DAMN Bank Bits
The DAMN banks store the results of a set of cuts applied to each event. More
details on the DAMN bank mechanics and the DMM processor can be found in the Chapter
Data Masks of the User Manual.
Within a DAMN bank each event has a set of bits, each bit indicating whether that
event was tagged by the corresponding cut. The convention on the bit setting is that
the it follows from the name of the cut. If a cut is called X then an event
identified as being of type X will have the X bit set to 1. (e.g. an event identified
as having a cluster will have the CLUSTER bit set to 1, otherwise it will be zero).
Whilst this often means that "bad" events have bits set to one, this is not always
the case (e.g. an event identified as being the initiator of a bunch of retriggers
will have the FIRST EVENT bit set to 1, but this does not mean that the event is
necessarily "bad", similarly with the MUON FOLLOWER bits).
We now use bits in both DAMN bank 0 and DAMN bamk 1. As a reminder damn bank 1 is referenced in DMM as bank -1.
Currently the following bits are defined:-
DAMN Bank 0
===========
BIT NAME
=== ====
0 RETRIGGER
1 FIRST EVENT
2 BURST
3 CHARGE
4 CLUSTER
5 QVT
6 Q/NHIT
7 CRATE ISOTROPY
8 AMB
9 FTS
10 OWL
11 JUNK
12 RING OF FIRE
13 NECK
14 ESUM
15 HIB
16 QCLUSTER
17 MUON
18 MUON FOLLOWER SHORT
19 MUON FOLLOWER LONG
20 IN TIME CHANNEL
21 FLASHER GEOMETRY
22 NHIT BURST
23 OWL TRIGGER
24 SUBRUN BOUNDRY
25 ZERO ZERO
26 MUON FOLLOWER BLINDNESS
27 MISSED MUON FOLLOWER SHORT
28 MISSED MUON FOLLOWER LONG
29 WILKERSON-ELLIOT BLINDNESS BIT
30 SHAPER BURST
31 MUX BURST
DAMN Bank 1
===========
BIT NAME
=== ====
0 UPWARD MUON BLINDNESS BIT
1 REVERSE FORK
2 FORK
3 NCD OSCILLATORY NOISE
4 FLAT TRACE
5 NARROW PULSE
6 RUN BOUNDARY
7 NCD PULSER
8 SHAPER OVERFLOW
9 MUON NHIT BURST
10 NCD MUX-SHAPER CORRELATION
11 NCD CORRELATION TIME
12 MULTI NCD
13 MUON HIT ALL CRATES
14 THIRD REFLECTION
15 MULTIPLE LARGE PEAK
16 NCD POSITIVE SIGNAL
17 NCD FREQ-DOMAIN FLATNESS CUT
18 NCD FREQ-DOMAIN FORK CUT
19 NCD SPIKE AREA CUT
20 NCD FREQ-DOMAIN SYMMETRY CUT
21 NCD FREQ-DOMAIN OSCILLATION CUT
22 NCD NRE PULSE TAG CUT
23 NCD GENERAL RECORD CUT
Any event that is within 5 microseconds of a previous event is
marked by this tag. The first event in a set of retriggers is NOT
marked, but is designated by the "First Event" tag.
This tag is used for the first event of a retrigger burst (see tag
0).
If 3 or more events occur within 1 milliseconds they they are all
marked by the burst tag. Events that have already been marked by
the "Retrigger" tag do not count towards the burst but will be
tagged if they occur in the dead time of the burst.
This tags an event if any calibrated channel has any pedestal
subtracted charge less than -80 or if any channel that doesn't have
a calibration has any charge that falls outside the range 300 to
4000.
This tags events that have more than 6 hits in a paddle card or
more than 200 hits in a crate.
This tags events where:
( The highest charge tube in QHL is more than 1000 pedestal
subtracted counts above the events average QHL charge
OR
The highest charge tube in QLL is more than 80 pedestal subtracted
counts above the events average QLL charge )
AND
The high charge tube is between 60 and 250 ns before the median
tube time in the event.
Using pedestal subtracted charge and a hardwired gain of 32.3 first
throw away the 10% of the tubes that have the largest charge and
then divide this by 0.9* NHIT. Tag the event if this ratio is below
0.25.
This tags events where more than 70% of the hits occur on one crate
and more than 80% of those hits occur on two adjacent cards,
including a wrap around to count cards 0 and 15 together.
This cut using the Analog Measurement Board information to place
cuts on events based on the value of the Integral/NHIT distribution
of their ESUMHI pulse AND/OR the Peak/NHIT distribution of the
ESUMHI pulse. Events are cut if they lie more than 3.7 sigma away
from the mean in either the integeral or peak cuts.
