On May 24, we figured out that central10-protected was misconfigured.  It was supposed to require TOFmult>th0, where th0 is 150, in order to break the degeneracy between very peripheral and very central events in the ZDC.  Instead, it required that the East ZDC (Au side) was greater than th2 on unattenuated sum, which is set to 650.  The central10 trigger was set up correctly.  Here I make a quick comparison between central10 (trigger id 410101) and central10-protected (trigger id 410112).  Also affects trigger id 410102.  We decided to update the central10-protected trigger id and change its name to central10a-protected.  central10-protected is suspect, even though it took 43M events.

After correcting for prescales and deadtime, the ratio (central10-protected/central10)/ (vpd-zdce-tac-protected/vpd-zdce-tac)  is approximately 1.3, meaning that the criterion for central10-protected is approximately 30% looser than that for central10.

Below is a comparison of the unattenuated east signal vs. tof multiplicity.  You can see that central10-protected has no TOF mult cut, and cuts off the low tail of the zdce unattenuated sum.  It's not a very large piece, but it is something to keep track of: looks like 0.2% of central10 do not have ZDC Unattenuated East < 650.

For historical reasons, "ctbsum" is actually the TOF multiplicity at level 0.

The effect can also be seen in the multiplicity distributions. There are clearly more (and broader) entries at low multiplicity in the central10-protected.

I normalized the two refmult histograms at refmult 250 to look in more detail at the distributions.  There appears to be a subtle loss of the low end of the high multiplicity tail in the central10-protected, vs. the central10, but this might be due to lower overall efficiency in central10 vs. central10-protected, due to the lack of pileup protection.   Black is central10-protected, red is central10.

In summary, there are some subtle differences introduced by the misconfiguration of the central10-protected triggers, so analyzers should be careful about using this dataset of 43M events.  Might be usable with work.

As another QA, below is the mean refmult vs. vertex z for 410007 (vpd-zdce-tac), 410101 (central10) and 410112 (central10-protected).  The VPD vertex z cut is clearly there as a bias,  but it does seem to be somewhat offset towards the west (cuts from -43 cm to 59 cm).  I attach it as a pdf so that people can read the fit parameters (double error function).