Transcript C-side: status and plans for current week

ECAL status and plans
by Yu. Gilitskiy, T. Kvaratskheliya, I. Machikhiliyan
and
special thanks to P. Shatalov for very useful laboratory studies of
electronics part
 Current status and problems
 Recent activities:
 C-W noise studies summary;
 Lab measurements of LED Monitoring system
performance;
 Future plans
Current status-I
 optimization of voltage delivery/control
boxes – finished;
 integrators for PM current measurement:
installed, adjusted and connected to
dedicated control box;
 PIN-diodes system: fine gain tuning to
account for large PM gains spread over
ECAL is finished; results of this procedure
are to be checked;
Current status-II: access
summary
malfunctioning PM channels:
 few power lines are fixed;
 C-side: about 30 connectors on signal cables
are repaired (FEB side); More problems are
expected. Typical pattern: small / unstable PM
signal;
 wall side ~ 40 PM/C-W doublets are replaced
 A-side: residual problems in cabling (swapped
connections, wrong mapping, bad contact, etc)
are fixed;
 List of few problematic FEB channels was
sent to Jacques;
Current status-III: access
summary (cont)
Optics of the
monitoring system:
 3 optical bundles
are found to be
damaged
irreparably: LED
groups A3.08.4;
A3.11.4 and A1.05.1
(partially)
 Besides: 5 (C) + 8
(A) cells with absent
/ week LED signal
due to damaged
single optical fibers
Short-circuited power lines
 After voltage delivery / control boxes modification: got 7
cases of short-circuits in low voltage power lines (2 in
March, 5 in April)
All cases show the same pattern:
 short-circuit is in +6V chains;
 short-circuit is due to not recovered resettable fuse (new type is
used in modified version) in the voltage delivery / control box;
 short-circuit happens after (? incorrectly preformed ?) low voltage
power supply (MARATON) shut-down (e.g. power cut);
Two versions of what is happening:
 small fraction of fuses is defective. As soon as all of them will
malfunction and replaced, the problem disappears;
 abnormally high currents sometimes appear at the moments of
ECAL LV power sources shut-down for reasons unknown. These
reasons must be understood;
Short-circuited power lines-II
Series of laboratory tests:
 fuse always recovers while it experiences one
single trip (several current and voltage settings
are checked);
 latest news: fuse can burned if:
 after the first trip it trips again several times with
time interval in-between less that ~30 seconds. Than
it goes to an intermediate state.
 If the fuse stays in this state with the voltage
supplied for a certain time, it becomes damaged;
NOT TO SWITCH ON ECAL LV RIGHT AFTER
POWER CUT, BUT WAIT ~30 MINUTES MIGHT
BE A SOLUTION
Short-circuited power lines-III
 Our proposal is:
avoid unnecessary switching ON/OFF MARATON LV;
if there is a need to switch it OFF (e.g. for preshower people) – contact us first,
we will check current ECAL conditions;
each turning ON/OFF of MARATON should be mentioned in CALO
logbook together with the status of ECAL MV Agilent power supply;
 request to ECS group: is that possible to record automatically all moments
when MARATON’s and also Agilent’s channels are turned ON/OFF?
Also useful: to mention in CALO logbook time of all power cuts
 Correct procedure to switch OFF the ECAL:
 Set zero HV codes on ECAL PMs (wait for confirmation from SW);
 Set zero MV on Agilent power supply (wait for confirmation from SW);
 Switch OFF LV (MARATON). All 4 dedicated channels of MARATON power
supply must be turned OFF at once. DO NOT TURN OFF single channel, it can
cause MARATON supply failure;
In case of any emergency shut-down Agilent is turned off first, only then –
MARATON, not contra vise!
C-W noise studies-I
 Pick-up noise appears regularly in time with the frequency
as twice as high as the one of C-W base internal pumping
oscillator f (f=60 kHz). Pulses have complex shape, typical
amplitude less than 3 mV and duration much less than the
period of internal oscillator;
 On the pedestal distribution the effect of the noise
manifests itself as rare tails around the core Gaussian
distribution of “ideal” pedestal;
C-W noise studies-II

Data from all ECAL R/O crates (CAT DAQ + random
trigger + LEDs OFF);
 Two settings for C-W base feeding medium voltage U MV
- 50V and 80 V;
 PM gains range: from 1K to 300K;
 Noise characterizing parameters are chosen to be:
fraction of entries outside the range <P> −2σ ≤ I(ADC) ≤ <P>
+2σ (<P> − 3σ ≤ I(ADC) ≤ <P> + 3σ ), where I(ADC) is ADC
reading, while <P> and σ are position and width of the pedestal
as obtained from Gaussian fit of the core distribution.
ii. noise sweep, estimated according to the minimal and maximal
values reached by ADC readings in the current r/o channel;
i.