This tags events where the median time difference of hit PMT pairs
within 3 m of each other is greater than 6.8 ns. Pairs with large
time differences (> 25 ns) are not included in the median and at
least 15 pairs satisfying the distance and time difference criteria
are required before the event can be tagged.
This tags events where the number of OWLs + BUTTs is 3 or more.
This will remove orphans, ECA data and events containing the same
PMT more than once.
Events are tagged where more than 70% of the hits occur on a crate
and more than 80% of those occur in a ring around the edge of the
crate defined by cards 0-3 or 12-15, or channels 0-7 or 24-31.
This cuts events containing neck tubes. It requires that either
both tubes in the neck fire, or that one of those tubes fires and
it has a high charge and is early. High charge is defined by a neck
tube having a pedestal subtracted charge greater than 70 or less
than -110. Early if defined by the neck tube having an ECA time
70ns or more before the average ECA time of the PSUP PMTS with z
les than 0. After the cable changes to the neck tubes this time
difference changes to 15ns.
Tags an event if the ESUMLO AND ESUMHI trigger bits are set, AND
NONE of the following trigger bits are set: (NHIT100LO, NHIT100MED,
NHIT100HI,NHIT20, NHIT20LB, OWLN, OWLELO, OWLEHI, PULSE_GT,
PRESCALE, MISS).
This cut will tag events occuring in a time window 0.1s before and
0.6s after a HIB trigger. (Once this system is up and running)
This is a merged version of the charge and cluster cuts. It
requires that any bad charge, must occur with at least 3 other
tubes in a window of 5 tubes. A more aggressive charge cut is used
requiring that a tube be 250 counts below pedestal (400 for QHL) or
2000 counts above (300 for QLX).
This tags muons by requiring that the event have 5 or more owls and
occurs more than 5us since the last owl Event. It also requires
that the event have 150 hits or more and that the time rms is less
than 90ns. Finnaly it requires that the event does not fail the
neck tube part of the neck cut.
This tags events all events in time window of 20s following a muon.
This tags events all events in time window of 60s following a muon.
This tags events having fewer than 60% of the tubes within 93 ns
coincidence ("in-time"). The number of in-time hits is the maximum
number in any position of a 93 ns sliding window applied to the PMT
time spectrum of the event.
Clusters are looked for in channel (4 hits in eight channels) and
real (4 hits within 1m) space. If the separation between the
average position of the cluster and the rest of the tubes is
greater than 12m the event is tagged.
This looks for 6 events above 40 hits within 4 seconds. Like the
burst tag all events within the imposed dead time are cut, but
events identified by the retrigger tag do not count towards the 6
events.
This tags events where the OWLESUMHI trigger fires.
This tags all events that occur within 60s of the start of end of
every subrun. This is needed as any tag that uses the RAA can
potentially miss events at the start and end of subrun. The length
of the time window is set to be the same length as the longest used
by another cut, currently muon follower long.
This tags all zero-zero events. This are events which have GTIDs
that have 0 set for their last eight bits.
For the salt analysis the collaboration has decided to implement a
blindness scheme using muon followers. The muon follower cut for
salt data has been changed to allow events in a certain time window
after the muon to leak through the cuts. These events are tagged by
the SALT BLINDNESS bit instead. Anyone carrying out any neutrino
analysis with salt data must not test this bit until the decision
to open the box has been made by the collaboration. This bit is now
open for salt data and can be tested.
This scheme is also being used in the NCD phase in the same way as
salt. Until the box is opened this bit should not be examined for
any NCD data without the express permission of the ACC/ACC++.
This cut is desinged to remove neutron followers from missed muons,
atmospheric neutrinos and other processes that create neutrons with
visable precursors. The cut removes any event that occurs within
250ms of a NHIT>= 60 hit event. In teh salt and NCD phase a cut of
150 hits on the precusor is used.As this cut breaks muon follower
blindness this bit must not be tested until the collaboration
decision to open the box has been made. For salt this box is now
open, but the NCD box is still closed.
This cut is the companion to the missed muon follower short cut.
It removes any event that occurs within 500ms of a trigger event
event. As this cut breaks muon follower blindness this bit must not
be tested until the collaboration decision to open the box has been
made.
This bit is used to implement the Wilkerson-Elliot blindness
scheme. The bit is NOT SET for events that CAN be analysed until
the box is open. Blindness is compromised if a test on this bit is
not implemented somewhere in your analysis chain. The scheme works
by setting this bit for a randomly chosen fraction of events. The
exact fraction of event is unknown but lies between 70 and 90%,
leaving most studies with almost full statistics, but leaving any
flux results blind. This box is now open for salt, but is still
applied for NCDs, with a differnet unknown fraction with the same
limits.
This tag is the first of two NCD burst cuts. This works solely on
shaper events. If there are 4 or more shaper events, with a shaper
charge greater than 1 count, inside 100ms, the burst is removed.
