Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
The Laser Workshop
AdiAdi BornheimBornheimCalifornia Institute of TechnologyCalifornia Institute of Technology
CERN, September 16, 2005CERN, September 16, 2005
Laser Long Term Performance& Pulse Width Issues
September 16, 2005 A.Bornheim - Laser Workshop - CERN 2
Outline
Part 1 : Laser long term performance 2003 - 2005
Interlude : Convoluting Pulses
Part 2 : APD/PN pulse width dependence.
September 16, 2005 A.Bornheim - Laser Workshop - CERN 3
Laser Long Term Performance
What is long term ?‘Natural scales’ :
LHC cycle : 24 h 1 dayAccelerator ‘maintenance cycle’ : 1 week (SPS), @ LHC ? 7 days
YLF Lamp aging : 500 h to 1000 h 20 – 40 days‘Various’ other laser parts (eg. flowtube) : 1 year 365 days
Accumulation of luminosity for In-situ calibration : few months ~60 days more are startup >>60 days
Over which time scale do we have to achieve the stability requirements ?
September 16, 2005 A.Bornheim - Laser Workshop - CERN 4
Laser Operation 2003 – Beam Periods
Time [hours]
Int_
rel
‘normal’ & stable running heavy ion with SM0 h.i. with SM1
air conditioning problem
laser internal monitoring problem
440nm500 nm
800 nm700 nm
2003 was the only time we ran the laser on green (and red). The benefit is unclear.
September 16, 2005 A.Bornheim - Laser Workshop - CERN 5
Time [hours]
Int_
rel
Laser Operation 2003 - Setup Period
20.08. 28.08. power tuning for ECAL
September 16, 2005 A.Bornheim - Laser Workshop - CERN 6
Int_
rel
Time [hours]blue monitor broken
air conditioning problem
Beam on ECAL Beam on ECAL
Laser Operation 2003 – Beam Periods
September 16, 2005 A.Bornheim - Laser Workshop - CERN 7
Time [hours]
Int_
rel SM cooling problem SM LV problem
laser internal trigger
blue laser cooling ?
Laser Operation 2003 – Beam Periods
September 16, 2005 A.Bornheim - Laser Workshop - CERN 8
YLF lamp aging - 2003In
t_re
l
Time [hours]Time [hours]In
t_re
l
440 nm 800 nm
Pump lamp for pump laser degrades over time→ Pulse energy degradation for constant pump current.
For blue laser (runs at higher pump current, here 25 A) :Mean degradation : 0.41%/day - 12.4%/month
For red laser (runs at 20 A):Mean degradation : 0.057%/day - 1.72%/month
This can be compensated by increasing the pump current and replacing the pump lamp.The pump current adjustment can be automated with a feedback system.
It was decided to not touch the lasers during the period shown above to ensure stable running.
September 16, 2005 A.Bornheim - Laser Workshop - CERN 9
Hardware Trouble – Spring 2004
0.36
0.38
0.4
0.42
0.44
0.46
0.48
0.5
600 650 700 750 800 850
. Time [hours]
Int_
rel
-1.28 %/day
-1.97 %/day
-1.67 %/day
Was later traced to damaged optical components. Laser 2 had a hardware problem at the same time - which was later traced to a broken flowtube.
At the time it was decided not to intervene but to continue data taking !
Laser 1
440nm
September 16, 2005 A.Bornheim - Laser Workshop - CERN 10
Comparison Degradation 2003 / 2004
Note : Intensity Scale Arbitrary !!!Aligned to have common ‘0’.(However : Scale agrees with power settings)
2004 : ~ -1.6%/day (w.r.t. 0.5)2003 : ~ -0.4%/day (w.r.t. 0.28)0.36
0.38
0.4
0.42
0.44
0.46
0.48
0.5
600 650 700 750 800 850
.
Time [hours]
Puls
e En
ergy
2004
2003
440nm
September 16, 2005 A.Bornheim - Laser Workshop - CERN 11
Pulse width history - Spring 2004
Time [hours]
Puls
e W
idth
[ns]
Hardware problems can cause non-standard behavior !Pulse width seems to be not affected as much as pulse energy. Anti-correlation not as strong as normally.
Time [hours]
0.36
0.38
0.4
0.42
0.44
0.46
0.48
0.5
600 650 700 750 800 850
.
