Polarozattion Pattern in FY03 run

Fill #        Run#                 Pattern    SpinBit
3612 - 3614   4123008-4123133	   1          ok 
3615 - 3627   4123134-4125069      3          ok
3634 - 3793   4126001-4147016      1          ok
3796 - 3799   4147019-4148018      4          ok
3801          4148019-4148033	   4          bad
3803 - 3810   4148037-4150010      5          bad

+ = up
- = down
0 = unpol
x = not filled (abort gap)

Bunch/2       0         1          2         3          4         5
              01234567890123456789 01234567890123456789 01234567890123456789
Pattern 1: Y= ++--++--++--++--++-- ++--++--++--++--++-- ++--++--++--++-xxxxx
           B= +-+-+-+-+-+-+-+xxxxx +-+-+-+-+-+-+-+-+-+- +-+-+-+-+-+-+-+-+-+-

Pattern 2: Y= +-+-+-+-+-+-+-+-+-+- +-+-+-+-+-+-+-+-+-+- +-+-+-+-+-+-+-+xxxxx
           B= +-+-+-+-+-+-+-+xxxxx +-+-+-+-+-+-+-+-+-+- +-+-+-+-+-+-+-+-+-+-

Pattern 3: Y= +-+-+-+-+-+-+-+-+-+- +-+-+-+-+-+-+-+-+-+- +-+-+-+-+-+-+-+xxxxx
           B= ++--++--++--+++xxxxx ++--++--++--++--++-- ++--++--++--++--++--

Pattern 4: Y= -+-+-+-+-+-+-+-+-+-+ -+-+-+-+-+-+-+-+-+-+ -+-+-+-+-+-+-+-xxxxx
           B= ++--++--++--+++xxxxx ++--++--++--++--++-- ++--++--++--++--++--

Pattern 5: Y= -+-+0-+-+-+-+-+-+-+- +-+-0+-+-+-+-+-+-+-+ -+-+0-+-+-+-+-+xxxxx
           B= --++0--++--++--xxxxx ++--0++--++--++--++- -++-0-++--++--++--++
              0         1          2         3          4         5
Bunch/2       01234567890123456789 01234567890123456789 01234567890123456789

Pattern 2 was used early in the run for tests. No production runs were taken with this pattern.

Note: There is an irregular bunch in the blue ring for pattern 3 and 4. It's bunch 28 (14 at bunch/2 above), which is 3rd UP in raw. Be careful.

Note for transverse spin case : As it was last year, "+" means polarization is pointing "up" (pointing sky) at CNI polarimeter(12 o'clock). Since there is a snake between CNI polarimeter and STAR IR (6 o'clock), polarization is pointing "down" (pointing ground) at STAR IR.

Note for longitudinal spin case : At best knowledge at this point, supported by few experimental evidence as well as spin tracking program, spin rotator turns incoming polarizaion "up" into "plus helicity" (polarization is parallel to momentum) and "down" into "minus heliciy" (polarization is anti-parallel to momentum) at STAR IR for both ring. At PHENIX, this is opposit due to how IR looks like (at STAR both beam go from inner ring to outer ring. At PHENIX it's outer to inner ring).

Combining these, here is the summary table:

   Pattern     At CNI      At STAR IR (Trans)      At STAR IR (Long)
   +           Up          Down                    - or anti-parallel to beam direction
   -           Down        Up                      + or parallel to beam direction

Here is an example fortran subroutine (spinbit.f) to get 8 bit spin bits from bunch id. See details on spin bits for what 8 bits means.

Problem in last 3 days

Last 3 days of run, we apprently have problem in spin bits. For these runs, bunch counter seems to be working ok. Steve found this (see bottom of the page).

This is not problem of offline software. I see the same at trigger data, which is separate file from daq. Below, plots are # of spin bits counted as function of bunch id in trigger data.

  Black  = # of events
  Purple = # of "Filled bit"
  Red    = # of "Up     bit"
  Green  = # of "Down   bit"
NI  Blue   = # ofUnpolck" bit"

Upper plots are for Yellow Ring, and lower are for Blue Ring.

If everything is ok, Black=Purple=(Yellow + Green + Blue). So you should NOT see any Black nor Purple lines, because Purple should be on top of Black, and one of Red/Green/Blue should be on top of Purple.

Run 4148013 (fill 3799) looks good

Run 4148022 (fill 3801) looks bad.

Run 4149018 (fill 3803) is after pattern change, and looks bad too

 
  Yellow :about 1/2 the time, we loose all spin bits
  Blue   :about 2/3 the time, we loose filled bit
          about 1/2 the time, we loose up/down/unpol bit

In event dumps, I see

   - about half the time, it's OK (do not loose any bits)
   - many unfilled-unfilled (all bits lost)
   - filled but no spin info (some in blue ring)
   - not-filled but there is spin info (few in yellow ring)
   - never see wrong bits on

Osamume says C polarimeter DAQ was cross checking spin pattern from CDEV with one from v124, and no inconsistency was reported. So is at Phenix. This seems to be STAR local problem.

It seems for me a timing problem between v124 and trigger DSM, or between RHIC strobe and spin bits within v124. Exact cause of this is still unkown.

From trigger data, using the bunchid->spinbit code (link above), I have checked all available trigger data and found few bad runs:

Fill 3620  run 4124045-47 ppTune
           run 4124051-57 ppTrans-1
Fill 3652  run 4127044    ppTrnas-1
Fill 3654  run 4127045    ppTrans-1
Fill 3659  run 4128056    ppFPDTOFu
Fill 3801 and later (run4148025 and later)

These runs need to be looked at carefully, before it goes any spin analysis. The full list can be found here .

OGAWA, Akio

akio@bnl.gov