CERN_Timing_Overview_-_GSI_

CERN Timing Overview
CERN timing overview
and
our future plans with White Rabbit
Jean-Claude BAU – CERN – 22 March 2012
1
Sequencing models
• Goal
• Strong coupling concept
• Loose coupling concept
• Interaction Loose/Strong coupling
Timing distribution
• Messages sent on the timing network
• Local timing
• Client timing libraries
Future of the CERN timing
• Overall view
• First White Rabbit implementation
Overview
Sequencing
Sequencing models
 Main goal
Cycle
Extraction
Extraction
Injection
Injections
time
Cycle 1
Sequencing models
 Strong coupling concepts
Cycle 2
Extraction
Basic Period
Injections
time
Cycle 2
BP 1
BP 2
Cycle length = N * Basic Period
Basic Period length = currently 1200ms
Sequencing models
 Strong coupling concepts
Injections
Beam
Client
Acc
time
Cycle 3
Extractions
Inj.
Acc.
time
Cycle 1
Cycle 2
All cycles are linked together :
All cycles of a beam are always played
Sequencing models
 Strong coupling concepts
time
Client
Acc
Spare beams
Beam A
Inj.
Acc.
Beam B
Beam C
Beam D
Beam E
time
Beam A
Beam A
Beam B
Beam C
Beam D
Beam F
Sequencing models
 Strong coupling concepts
A spare must be on the
shadow of its parent
Beam Coordination
Diagram
A
Client
Acc
Phase
Inj.
Acc.
D
E
B
C
F
H
time
Duration
A
A
B
B
C
D
E
F
G
I
time
Duration
•
•
•
A BCD is executed in a loop
Each accelerator has its own phase
All accelerators in a BCD have the same duration
Sequencing models
 Strong coupling concepts
Sequence
Sequence 3
S
e
q
.
s
e
l
e
c
t
o
r
Sequence 2
Loop waiting condition
Loop waiting condition
Executed 1 time
Executed 1 time
Normal
Operation
Coast
Prepare
Coast
Output
BCD
Sequencing models
 Strong coupling concepts
Coast
Recover
Coupling/Decoupling
Manual
Acc 1
•
Sequence 1
Sequence 2
Acc 2
•
•
Decoupled acc. play different
BCDs & seq.
No beam can be played
Can be recoupled at some key
points
Automatic
Coupling point
Acc 1
Normal
Operation
Decoupling point
Coast
Prepare
Acc 2
Sequencing models
 Strong coupling concepts
Coast
Coast
Recover

Advantages
◦ Manage by one
timing data master
◦ Optimize the usage
of the accelerators
Sequencing models
 Strong coupling concepts

Constraints
◦ Maintenance
◦ Complex
◦ Find a common
basic period of time

When to apply this model ?
◦
◦
◦
◦

Frequent beam transfer among accelerators
Short cycle length
Optimization of the accelerators
Very close accelerator schedule (maintenance)
Use at CERN for LEIR, BOOSTER, CPS ,
SPS
Sequencing models
 Strong coupling concepts
Loose coupling
Unpredictable time
Collisions
Inj.

Filling
Inj.
time
Used when :
◦ The duration of the cycle is unpredictable
◦ The cycling time of the accelerator is long compared to its
injector


Need to be synchronized with injector only at injection
points (RDV)
Need to wait the injector at the RDV point
Sequencing models
 Loose coupling concepts
LHC Injection
LHC injectors Data Master
(Strong coupling)
LHC Data Master
(Loose coupling)
Beam request (Type, Ring, Nb batches, ….)
Unpredictable
time
Forewarning Injection
Predictable
time
Injection
time
Sequencing models
time
 Interaction Loose/Strong coupling accelerators
Timing network
Timing Data Master
Triggers
External triggers
Telegram
Cable id
UTC time
Diagnostics
Distributed timing

Used to trigger
◦
Local counters
◦
Real Time tasks

High priority messages

Describe the played Cycle and the next one
◦
Particle type, beam destination, …

Sent every Basic Periods
Low priority messages

Identification of the timing cable

◦
Auto configuration of the computer

Low priority messages

UTC time for time stamping
Low priority messages



To check the quality of the transmissions
Low priority messages
 Messages sent on the timing network
Time window
UTC millisecond ticks
Msg 1
Msg 2
t0
Msg n
Time
t0+1ms
RT
Task
Msg 1
Messages sent on the timing network
Msg 2
Distributed timing
 Messages sent on the timing network
Timing Receiver card
Msg 1
Msg 2
Msg n
Timing network
Clocks
External starts
Pulses
RT
Task
Distributed timing
 Local timing


Trigger external devices
Chain counters among timing
receivers
Complex timing layout
Distributed timing
 Local timing
Front-end timing libraries
Applications
(FESA, …)
DB
Concept of triggers/fields
Timing abstract layer
Transformation
Timing low level layer
Concept of
triggers/payloads/Telegram
GMT network
GMT specific
WR specific
GMT
Receiver
WR
Receiver
Distributed timing
 Client timing libraries
To be defined
White Rabbit network
Overview



http://www.ohwr.org/attachments/913/wrCernControlAndTiming.v1.1.pdf
Complex to manage redundancy for Timing & Data
WRDM with two ports for the redundancy
Future of the CERN timing
 Overall view
VLANs
Future of the CERN timing
 Overall view
WRDM: Master/Slave




Consist of two synchronized WRDM running
exactly the same thing  Produce the same
messages
Only one at a time sends its messages on the
WR network
The switch between the WRDM should be
transparent
Main goal :
◦ Fast upgrades during a technical stop
◦ Reduce intervention time in case of hardware failure of
the WRDM
Future of the CERN timing
 Overall view
WRDM: Solutions
Future of the CERN timing
 Overall view
AD& ELENA decelerators
Loose coupling
AD -> Renovation
 ELENA -> New accelerator
 Main constraints

◦ AD injection : Can’t wait on the flat top.
 synchronization at the start of the
ramp
◦ Cycle length unknown (Stop)
◦ AD ejections to ELENA
Future of the CERN timing
 First White Rabbit implementation
Strong coupling accelerators
Stop
Inj.
Ej. to ELENA
WR deployment
WRDM
AD & ELENA
WR/GMT
Gateway
GTM
GTM
Receivers
GTM
Receivers
GTM
Receivers
Receivers
 Deployment for end of 2013
◦ Only a WRDM, No WR nodes foreseen
◦ WR to GMT gateway (end 2012)
◦ Use of GMT receivers
 Deployment for end of 2014
◦ AD in production, ELENA in commissioning
 2 WRDM ?
 Deployment for end
◦ Both in production
Future of the CERN timing
 First White Rabbit implementation
of 2015