LER trickle injection monitoring
The postscript file for this page can be found here.
The plots drawn on this page are the following.
This plot shows the distribution of the number of hits per event in the HER
This profile plot shows the mean number of hits per event in the different
regions of the HER Weaver plane. Vertical error bars show the RMS of the
The trickle injection is a nice injection scheme, introduced by PEP
beginning of 2004 to maximize the integrated luminosity. Normally, the data
taking is periodically interrupted to let the accelerator crew refill the
beams. During these phases, the detector is ramped down to avoid damages and
so these periods are complete DAQ downtime. Then, when data are taken, beams
loss slowly charges and so the luminosity decreases until the next injection
period starts, which stops the data taking again.
When the trickle injection is ON, bunches are refilled at low frequency (5 Hz
for the HER, 10 Hz for the LER) to keep currents and lumi as constant as
possible. In this way, refill periods with BaBar down are no more needed and
so more data can be integrated as the machine duty cycle is higher. Yet, adding
charges in a bunch creates a lot of background for a short while the bunch
is circulating in the ring. As the time goes, the bunch gets more stable and
is finally undistinguishable from those which have not been refilled.
Therefore, the BaBar DAQ is synchronized with the trickle injection: it stops
when newly-injected buckets go through the detector. Indeed, events generated
in these conditions are so big and numerous that they would just fill the DAQ.
This unavoidable deadtime is more than compensated by the average luminosity
and duty cycle increases. A nice way to see this temporary DAQ inhibit window
working and to estimate the trickle injection impact on DIRC data consists
in using 2D plots showing the distribution of some quantities in the
The Weaver plane (named from Matt Weaver, one of the few key BaBar operations
people) shows on the x-axis the number of revolutions after the bunch injection
and on the y-axis the newly-injected bunch phase (i.e. its position over the
ring). As BaBar is located at a constant phase, the trickle injection inhibit
window is a white rectangle in the Weaver plane: events whose coordinates in
the Weaver plane would be inside this area are purely ignored. It is important
to understand that the origin (0;0) of the Weaver plane is resetted each time
a new bunch gets refilled: all 2D points are defined with respect to the
position of the last bucket where charges were added. Each point in the Weaver
plane is an event.
If the inhibit window is well-defined, the structure of the Weaver plane
outside it should be flat. To test this assumption, the Weaver plane has
been divided into several regions either in-phase or in-time with the
inhibit window. The region E -- 'the furthest' from the inhibit window --
is often used as reference.
Due to some code feature, the boundaries of the regions in the Weaver plane
(draw in red on the plot above) are currently not printed in the DIRC QA plots
produced by OPR -- January 2006; release 18.6.1c. This should hopefully be
fixed in the near future.
Back to the DIRC QA plots page