ebpms and orbit feedback electronics for...
TRANSCRIPT
Guenther Rehm 1
EBPMs and Orbit Feedback Electronics for Diamond
IWBS 2004
Guenther Rehm
Guenther Rehm 2
Outline
• BPM locations and cross sections
• BPM response calculations
• EBPM performance examples
• SOFB plans
• FOFB plans
Guenther Rehm 3
Number of BPMs
Button / octagonal+oval120+48SR
204Sum
Stripline / circular7BTS
Button / elliptical22Booster
Stripline / circular7LTB
-0LINAC
Type / cross sectionCountLocation
All with Libera EBPM electronics
Guenther Rehm 4
Storage ring BPMs
• Primary BPMs:– Increased sensitivity through smaller aperture
– Mechanically decoupled through bellows
– Position monitored relative to reference pillar
• Standard BPMs:– BPM blocks welded into vacuum vessel
– Mounted on “anchor” stands
– Position monitored relative to quad centre
Guenther Rehm 5
BPM locations in SR
Standard BPM
Primary BPM
-50 -40 -30 -20 -10 0 10 20 30 40
-20
-10
0
10
20
-40 -30 -20 -10 0 10 20 30 40
-10
-5
0
5
10
Guenther Rehm 6
BPM response calculation
• MATLAB based boundary element solver– Fast: 5.5 sec on P4/1700 for 722 boundary
elements and 441 beam positions
– Precise: results checked with finite element solver (Vector Fields OPERA)
• Geometrical manufacturing uncertainties have been modelled using Monte Carlo simulation
Guenther Rehm 7
Calibration Factor for Primary BPM
2 4 6 89
10
11
12
13
14
offset from centre [mm]
ca
libra
tio
n fa
cto
r [m
m]
Kx
Ky
Guenther Rehm 8
0 2 4 6 8 10
0
1
2
3
4
5
6
7
8
9
10
beam x offset [mm]
beam
y o
ffset
[mm
]
BPM reading
real position
Nonlinear 2D BPM response
0 2 4 6 8 10
0
1
2
3
4
5
6
7
8
9
10
beam x offset [mm]
beam
y o
ffset
[mm
]
BPM reading
real position
Needs to be corrected
before nonlinear
beam dynamics studies!
Guenther Rehm 9
RMS noise @ TBT
-80 -70 -60 -50 -40 -30 -20 -10 010
0
101
102
103
Pin(dBm)
RM
S (
µm
)
300 mA30 mA
Guenther Rehm 10
Resolution with 1 kHz BW
-80 -70 -60 -50 -40 -30 -20 -10 010
-1
100
101
102
Pin(dBm)
RM
S (
µm
)
300 mA30 mA
Guenther Rehm 11
RMS noise with
1 kHz Bandwidth
3.3/2.21.35/1.51-10 mA
1.3/0.90.54/0.610-60 mA
0.65/0.450.27/0.360-300 mA
Standard x/y in µmPrimary x/y in µmBeam current
Guenther Rehm 12
Control and Instrumentation Areas
(CIAs)
RA CK 3DIAGNOSTICS
AND CONTROL
RA CK 4DIAGNOSTICS
AND CONTROL
RA CK 2PROTECTION
AND CONTROL
RACK 1CONTROL, NETWORK
& TIMINGDISTRIB UTION
RA CK 5SPARE
RACK 8DIP,QUAD & SEXTPOWER SUPPLIES
AND CONTROL
RACK 9DIP,QUAD & SEXTPOWER SUPPLIES
AND CONTROL
RACK 7DIP,QUAD & SEXTPOWER SUPPLIES
AND CONTROL
RACK 6DIP,QUAD & SEXTPOWER SUPPLIES
AND CONTROL
RA CK 10SPARE
RA CK 13INSERTION DEVICE
MOTOR DRIVESYSTEM AND
CONTROL
RA CK 12STEERING
POWER SUPPLIESAND CONTROL
RA CK 14INSERTION DEVICE
MOTOR DRIVESYSTEM AND
CONTROL
RACK 11STEERING
POWER SUPPLIESAND CONTROL
RA CK 15SPARE
RA CK 20SPARE
RACK 19FRONT END VA CUUM
INSTRUMENTATIONAND CONTROL
RACK 18FRONT END VA CUUM
INSTRUMENTATIONAND CONTROL
RACK 17VAC UUM
INSTRUMENTATIONAND CONTROL
RACK 16VAC UUM
INSTRUMENTATIONAND CONTROL
OUTER SHEILD WALL
