|
What to look for in the EMC OEP plots.
|
This page is intended for emc shifters. A more detailed picture is given here
Postscript files containing the EMC OEP plots are produced automatically
from the hbook file generated by FastMonitoring in IR2. The directory
containing the most recent OEP plots is here.
For each page, reference plots are shown for "normal" conditions, and for
some known problems. Generally, green things are good things and red things are bad. Red things are things that the EMC shifter should do put in their report.
PAGE 1 - summary
The first page is just a summary page.
PAGE 2 - digi timing
The top plot contains the hit information for all digis (hits in crystals). Things to notice are:
A peak centered on &asymp 6300 ns.
The normal width of the peak.
The normal continuum of low energy digis (background (bg) noise) upto LogE &asymp -2.
Some bad things to look for are;
A noisy crystal that produces a line like this.
High bgs that produce blobs next to the peak like this.
The second Digi Timing plot only shows high energy digis and is fitted (badly!) with a Gaussian. Just check that the peak is at &asymp 6300 ns.
PAGE 3 - digi timing per trigger line
The Digi Timing for Trigger Lines Plots show the digi timing again, but each of the 4 graphs corresponds to a different kind of physics event, and so to a different triggering system. The top left plot should have a lower peak than the other three.
PAGES 4 and 23 - digi multiplicity vs event number
The plots on page
4 (no cuts) should look something like
the reference above.
The plots on page 23 have cuts so the digi multiplicity is accordingly lower;
it should be around 30
digis/event for the 10MeV cut
and about 3/event for the 100MeV cut. As long as the lines are mostly smooth and mostly flat all is well. The occasional spike is normal, especially at the end of the run.
PAGE 5 - noisy channels
Page 5 shows the noisy channel map. The x-axis is the theta index
and the y-axis is the phi index of the crystals. We normally expect this
plot to be empty.
If there is a noisy crystal, or a group of them,
you should try and note the theta and phi indices of the noisy crystals,
and send an email to
babaremc
so that these channels can be masked out (and note in your shift report).
PAGE 6 - dead channels
Page 6 is also a map of all the crystals in the calorimeter.
Here, any crystal that has at least 1 hit during the run is coloured
blue, so any white holes are dead (or masked out) crystals.
Different white dots mean different things (to put in the report):
One dot is a crystal.
Four crystals next to each other in
theta represent a "CARE" chip.
Three CARE chips next to each other in phi represent an "ADB" (12 channels).
Two ADBs next to each other in theta represent a "fibre" (i.e. 24 channels).
Three fibres next to each other in theta represent an "IOB" (72 channels).
If you see a NEW dead region,
send an email to
babaremc and note it in your
shift report.
In the plot above, there is one CARE chip masked out at theta=53-56, phi=21,
and four individual dead channels, giving a total of 8 dead channels.
PAGE 7 - Hitmaps
Page 7 shows four hitmaps, each with &theta (x-axis) against &phi (y-axis). The difference between these plots and the map on page 6 is
that here the colour of each
bin represents how often that crystal was hit during the run.
The energies next to each plot refer to the applied cut. Ignore SLIGHT changes in the top left plot, the top right is what DQM looks at.
Examples of things you'd note here are:
The "no cuts" plot on the top left has two noisy crystals around
theta=25, phi=39. These are common and aren't noted.
The "10MeV" hitmap on the top right shows 3 slightly warm crystals in the
middle of the barrel, worth noting.
A warm care chip at
theta=53-56, phi=21 is worth noting.
The "100MeV" hitmap on the bottom left has several quiet crystals, not usually noted.
The "1GeV" hitmap on the bottom right also has several crystals with zero
hits at this energy, worth noting.
There is a warm CARE chip visible at theta=36-40,
phi=110, agian worth noting.
PAGE 8 - Projections of hitmaps
Page 8 shows the projections of the hitmaps on the previous page in theta (top)
and phi (bottom). Higher peaks and/or lower troughs than sual are worth noting, see here on the left hand side.
PAGE 9 - Digi multiplicities
Page 9 shows the distributions of number of digis per event, with different
cuts on the digi energy. The top left plot has no energy cut, and should
look something like the reference plot. The mean of the distribution is
usually in the range 830-860. Sometimes bad background conditions can show up as a secondary peak like here . Or in some cases, there are
even a lot of events containing more than 6000 digis like here . So, high backgrounds are worth noting.
For the top right plot, the mean of the distribution should be around 30
and there should not be many entries going above 120 digis or so.
If there are high backgrounds,
the tail of the distribution may be much longer, and the mean higher, like
this .
PAGE 10 - Digi energy
Page 10 is a histogram showing the energy (or log(E)) of all digis, and
should look something like the reference above.
PAGE 11 - Energy profiles in &theta and &phi
The top plot on page shows the average energy of a digi as a function of
theta.
The bottom plot shows average digi energy as a function of phi.
There should be peaks at phi values around 30 and 90, which correspond to the vertical plane.
PAGE 12 - Event and Emc TC Damage
The bottom plot on page 12 shows the fraction of events containing different
types of dataflow damage. It is not unusual for this plot to look something
like the reference above. The most common types of damage are in bins
12 and 13 - OutOfOrder and OutOfSynch. If these occur at a frequency greater
than 0.5 percent, it is worth sending an email to
babaremc.
PAGE 13 - Emc TC Damage vs ROM
This is the same as 12 really. It shows where the fault lies. Here the faults lie on 03 and 06; normally written 0-3 and 0-6. BUT because of a number inf convention it's actually 0-4 and 0-7 (i.e. add 1 to the last number!). Meanwhile, in
this plot, there was TC damage (OutOfOrder, OutOfSynch,
FLINK TimeOuts) from all ROMs in crate 3.
If the amount of damage (as seen on page 13) is &ge 10-3 then note it in the report. Lower, but consistent damage (i.e. over many runs) is also worth reporting.
PAGE 14 - Emc TC Damage by Fiber
Page 14 is the same, except here you note whether the damage is on all the 'fibres' A, B and C.
PAGES 15-17 - Channel Maps
The EMC shifter doesn't need to worry about these.
PAGES 18-22 - Energy profiles
The next 5 pages show the average energy of digis in each crystal vs the
channel number of that crystal. You should expect a level saw-toothed pattern. If a particular bit of hardware is
behaving strangely, this might show up as a spike in one
of these distributions, like here in crate 7.
This is worth noting in the shift report.
Directory Location: BFROOT/www/Detector/Calorimeter/Operations/EMC_Shifter.html
Page contact:
Babar Emc,
|
Last update: Graham Jackson, 02 Jul 2007
|
|
