Trigger work done during the January 2003 shutdown.
Rex Tayloe, Andrew Green, Richard Imlay.
Section 1: Work outline: The following is an outline of the planned trigger changes to be made ostensibly during the January shutdown. The number in parentheses is the planned order of the tasks. Thanks to everyone for the discussions that helped to clarify the trigger work.
Mainstream trigger software (default running mode)
(6)Prescale factors increased for Tank, Veto, and Michel triggers by about a factor of ~2 each (or more). This is just a matter of changing the file trigger_conf_file and the subsequent trigger type definitions in the data base, then documenting the final rate from each trigger.
(7)Add calibration triggers. For now, the "E3" external input will be used, and the length of the NIM input signal will determine the calibration event type.
Laser event: use a 450 ns NIM signal to E3 (will assert 4 or 5 trigger FIFOs). Event length is still 32 u-sec (32 TSAs).
Cube event: use a 150 ns NIM signal to E3 (1 or 2 trigger FIFOs) to signal a cube hit. The full trigger is: cube + DET4 (in a ~400-800ns window). The event length would be 9.6 u-sec.
Tracker event: use a 650 ns NIM signal to E3 (6 or 7 trigger FIFOs). This trigger will include cube + tracker coincidence (un-prescaled), as well as prescaled (on the LSU side) tracker triggers. The relative prescaling of tracker and tracker + cube is up to Morgan and Richard I. The event length would also be 9.6 u-sec.
Fake beam event: use a 850 ns NIM signal to E3 (8 or 9 asserted FIFOs). This trigger uses a "14"(p-bar stacking Booster event) and-ed with a "1F"(Booster extract) event from the IRM device (a device from beams division from which we can extract the various beam timing signals transmitted throughout the lab). This event is combined with a laser pulse a few us into the event window. The purpose of this trigger type is to check the possible change in our calibrations when synchronous to the Booster ramp.
(8)Add physics triggers.
Gamma: DET1 && !VETO1 up to 1ms following a nu candidate -- defined by DET2 && ! VETO1 inside a beam window.
Beta: DET2 && !VETO1 up to 30 ms follwing the same nu candidate. We will probably add a !VETO1 holdoff of about 1 u-sec.
Beta-prime: DET2 && !VETO1 up to 30 ms following DET2 && ! VETO1 inside a strobe window. This gets the accidental rate, and rate caused by cosmics. This rate and the following (#4) can also be compared with the out of time nu-candidates in a beam window.
Gamma-prime: DET1 && !VETO1 up to 1ms following DET2 && ! VETO1 inside a strobe window. This along with the Beta-prime trigger will provide more info for background estimates. I think that enforcing symmetry between the strobe and beam triggers is important. The rates should below enough that the additional event rate will not be a problem.
(9)Super Nova trigger modification. Push the DET setting to 60 hits instead of 100. Also use DET5 as the main trigger condition. The rate will be about 10 Hz.
(2)Expand TriggerShop to include a way to monitor the un-enable problem. More specifically to count the number of trigger FIFOs that are asserted when it happens, and what bits are set. For example, print an array of FIFO words when the trigger un-enables.
(1)Look at the data for events (trigger headers) which exactly preceed a "trigger UNENABLED" message in the daqLog (the trigger does not respond to the status register until buildTriggerData word has finished a contiguous set of FIFO loads). What kind of event is there? How many trigger FIFOs did buildTriggerDataWord assemble for that event?
(3)Introduce diagnostic code in triggerloop.c which provides more info about the un-enable problem.
Work with Richard V., Gordon on trigger hardware fixes.
(4)Enable bit spontaneously toggling. What is the source of the problem? Nailing this problem is essential to the future success of the detector.
(5)Investigate Broadcast errors. The main symptom is that after the broadcast card has delayed for a second or two, there is approximately 1/50 chance of the broadcast being incorrect (bad START event or bad RESUME event). There is an additional problem of the broadcast occasionally being missed altogether. Another class of broadcast error is a missed broadcast (diag = 0xffffffff), which is less frequent. Is there some state that is shown in the schematics that would cause these bad broadcasts?
Section 2: Justification for new trigger types in the mainstream trigger software.
The reason that we opt for the varying of the E3 NIM pulse is to give us the ability to trigger on calibration + DET bits. If we just have E3 and the calibration type (laser, cube, tracker) encoded in crate13, then we do not have that ability.
The primary motivation for this trigger is to get a better handle on neutral current p0 production via nm + 12C ® nm + p0 + n + 11C*, which has a 35% BR. Neutrons produced in the MiniBooNE tank will thermalize with a 186 microsec time constant and then undergo a capture reaction n+p to g+d. The g has a fixed energy of 2.2 MeV. Background events in the oscillation search from m misid and from NC p0 production will have on average more neutrons and therefore more 2.2 MeV g's than ne CC events. Mu misid events with m- capture have one extra neutron compared with nue events. p0, on average and extra 0.5 to 1 neutron. In addition, there is a m capture component that will also give g's from a subsequent n-capture. These are in Richard Imlay's presentations at Mini-BooNe meetings Nov. 2000 and July 6, 2000. The problem is that we may have too much background from radioactivity and cosmics. However, this background can be understood by having this same trigger follow strobe events.
There are a couple of major reasons to have a b trigger. First, we would be able to have another handle on the neutrino flux to the detector via the CC interaction, nm + 12C ® m + 12Ngs, 12Ngs ® e+ + ne + 12C. The lifetime of 12Ngs is 15.9ms, hence the 30ms window for this trigger. The second reason is that most of our m misid will come from m's that do not decay to michel electrons, but are captured by a 12C nucleus, resulting in the subsequent b decay (30ms lifetime) of 12Bgs . For more information, see Richard Imlay's presentation of April 28,2000, which Sam Zeller included in the copies of the LSND talk given to MiniBooNE in November. Once again, backgronds may be separated by having this same trigger trail strobe events.