Puls
e En
ergy 440nm
September 16, 2005 A.Bornheim - Laser Workshop - CERN 12
Laser Operation SM10 – Fall 2004
750 hours
!!
?
Fast Monitor Data
Cooling Problem
Electronics/Cooling ProblemWidth
Height
Overall performance good. Three ‘hick ups’ during 750 hours operation – one of unknown source.Two ‘non standard’ – height and width change in the same direction – incidents.
Pulse Width
Fast monitor (event by event) averaged over one run.
Fast Monitor Data
440nm
September 16, 2005 A.Bornheim - Laser Workshop - CERN 13
Laser operation (440 nm) 2005
23
24
25
26
27
28
29
30
31
32
1000 1200 1400 1600 1800 2000Time [ h ]
Mea
n T
iS W
idth
[ n
s ]
5.8.
2005
1.8.
2005
15.8
.200
5
20.8
.200
5
25.8
.200
5
30.8
.200
5
1.9.
2005
5.9.
2005
10.9
.200
5
15.9
.200
5
1
1.2
1.4
1.6
1.8
2
2.2
2.4
1000 1200 1400 1600 1800 2000Time [ h ]
Mea
n T
iS E
ner
gy
20.8
.200
5
25.8
.200
5
30.8
.200
5
1.9.
2005
5.9.
2005
10.9
.200
5
15.9
.200
5
15.8
.200
5
5.8.
2005
1.8.
2005
Laser 1Laser 2
IPump=23 AIPump=22 A
To be adjusted in LaserSupervisor
Operation so far good.Continue to optimize retuning procedure to match pulse width and height.
~0.5%/day
September 16, 2005 A.Bornheim - Laser Workshop - CERN 14
Pulse Timing Drift - 2005
Pulse Time is anti-correlated to the pulse energy and correlated to the pulse width !The return signal from the laser DAQ guarantees proper timing of the ECAL readout.The timing of the TiS pulse drifts over several LHC clock cycles on the time scale ofweeks.
50
100
150
200
250
300
350
1000 1200 1400 1600 1800 2000
Elapsed Time [ h ]
TIS
Pul
se T
ime
[ ns
]
5.8.
2005
1.8.
2005
10.8
.200
5
15.8
.200
5
20.8
.200
5
25.8
.200
5
30.8
.200
51.
9.20
05
5.9.
2005
10.9
.200
5
15.9
.200
5
Compare previous Page :There can be shifts in pulse timing independent of pulse energy/width
Pulse width scan
September 16, 2005 A.Bornheim - Laser Workshop - CERN 15
Laser Source Performance
1.8%1.4%1.1%
1.3% 1.3%
3.5%3.7%
1.7%
Stability over periods of 5 days :
The lasers were tuned to optimize the pulse width stability – seems to work.2003 : 440 nm 800 nm
Pulse Height : 2.6% / 25 hours 3.2% / 25 hours1.5%/ 30 min 2.8% / 30 min0.4%/day (long term) 0.06%/day (long term)
Pulse Width : 2.7% / 25 hours 2.6% /25 hours
September 16, 2005 A.Bornheim - Laser Workshop - CERN 16
Short Term Stability - 2003
440 nm
800 nm
495 nm
700 nm8.2 %3.2 %
7.6 %2.6 %tref : 330 - 335 h
440 nm
800 nm
495 nm
700 nm
pulse width [ns]
5.7 %2.6 %
6.8 %2.7 %
tref : 330 - 335 h
25.7 ns
23.9 ns 36.3 ns
65.8 ns
440 nm 800 nm
pulse time [ns]
2.8 ns 1.5 ns
pulse height
Note : The higher the pump current the better the short term stability. In 2003 we used a higher pump current than in 2004 & 2005.
September 16, 2005 A.Bornheim - Laser Workshop - CERN 17
Convoluting Pulse ShapesThe sampled pulse is a convolution of the electronic shape and the laser pulse shape. We then estimate the energy from the pulse height.
0
0.2
0.4
0.6
0.8
1
0 25 50 75 100 125 150 175 200 225 250
Time [ ns ]
Rel
ativ
e P
uls
e H
eig
ht
20 ns25 ns
30 ns35 ns40 ns
Shape Function ))(exp())(
()( maxmax
peakpeak
peak
TTt
TTTt
tf −×−
−−= αα
Laser shape (Gauss with FWHM 20 ns, .. , 40 ns)convoluted with shape function.