C-W noise studies-III
 Minor dependence on U MV
 No significant
dependence on PM gain
in LHCb ECAL operational
range;
No correlations between
channels;
 The clear difference
(factor 1.6 in the tails
sweep) is observed
between Inner cells and
the rest of ECAL channels;
Different gain settings from 1K to 300K
in the same r/o crate #20
Noise sweep (in Et terms) vs r(cm). Black,
red and blue – Inner, Middle and Outer cells
C-W noise studies-IV
 Summary/Conclusions:
 channel-to-channel difference in the noise level can be neglected for all
cells, belonging to the same section (at nominal gain settings);
 pedestal distributions for groups of channels of the same section type can
be summed up. Reason: resulting net distributions (either one per ECAL
section or one per readout crate) comprise much higher statistics in
comparison with the ones for individual channels and therefore more useful
for further studies
 Details can be found in Internal note LHCb-2008-016
 Beyond the note. News from the lab: difference between sections could be
attributed to the interference in grounding lines of the flat power cable. The
size of effect depends on the distance between neighboring C-W bases on
the line, i.e. cell size;
 Further step - MC studies:
 it is proposed to use for Monte Carlo simulations the description of C-W
noise in the form of three net pedestal distributions summed up over channels
of readout crates #13 (Inner), #12 (Middle) and #9 (Outer), which are obtained
under the conditions of nominal PM gains and “pessimistic” value of U MV
equal to 80 V.
 use these distributions to implement C-W noise in Monte Carlo simulations
and to understand its impact on ECAL performance (in progress, by A.
Kozlinskiy)
ECAL monitoring system
performance lab studies
 LED drivers: time delay between arrival of
LED fire signal and LED flash: 39.6±1.8 ns
(on the basis of 55 LED drivers);
 PIN amplifiers (on the basis of 31
amplifier):
 time delay 12.5±0.5 ns
 gain dispersion: r.m.s. < 0.9%
Further plans
 ECAL system now is in more or less final shape.
The majority of channels is operational;
 Next step: study of the whole detector
performance. In that purpose time sharing with
other commissioners is needed:
 near future: collect several samples with significant
LED statistics in order to develop analyzing algorithms
(will be used further for LED based calibration and
monitoring);
 later, periodically: several hours of DAQ in order to
study detector behavior at different PM gain / LED
intensity settings and to understand long-term stabilities
of the PM and PIN responses;
Spare slides
Brief comments on modification
Goal: stabilization and
increase of low voltage on
ECAL C-W bases
Change: new type of
(resettable) fuse in LV chains.
Fuses of old type have large
dispersion in internal
resistance R, which, together
with large value of R itself,
resulted in noticeable voltage
drop on it. As a consequence,
there is an undesirable
decrease of LV on the C-W
bases. The value of the effect
depends also on the load of
the line (i.e. number of C-W
bases served by it).
Fuse data sheet
Laboratory studies: burn tests
1.
Operational conditions check:
at U=8V no trip up to I=0.4A; Internal resistance value R is 1.6 Ohm and
does not change
2.
Trips at “soft” conditions: U=8V, I varies from 0.5 up to 3A:
1.
2.
3.
After the first trip at 0.5A R increases up to 2.3 Ohm
Next trips: fuse recovers to the value R=2.3 Ohm; recovery time is
~10min
Trips at “hard” conditions:
1.
2.
3.
high U=24V, I varies from 3 to 5A. Fuse recovers to the value R=2.3
Ohm; recovery time increases to 60 min;
Data sheet max U=30V. R increase to 3.9 Ohm, but fuse is still
operational
I=10A and U=60V. Fuse recovers to the value R=3.9Ohm; it is still
operational
Even at the abnormal current/voltage conditions, which can not be
achieved in the ECAL system, it was not possible to reproduce
an effect of non-recovery in the lab