There is indeed now a policy to avoid having bunches which collide only in IP8 as much as possible, as these showed much higher losses than other bunches. At least one dump was caused when these bunches fell below the sensitivity threshold of the position monitors, even though the other 1368 bunches were fine.DCWhitworth wrote:Yes a new record as announced here - https://twitter.com/#!/CMSexperimentLongshot wrote:Beams just dumped on a 170 pb^-1 fill! Is this a record for the LHC so far?
Now filling with a new scheme. Still 1380b, but few more expected bunch pairs at IP 1 and 5, (CMS/ATLAS) and a few less at point 8 (LHCb). Anyone know why the change?
Not quite sure why the new scheme but IP1 and 5 are CMS and ATLAS so they'll take all the collisions they can get. The thought appears to be to avoid IP8 only collisions (i.e. bunches that only collide in IP8), not sure why need to read a few more reports.
When two bunches collide, those particles which are not involved in actual collisions still receive a sideways push due to the electromagnetic field of the other beam. This 'beam-beam kick' is compensated using various corrector magnets (there is also a 'beam-beam tune shift' which is compensated by tuning the quadrupoles). The size of the beam-beam kick depends on the number of IPs each bunch collides at and the arrangement of neighbouring bunches at that IP. The isolated bunches which collide only at IP8 do not receive much kick, they are then over-compensated by the corrector magnets (which are trying to correct the average kick for all bunches) leading to a bad orbit and lots of losses.