Parameters for shape function adjusted to matchvalues measured as from Renauld’s talk 16.03.2005.
0
0.05
0.1
0.15
0.2
0.25
0.3
300 400 500 600 700 800 900 1000
Time [ ns ]
Pu
lse
hei
gh
t
APD
PN
0.28321
0.28322
0.28323
0.28324
0.28325
0.28326
0.28327
0.28328
780 785 790 795 800 805 810 815 820
Time [ ns ]
Pu
lse
hei
gh
tEffect strongly depends on the rise time !Sizable for APD - Negligible for PN.
50 ns
September 16, 2005 A.Bornheim - Laser Workshop - CERN 18
Pulse Shape Convolution (con’t)
0.8
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
20 22.5 25 27.5 30 32.5 35 37.5 40
Pulsewdith [ ns ]
AP
D/P
N
APD/PN Ratio from previous page
Effect (slope) is of the order of 0.006/ns.But this number is strongly (within a factor 2) rise time dependent.
September 16, 2005 A.Bornheim - Laser Workshop - CERN 19
Pulse Width Dependence in Data
∆R = 0.085
∆T = 4 ns
Normalize R to 1 ⇒ Slope : ~ 0.003/ns
Plot from R. Salerno
Marc, WACH2002, Paris, 2002 Data
Pulse width ~40 ns ??
Slope : ~ 0.002/ns But : Pulse width determination different.
Note : Here very different setups are compared (electronics, width measurement, etc.).
September 16, 2005 A.Bornheim - Laser Workshop - CERN 20
Pulse Width Scan – Spring 2004Compare : Adi’s talk on 06.10.2004
IPump : 25 A
IPump : 19 A
Nominal IPump for last SC beam period: 22 A
TiS Pulse Energy vs Pulse Width
Attempt to compensate pulse energy change with the intensity regulator ‘online’ to isolate pulse width effect. Since pulse height non-linearity is much smaller this is not necessary.
APD/PN pulse width dependence
0.006/ns
From MATACQ
From fast monitor
From fast monitor
September 16, 2005 A.Bornheim - Laser Workshop - CERN 21
Laser Pulse Width Correction - 2004Please recall Patrice’s presentation at test beam meeting on 18.05.2005 :
Period 1 Period 2 Period 3
Period 1 : Stable runningPeriod 2 : H4 DAQ trouble (timing ?) Period 3 : Running After period 3 : Temperature step, HV scan, laser scans, token ring broken …
Note : The laser problems in this period have nothing to do with the H4 DAQ trouble.
September 16, 2005 A.Bornheim - Laser Workshop - CERN 22
Pulse Width Correction on SM10 - 2004
Uncorrected
Corrected
ECAL APD/PN, Single Channel Monitoring History
Laser Pulse Height
Laser Pulse Width
450 Hours
Fast Monitor Data
Data analysed : Part of Period 1 (not all the data was re-reprocessed to fix PN data) and Period 3. Period 2 is problematic - and thus not used.
Pulse width correction :APD/PN_cor = APD/PN+c·PW_Laser
September 16, 2005 A.Bornheim - Laser Workshop - CERN 23
Pulse Width Correction Performance
Uncorrected
Corrected
Monitoring stability over 450 hours : ~0.1 %
Judge performance (monitoring stability) by projecting values onto Y-Axis for each channel (actually 400 for the plot shown) and determine the RMS.
Period 1 and Period 3 combined.
September 16, 2005 A.Bornheim - Laser Workshop - CERN 24
Summary & Outlook
The long term laser performance and our understanding of its limitations improved over the three years the system is in operation.Pulse energy and pulse width can be kept stable within the requirements for several weeks. We continue to improve the operation of the laser to minimize the impact of maintenance operation and hardware failures.The pulse width dependence of the APD/PN ratio remains a critical issue. It appears that a width stability on the level of <1ns is needed.
Analyse SM5 pulse width scan data as soon as rrf-files are available.Perform further scans on SM5 and on further SM as they become available. Presumably the effect is not channel-to-channel depedent.