MEG 2009 Run
Run
Coordinator’s
View!
Peter-Raymond Kettle
MEG Review February 2010
1
Review 2009
DC – HV stability
Successfully finished first MEG Physics Run
(Sept. – Dec. 2008)
APD - electronics
However – Major Issues to be Investigated/Rectified:
LXe - calibration
• DC: high-voltage stability (He-diffusion problem)
• Calorimeter: LXe light-yield LY (reduced absolute LY  )
• Calorimeter: PMT gain-drift (reduced gain with >Qbeam)
• TC: fully functioning fibre detector (new APD readout)
Peter-Raymond Kettle
MEG Review February 2010
2
Situation Spring 2009
Post Review February 2009:
Back to “Square One” –
TOTAL Detector DISMANTLED
for Maintenance/Repair /Improvement
Shutdown 2008 / 2009
+
Kottmann Expt. Front-part E5
Scheduled until mid-July!
Calorimeter –
> cooling-power (<LN2)
Modify purification
System – new Getter
Dismantling of all DCs
New anode-prints
+ wires +
extensive
tests “aquarium”
& “He-cabin” +…
New APD front-end
electronics
+ control interface
(HV stability
> )
(reduced noise
+ faster -> trig)
Peter-Raymond Kettle
(< contamination)
MEG Review February 2010
& Liquid Pump (LY)
3
Further planned Implementations 2009
(I) Upgrading from DRS2 to DRS4
• differential i/p
• internal clock & synchronization
• on-board timing calibration
• 3.2 GSPS possible XEC-1.6GHz
DC  0.5 GHz
but >0.7GHz
(2) Change of degrader thickness at BTS focus
• 300m Mylar to 200m Mylar
• to match 6 % air contamination
content in COBRA (previously variable)
& centre stopping distribution in target
One main aspect which ran in parallel during the first
half of the year was the ANALYSIS of 2008 DATA!!!!
Peter-Raymond Kettle
MEG Review February 2010
4
Schedule 2009
• Kottmann et al. beam time until mid-July
• DRS4 installation end-July
• Parasitic Run (detect. set-up, beam tests e.g. Deg., e+)
beg.-Aug. - mid-Sept.
+
• e -test + CEX calibration run mid-Sept. – beg-Oct.
Peter-Raymond Kettle
MEG Review February 2010
• Remainder of year
- Physics data-taking
expectation ~ 2½ months
5
Schedule Modification
“Exciting Results” from
Lamb-shift Expt.
Caused a shift of
MEG start up by
3 weeks!!!
CONSEQUENCE:
In order to enough time
To calibrate detector
&
obtain usable statistics

