Possible features in the calibration plots
Most of the time, calibration plots look like this:
plots on pages 1 and 2 are widely green (apart the sector 5 which has
always been messy for obscure reasons) and no big blue cluster shows up
in page 6. Calibrations like this are successful.
But, seldom, some plots can be very different from the 'reference' ones: does
this mean that the last calibration was bad? Well, not automatically...
Otherwise, things would be too simple!
The different cases considered below are:
- Muted LED: BAD calibration
The sector 10 LED didn't emit any light during this calibration, preventing
the PMTs from this region to be properly calibrated. Therefore, this
calibration is bad although it was loaded in the DB as the online
validation process is turned off or is accepting everything.The LED can
emit no light for two main reasons: either it is broken or it doesn't
receive triggers from the DCC (bad board or bad connection). In this case,
the cable connecting the DCC to the light generator crate was not plugged
- Magnetic field inside the DIRC shield:
In that case, the DIRC was calibrated by mistake by the pilot while the
solenoid was OFF but the bucking coil ON. As the bucking coil is here to
compensate the magnetic field from the solenoid, it was generating some B-field
inside the DIRC magnetic shield which was affecting the PMTs. Hence, T0 was significantly modified for several PMTs which made this calibration bad.
The DIRC can be calibrated properly when both fields
are nominal or at zero. The two other states (one field ON and the other
OFF) would lead to bad results.
- HV not nominal: potentially BAD calibration
Some PMTs show reduced efficiencies in the outermost part of sector 4.
Even more PMTs show a large shift in the T0 calibration.
Such pattern was due to two HV spare boards which were put in HV crate 3 to
replaced two broken boards. After being put in, the nominal HV values were
not reloaded and so the default values were not appropriate for the powered
channels. Yet, as these boards had been in some HV crates in the past, the
nominal V0 voltages were not that far from the expected values (in average,
50% above and 50% below!). So in fact this was not a real problem (fixed after
a few hours): calibrations didn't diverge and the runs taken in the meantime
were flagged 'OK' , both by the DIRC and by the PID group.
Always reset the nominal HV after having changed some boards!
- DFB changed: a priori GOOD calibration
The T0s include electronics-related effects. Hence, after changing a front-end
boards, fluctuations in the T0 plots are expected. That's what we can (hardly
next time a DFB gets changed, replace the current plots by
nicer ones!!!) see on the upper border of sector 5 ('blue' radius). The
DFB 'radial structure' should be visible on the plot: affected PMTs should
form a blue or black 'line'.
After a second calibration, the troubled area becomes invisible. The last
two calibrations have been taken with the same electronics and are hence in
agreement! So, after having changed an electronics board,
take two calibrations in a row and make sure they agree.
- DCC changed: a priori GOOD calibration
For the same reason as explained above, changing a DCC modifies the T0
correction in a whole sector. Again, take two calibrations
in a row and make sure they agree.
- look carefully at calibration outputs, as quickly as possible after it
has been taken.
- Take two successive calibrations to get rid of hardware changes.
- After a significant downperiod, some hardware change or anything which
may have interfered with the DIRC, do not leave IR2 before a calibration
has been taken!
Created: Tue Jun 20 20:03:39 PDT 2006 (Nicolas)
Last modified: Wed Jun 21 15:54:06 PDT 2006