The resolution looks quite stable regardless of whether or not the UTC electronics were modified, while the range doesn't, showing a slightly higher before the modification. 
 Autorun pass2 ntuples (2002) It is interesting that the upward offset after run 48500 seen in the latest pass2 ntuples  doesn't show up despite the fact that the same behavior is also seen in  the kinetic energy analysis . 
 Final pass2 ntuple (1998) A stable mean and resolution of range measurement is seen and also independence of the rate.

Work on improvement of the range resolution
1) Range Stack track fitting with multiple scattering (evaulation done in the UMC level, no more room for improvement)
2) Hardcoded parametered used in the fit (investgation underway)

The kinetic energy measurement (called ETOT in most case in E787/E949) is done by summing over the energy deposit in the target, the I-counter and the Range Stack. Energy loss in the dead material is taken into account. Up to date, the energy deposit in the kaon fiber is not added, since it is hard to be separated from the kaon energy. Obviously, it is independence of both the momentum and range measurement.
 
 
Ntuples samples Comment
  Latest pass2 ntuples (2002) The energy peak is now quite stable.  Previously a worse energy resolution observed in E949 due to a broad run range calibration is fixed, a significantly improvement is seen. More discussions can be found in this web site.
 Autorun pass2 ntuples (2002) Since the energy calibration files were not updated, there is less meaningful to interpret the plots. 
 Final pass2 ntuple (1998) Both peak and resolution of pion kinetic energy look stable in E787. The small offset below run 38000 is known to be due to the energy calibration and has been fixed in the final pnn1 data analysis. 

Work on improvement of the energy resolution
1) First order correction
     a) Run dependence effect (done)
     b) 3 MeV RS energy measurement deficit (underway)

2) Second order correction
     a) Somewhat worse energy resolution from the target and I-counter (Understood, because we used CCD pulse fit to extract
          the pion energy in the kaon fiber, introducing some uncertainty, but the known energy deficit in the target is reduced.
          More detail information can be obtained from this web site.)
     b) Rate dependence correction (underway)