Test beam time reduced
to absolute minimum
• No degrader optimization
• No positron beam tests
Peter-Raymond Kettle
MEG Review February 2010
6
Organization
Basic Run Layout
Parasitic Run
Beam optimization
in parallel with
Debug, Tune &
Calibrate
Full Run Part II
Full Run Part I
CEX Run + Trigger
Setup + calibrations
Detector monitoring
Pre-physics data check
Physics Data
(MEG + mixed
triggers)
Run Coordinator
2 Shift
Coordinators
 6 weeks tot.
13 Shift Coordinators
 1 weeks/person
Parasitic Run
Total of 55 persons for 592 shifts (Full Run Only)
to allow for flexibility + continuity:
Staggered & Overlapping shift system
Daily Run Meetings (on-site)
Weekly Video Run Meeting (Collaboration-wide)
later, weekly Video Physics Analysis Group Meeting
Web-based Schedule + Shift list + “On-call” List
Peter-Raymond Kettle
MEG Review February 2010
12 Hr Shifts:
1 DAY SHIFT (Beam Group) 10:00 – 22:00
1 NIGHT SHIFT
max.
22:00 – 10:00
1 Shift Leader Nights
Manned by Detector Experts
Full Run
8 Hr Shifts:
1 DAY SHIFT
1 EVENING SHIFT
1 NIGHT SHIFT
07:00 – 15:30
15:00 – 23:30
23:00 – 07:30
1 Shift Leader + 1 Crew Member
7
Practical Additions - Monitoring
MIDAS – DAQ control page:
•
•
•
•
•
web-access to all MSCB sub-masters & nodes
web-access to FE crates status, temp, fans, ON/OFF
web-access to DC HV-control
web-access to Online Analyzer
improved Offline analysis jobs for shift crew
All towards the long-term GOAL of Remote Shifts….
Peter-Raymond Kettle
MEG Review February 2010
8
Detector Synopsis
TCs:
• APD electronics induces too
higher noise level in DCs
 Switched OFF
• APD DAQ control problems
• still Laser temp. control problems
TCs
Beam:
• BTS He cold leak around JTvalve solved temporarily for
run
• Degrader change from
200m to 300m during Run
stopping distribution problem
Calo:
• Liquid & Gaseous
purification success
optimal LY achieved
• further study PM gain
stability
Calo.
DCs
Beam
DCs:
• HV instability problems
solved, fully efficient
• Resolutions?
Peter-Raymond Kettle
MEG Review February 2010
9
Trigger
+ Splitters
Trigger + DAQ Synopsis
Trigger:
Online
Cluster
Megonxx
Full Complement of 29 Triggers with pre-scaling used
E-resolution improved 20% (7.5%FWHM @ 55MeV)
 > thresh.
Z-resolution Zrec- ZTC(Qratio) improved ~23% (5.5cm)
Direction matching  e with fibres still missing !!!
APD(OFF)  XEC(PMT-index) + TC-bar(index,z)
where z from bar charge-ratio
Implementation of -trigger during beam=“ON” <TCALIB
Trigger#0 TC eff. loss ~25% due broken LVDS transmitter
for first-part of the run
lcmeg05
lcmeg04
lcmeg03
lcmeg02
lcmeg01
Offline
Cluster
lcmeg
Peter-Raymond Kettle
FE-electronics: mostly NEW!
DRS4 teething problems: synchronization + jitter
DRS4 +
part DRS3
DAQ:
• LT~84%, DAQ inter-run time reduced by > 4x
• DAQ online monitoring & control - many additions
• Online (backend) 2 TB storage
• Offline (lcmeg) 64 CPUs + now 150 TB disk
• “Lazylogger” autocopy Online  Offline
factor 2 compression offline
MEG Review February 2010
10
2009 Run Conditions
1. Target Angle: - similar slant angle  as 2008
• Conventional
= (20.6 ± 0.2)°
• Photogrammetric
(outside COBRA) = (20.4 ± 0.2)°
• Photogrammetric
(inside COBRA) = (20.3 ± 0.3)°
 - target
inclination
angle
2008 <> = (20.5  0.3)º
2. Degrader: two settings used during Run
200m & 300m Mylar & 94%He/6%Air
Target Inclination 2009
 = (20.0 ± 0.3)° provisional
3. Beam Intensity: - on average 10% > IPROT compared 2008  2 main modes tuned
“Normal” – same slits as 2008 & “Normal*8” – slits diff., same absolute rate as 2008
Mode
R Measured Rate COBRA at 2mA
200m Degrader
R Rate COBRA at 2mA
300m Degrader
“Normal” (slits 115)
4.3107 +s-1 (2009)
3.7107 +s-1 (2008)
“Normal*8” (slits 105)
( physics data)
3.7107 +s-1 (2009)
3.2107 +s-1 interp. (2009)
Stopping rate for physics Run (300m degrader) at 2.2 mA ~ 2.9·107+s-1
Peter-Raymond Kettle
MEG Review February 2010
11
Degrader Question 200m
GEM Monte-Carlo 2008 Shows stopping US of target centre by > 50m CH2
for a target thickness of 205 m 4.3% Air
300m
Degrader
GEM 2008
4.3% Air
Equivalent to 140 m CH2
Along l or ~ 100 Mylar
t=205 m
l=585 m
Remember 100m of degrader
Equivalent to ~
24% of target thickness ONLY!!!
Therefore should just shift peak of
stopping distribution inside of target
Conclusion:
Remove 100m of Mylar i.e. go to 200m thickness to centre stopping distribution
2009 Physics Run started with 200m Degrader since optimization skipped due to time
Measured
200m
Measured
Vertical
Profile
0.5 mm Aare
Peter-Raymond Kettle
Horizontal
Profile
0.8 mm high
Vertex Distribution
Using Michels
MEG Review February 2010
12
GEM vs Data
GEM 2009
6% Air
strong asymmetry in TC US/DS hit-maps seen???
100
200
300
200 m Degrader
Example MEG-trigger Data 2009
Run# 53000
GEM 2009 –
confirms central shift
BUT!!!
Data features not all
reproduced e.g.
• Asymmetries
• stopping with no target
GEM/Data:
Peter-Raymond Kettle
300 m Degrader
Example MEG-trigger Data 2009
Run# 59040
Timing problem TC-US
Enhances Asymmetry
For MEG-data
TIC-asymmetry DS/US ~ 4.3
TIC-asymmetry DS/US ~ 1.3
MC
Trg# 22 data
Trg# 0 data
300 micron Deg.
1.66
1.37
1.4
200 micron Deg.
1.7
3.0
4.6***
MEG Review February 2010
*** US-TC
enhancement
13
Degrader Conclusions
Motivation:
from _ 1 & 3
Originally 2008 degrader of 300 microns changed to
200 microns
in 2009
to optimize
200
COBRA
2009
(200,105) / mC 
 stop6%
R
Michel
stop-distribution to centre of the target for now fixed
Air-admixture
2009
spect  2008
 2008
300
COBRA
(300,115) / mC 
amongst other information, this was supportedby
GEM
RMichelMonte-Carlo
stop 2008
spect  2009
from _ 2 &hit-maps
3
However – strong asymmetry in TC US/DS
seen???
300
COBRA
2009
(
300
,
105
)
/
mC