450 CAB LE TRAYD
ISTR
IBU
TIO
NB
OA
RD
DIS
TR
IBU
TIO
NB
OA
RD
EXACT REQUIREMENT FOR150\ U+ 2205WILL BE DEPENDANT UPON
CABLE QUANTITIES &EQUIPMENT LOCATIONS150\ U+ 2205 DUCTS
OUTER SHEILD WALL
24 temperature stabilized CIAs for 19” racks
Guenther Rehm 13
Pro
ce
sso
r
Controls Network (EPICS)
VME
eBPM eBPM eBPM eBPM eBPM eBPM eBPM
SOFB Setup
14+8 Corrector PSUs
PSU 1 PSU N
PS
U I
F
PS
U I
F
…
Centralfeedback
calculation
One CIA
Guenther Rehm 14
SOFB Details
• EPICS IOCs run inside LIBERA
• EPICS interface to PSU already available
• MatLab channel access makes application development easy
• Can be tested with “virtual accelerator”
• Expected to run at 10 Hz sampling, 0.5 Hz closed loop BW
• Will be available on day 1
Guenther Rehm 15
Virtual Accelerator and Software
Commissioning
Real
Accelerator
Virtual
Accelerator
The Virtual Accelerator is used
1) to simulate the control system environment as seen by the users
2) to provide a realistic test for AP applications
The Virtual Accelerator uses the Tracy–II code to simulate the physical
behaviour of the ring
Physics
Application
Software
Guenther Rehm 16
Basic Virtual Accelerator
Functionality
• Set magnet currents
• Read EBPM x/y average calculated using Tracy2 closed orbit
• Read EBPM x/y turn-by-turn buffer using Tracy2 particle tracking
EBPM MAGNET
TRACY 2
EPICS EPICS
Guenther Rehm 17
“measured” Response Matrix
Guenther Rehm 18
MatLab AT Based Feedback
Orbit Correct implemented using AT for Spear 3
Guenther Rehm 19
Design constraints for FOFB
• Global system, all data should be available
everywhere
• Low latency from hardware, main delay should
result from LP filter
• FB algorithm should be easily serviceable
• Corrector PSU interface is VME
• Robust system which continues to perform with partial faults
Guenther Rehm 20
Pro
ce
sso
r
Controls Network
VME
eBPM eBPM eBPM eBPM eBPM eBPM eBPM
Cell -m
Cell -n Cell +n
Cell +m
FOFB Setup (one CIA)
14+8 Corrector PSUs
PSU 1 PSU N
PS
U I
F
PS
U I
F
…
Event Network
Eve
nt R
x
AD
Cs
Photon BPMs
pBPM pBPM
FB
Pro
ce
sso
r
PM
C R
ocke
t IO
Guenther Rehm 21
FOFB Details
• FB data produced at 4-20 kSamples/s• Dedicated FB CPU board MVME5500 running vxWorks,
but no EPICS IOC, no network.• RocketIO in Virtex2Pro to run at 2.5 Gbit/s• PMC card with RocketIO will be board developed for
timing system• Connections inside rack can be galvanic, longer
distance will be single mode fibre• All connections between CIAs will be patched centrally• Communication is broadcast, no routing or location
information is required for any node.
Guenther Rehm 22
FOFB Delays
(simulated/estimated)
• Distribution of 168 sets of data to 168+24 locations: 30 µs
• Transfer to CPU: 10 µs
• Matrix multiplication: 30 µs (worst case)
• Write into PSU: 50 µs
• 200-400 µs delay for LP filter– > feedback BW >>100 Hz should be feasible
• Detailed simulation is required!
Guenther Rehm 23
Acknowledgements
• DLS: Mark Heron, Ian Martin, RiccardoBartolini, Tony Dobbing
• Instrumentation Technologies
• Supercomputing Systems