This tag is the second of two NCD burst cuts. This works solely on
MUX events. If there are 4 or more MUX events inside 100ms, the
burst is removed.
This bit is used to implement the blindness scheme that has been
proposed by the muon group. As for the WE tag the event is set for
events that should not be analysed. You cut on this bit to build
your data set. The scheme applies a zenith angle dependent
acceptance based on the zenith angle measured by FTR. Above
cos(theta)=0.6 the bit is set 1228/2048 of the time. Below
cos(theta)=0.6 the acceptance is an unknown continuous
function. The box is currently closed for all analyses.
This cut checks the logged scope trace of an event. If there are
two peaks within 1000 ns and the size of both are larger than 5
bins and the size of the first peak is smaller than that the secon
dpeak, this event is tagged as a reverse fork event.
The cut checks the logged scope trace of an event. If there are two
peaks within 1000 ns and the size of both are larger than 5 bins
and the size of the first peak is larger than that of the second,
this event is tagged as a fork event.
The cut checks the logged scope trace of an event. The baseline and
the spread of the trace are determined by the last 2000 bins. If
the spread is larger than 5, it is reset to 5. If it is smaller
than 3, it is reset to 3. If the scope trace oscillates beyond the
average minum the spread and the average plus 3 times the spread by
more than 5 times, this events is tagged as an oscillatory noise
event.
The cut checks the logged scope trace of an event. If the
difference between the minimum value and the average of the last
2000 bins is smaller than 3, this event is tagged as a flattrace
event.
The cut checks the logged scope trace of an event. If the FWHM
(full width half maximum) of the signal peak is smaller than 250
ns, the event is tagged as a narrow pulse event.
This tag events at the start and end of a run for which it is not
clear that both PMT and NCD systems are running. The start of the
run is defined by a point in time 1.1s after the first PGT or NCD
event, which ever is later. The same is done at the end of the run
except in reverse. One exception exists for the end of the run, is
this point of time is before the end of the last subrun, the start
if the last subrun is used as the defined point.
Events from the NCD random pulser appear on a separate NCD
channel. This DAMN bit is used to identify these events and prevent
accidental analysis of them.
The shaper overflow cut search for shaper events with ADC value
2047. Such events are shaper overflow events. The events following
a shaper overflow events within 15 us to 5 ms is tagged by the
shaper overflow cut.
The Muon NHit Burst tag is set if four events over 250NHit occur
within two seconds. Like other burst tags all events within the
imposed dead time are cut, but events identified by the retrigger
tag do not count towards the four events.
The NCD correlation cut searches for pairs of shaper and mux events
which are from the same string and the time interval between then
are within a window (5-15 micro seconds by default and is
adjustable in raa.dat). One status bit will be set in RAAR bank for
both events. NECL banks will be lifted under EV banks of both
events to store the correlated shaper charge and the time interval
between two events.
The NCD correlation time cut checks the time interval between a
pair of correlated shaper and scope events. If the time interval is
smaller than a threshold (default at 8.3) micro seconds and is
adjustable in filter_flth.dat), the events will fail this cut. This
cut is potentially useful to remove some "Seattle" events.
The multi NCD cut checks how many NCD events (shaper and mux) are
in one SNOMAN event. If there are more than one NCD events, the
SNOMAN event will fail this cut. It is expected that very few event
should fail this cut.
This tag is set when there is not at least one hit in every crate
in the detector. It will be used by the muon group to reject
events that occur when a crate has tripped off.
This tag is set when there is so-called third reflection on the tail of
NCD scope trace. This is a feature of fork events.
This tag is set when there are 4 or more large peaks in NCD scope trace.
This tag is set when the amplitude of positive peak on NCD trace is larger
than 20.
The flatness cut removes events with very little power in the
lowest non-zero frequency bin of the waveform power spectrum. This
is a good indicator of the presence of a significant pulse in the waveform.
The fork cut removes events with a peak in the power spectrum
around 12MHz. This corresponds to the frequency associated with
fork events.
The spike cut removes events which have a larger deviation in the
maximum height of a waveform compared with it's area. This is a
good indication of a large spike in a waveform, or other oddly
shaped waveforms.
The sym cut removes events which have a non characteristic zero
frequency intercept of the phase of the fourier transform of the
waveform. This is a good indicator that the waveform has a unusual
symmetry.
The oscillation cut removes events with large power at low frequency.
This is a good indicator of oscillations but also removes spikes.
The NRE bit flags the NRE calibration waveforms, which are used to
look for RCD conditions.
The NCD General Record (NGR) are the events with both shaper and mux
trigger. The NGR cut tags those events and all the events 30ms before
and 5 ms after those NGR events. The code of this cut also calculated
the dead time due to this cut.
Go Back to the
Snoman Companion Top Page