Changed back to 2008 situation
during
1 R MichelRun
stop 2009
spect  2008
Run
Condition

300


300
stop 2008

0.854

RMichel (300,115) / mC 
2008
200 Physics Data: 29/10 – 23/11
14.0D
 2009 _ 300DAQ-time:
 _ relative _ to _ 200
:
300 Physics Data. 26/11 – 23/12 DAQ-time: 21.2D
Relative Muon Stopping Statistics 2009/2008:
stop 2009
200
stop 2009
COBRA
spect  2009
2009
1 R Michel (300,105) / mC


0.854 R2008
(200,105) / mC
Michel
(i) DAQ-time 2009/2008 = 0.732 Dominant factor!
(ii) Relative time-weighted stopping-rates 2009/2008 = 0.866 (all 300 data = 0.94)
 (2009: stop200~56%, stop300~82%) – relative values from Michel data

Relative Stopping Statistics:
SS2009/SS2008 = 0.634
Dominated by run time!
However true relative statistics 2009 much better since spectrometer tracking efficiency
& trigger efficiency much better than 2008!  factor ~3.5 x 0.634 = ~ Factor 2.2
Peter-Raymond Kettle
MEG Review February 2010
14
Degrader Conclusions cont.
Conclusion: MC does not support “strong leakage” from DS-face for 200m case
detailed check & MC-study underway
Necessary to perform optimization beam test using different degrader thicknesses
Will be performed spring 2010
• measuring stopping-rate by counting Michel e+ using reconstructed tracks
originating from the target volume vs. several degrader thicknesses
• measure “No-target” stop distribution for above degrader thicknesses to compare
with & tune/optimize MC
Peter-Raymond Kettle
MEG Review February 2010
15
Further Aspects of 2009 Run
•
BTS: He transfer-line coupling + Joule-Thompson Fore-valve region, cold-leak
(meant BTS had to be warmed-up & JT-tower opened + additional pump-stand introduced
that could work in B-field environment)
Transfer-line
coupling
OPEN
FV < 7.5%
Fore-valve
CLOSED
He leak Detection
Open: 3·10-7mbls-1
Closed: >7·10-6mbls-1
BTS
isolation vacuum
JT-Tower
FV-OPEN
Problem solved for Run:
• New In-coupling transfer line
• New additional pump-stand
for isolation vacuum
Peter-Raymond Kettle
MEG Review February 2010
16
FE-Electronics
•
FE-Electronics:
(teething problems with DRS4 implementation took much longer than anticipated
involving both software & hardware modifications
also the new architecture 4-chips/2-chips necessitated the total re-doing of the zero-suppression algorithm; physical displacement of
cross-associated anode & cathode channels to different modules)
(i)
(ii)
(iii)
(iv)
•
DRS4 clock synchronization
DRS4 large dispersion in inter –channel, -chip, -board, -crate time-jitter
DRS4 spike & ripple suppression
DC zero-suppression algorithm
Noise problems:
(Fourier power spectrum analysis –showed various source frequencies- high frequency
range associated with APD electronics resulting in TC fibre system not being used)
(i) Environmental noise – seen by DC & TC systems
(ii) New APD Fe-electronics producer/propagator of high noise-levels on DC
readout
Peter-Raymond Kettle
MEG Review February 2010
17
LXe Aspects
Calorimeter:
2008 Light-yield < expected both for s & s (Q/A)/(Q/A)~ 1.25 expect LP ~ 1.92!!!
Contamination? new purifier installed Liquid & Gaseous
& No purification scenarios studied
• Light-yield proved to be stable @ 1% level during 2009 Run
• Achieved absolute LY gammas & alphas as expected
• long-term PMT gain-drift still not understood ~ 6% drop over run
monitored extensively by many tools – can be compensated for
with HV-adjustment if necessary – problem still being studied
• gain-variation with beam-loading well understood –new ideas
Gain Chronology
2009 Run
R9869 Gain Shift vs SKB
1.2
8
SKB
GainShift
1.15
SKB
7
1.1
6
1
5
SKB
Gain Shift
1.05
(Q/A)/(Q/A) ~ 1.92 !!!
0.95
4
0.9
Peter-Raymond Kettle
3
Rel gain-shift
0.85
MEG Review February 2010
1948
1902
1851
1783
1719
1634
Serial
No
Serial No.
1501
1436
1341
1258
1112
943
1032
867
793
714
636
545
475
359
223
2
120
0.8
19
SKB = cathode blue sensitivity
vs. serial no – shows possible link of
gain to too much alkali?
18
Beam Time/Data
In view of the extremely tight schedule for starting from “scratch” with the set up in E5,
also with Lamb-shift extension + numerous detector/electronics etc. problems to be
solved … Managed (in real days) 55% of 2008 physics data-taking
& approximately same number of triggers
MEG DATA
Parasitic Run: 1st September- 15th September ~ 2 weeks
Beam Tests/Tuning (~1 week)
Full Run Part 1: 16th Sept. – 30th October ~6 weeks
CEX 12th October– 24th October (12 days)
Full Run Part 2: 25th October – 22nd December ~8½ weeks
Physics Data 200m Degrader ~17.5 Days
Physics Data 300m Degrader ~25.5 Days
Run 2009
Normal Physics Data-taking:
• MEG 12-mixed trigger
300m (Runs# 57582-64482)
 5775 Runs a 2k events
 11.63 M Triggers
 Time 21.05.25.07
93 TB data
6 Hz Trigger Rate, LT~84% 300m Degrader
10 Hz, 75% LT 200m Degrader
• Daily LED-calibration beam “on-off”
• 3/week Full-calibration LED beam “on” +LED beam “off”
+ C-W (Li) + C-W (B) + s
Peter-Raymond Kettle
200m (Runs# 51824-57184)
 5313 Runs a 2k events
 10.64 M Triggers
 Time 14:00.31.17
MEG Review February 2010
Total of 93 TB Data
Taken 2009
19
Conclusions 2009
• Of the original expected ~11½ weeks of physics data-taking (MEG Schedule) minus
the 3 weeks extension granted to Lamb-shift expt. We managed to take ~ 8 weeks of
MEG data!
•
The major problem encounter during the 2008 Run (DC HV-instability) was solved and the
chambers ran with “full efficiency”
•
The LY of the calorimeter also “plateaued” at its expected value allowing /-discrimination
with beam “on” enabling continuous monitoring during MEG data-taking – LY stable over
extended period of time to ~ 1% level WITHOUT further purification!!!
•
The LXe PMT long-term gain-drift is not understood yet though enough tools available to
monitor & sufficient HV-reserve so not problematic. Beam correlated gain variations well
studied – strategy for UCN beam structure being further studied.
•
Reduction of the timing-jitter from DRS4 channels on different boards is being studied
together with the possibility of running at higher frequency
•
TC fibre-detectors which could not be used in 2009 due to large induced noise on the DC
electronics are being modified & expected to be fully functional for 2010
Peter-Raymond Kettle
MEG Review February 2010
20
Conclusions 2009 cont.
•
Fine-tuning of the degrader with beam and a check of the range-straggling in COBRA
without a target, for MC comparison/tuning, will be undertaken
•
The postponed test of a monochromatic positron beam which would allow the energy
dependent study of our positron spectrometer resolutions as well as the relative
acceptances, will be scheduled
•
Many of the outstanding questions concerning detector performance: DC cathode foil
aging,TC fibre detector noise suppression, DC-survey position, magnetic field symmetry,
chamber resolutions etc. are now being addressed or to be answered soon!
– MORE TIME NEEDED!
This overview was meant as an introduction to the following “Expert” talks
Peter-Raymond Kettle
MEG Review February 2010
21
Finally
A substantial portion of the year was
ALSO utilized to analyze our
2008 Physics Data
In parallel to solving the previous
outstanding problems.
The Preprint:
Is now ready for submission
for publication!
Peter-Raymond Kettle
MEG Review February 2010
22