CESR beams in order to sense nm-size
mechanical vibrations?
Report on machine experiments at CESR in June 2009
CERN:
Marek Gasior: BBQ electronics
(Andrea Boccardi: VME electronics)
Juergen Pfingstner: Beam measurements
Magnus Sylte: Vibration measurements
Hermann Schmickler: not much useful
CESR:
Mark Palmer, Mike Billing, operations crew
and the valuable support at “lightning speed” of John Barley
Outline
• Motivation
• Experimental Set-up
• First results
- amplitude calibration
- residual beam motion
- noise of detection system
• Conclusions and Perspective
CLIC stabilization requirements
• Mechanical stabilization requirements:
Quadrupole magnetic axis vibration tolerances:
Vertical
Horizontal
Final Focus
quadrupoles
Main beam
quadrupoles
0.1 nm > 4 Hz
1 nm > 1 Hz
5 nm > 4 Hz
5 nm > 1 Hz
• Main beam quadrupoles to be mechanically stabilized:
– A total of about 4000 main beam quadrupoles
– 4 types
– Magnetic length from 350 mm to 1850 mm
Present design approach
(CLIC stabilization WG, C.Hauviller et al.)
• Mechanical active stabilization in a feedback loop using
electromechanical sensors and (piezo) actuators
• Optimized mechanical design for
- girders, magnets and electromechanical alignment system
- best choice for number and position of actuators and sensors
- low Q mechanical resonances in order to avoid vibration
amplification
- mechanical resonances at the highest possible frequencies
• Mimimization of environmental noise
through isolation from vacuum chamber vibration, coolent flows,
cable vibration and microphonic coupling
• Experimental verification of the result of stabilization:
- construction of real hardware based on a quadrupole prototype, an
active stabilization system
Present work program: A type 4 quad ready for lab tests mid 2010.
Main Beam Mock-up
Work launched within the collaboration
•
Functionalities
–
•
Demonstrate stabilization in operation:
• Magnet powered, Cooling operating
• Configurations
– 1- Stand-alone
– 2- Integrated in Module
– 3- Interconnected
• Accelerator environment
Parts / Measuring devices
–
–
–
–
–
–
–
Floor (damping material)
Support
Pre-alignment
Stabilization
Magnet
Vacuum chamber and BPM
Independent measurement
Slide taken from C.Hauviller, ACE 2009
Main beam quadrupole
• Under final design.
• Plain material (incompatible with corrector magnet)
• Assembly methods to be tested (accuracy of some microns!)
Slide taken from C.Hauviller, ACE 2009
Necessary complementary verification ?
• The demonstration of the stabilization of the magnet (=Magnetic
field?) is based on “zero” signals of electromechanical sensors on
the outer shell of the magnet.
• The physical size of the sensors do not allow to mount them close to
the pole tips or inside the magnet.
• Pole tip vibrations, coil vibrations might exist without the outer
monitors measuring them.
• The limited number of monitors might not catch all vibrations.
Question:
can another physical process be used to verify the stability of the
magnetic field?  try a high energetic low emittance particle beam
Validation of Quad stabilization principle (1/2)
Standard
Quad
Standard
Quad
Standard
Quad
Standard
Quad
Standard
Quad
Standard
Quad
Stabilized
Quad
Calibrated mechanical
exciter
High
sensitivity
BPM
Validation of Quad stabilization principle (2/2)
• insert a CLIC quadrupole (fully integrated into a CLIC module with
a mechanical simulation of the environmental noise) into an electron synchrotron
• in frequency bands in which the intrinsic motion of the particle beam
is smaller than 1nm, observe the effect of quad stabilization on/off
• in frequency bands, where the particle beam moves more than 1nm,
the beam validation is limited to exciting mechanical vibrations of the quad
at larger amplitudes and measuring the gain of the feedback.
The performance of the feedback system at lower amplitudes would in this case
to be estimated from the signal to noise ratio of the actuators and sensors.
 objective of the test experiment at CESR-TA:
- what is the rms of the residual eigen-motion of the CESR beam as a function
of frequency
- what are the limits in noise performance of the BPM electronics?
experimental set-up
• Excitation of beam with a vertical orbit corrector dipole,
direct connection to dipole coil (Q10W)
• Observation of beam oscillations on vertical pickups with
modified BBQ electronics (Q8W)
heavy downsampling in special acquisition cards,
up to 17 minutes measurement time.
• Calibration of the system using a 300 um peak-peak
oscillation measured in parallel with BBQ system and
local orbit system.
• Various beam conditions, partial shutdown of injector
complex etc…
• 4 measurement shifts
• Very friendly and effective support by CESR team
Diode detectors on PU-Q8W
Data processing
NIM crate
DAQ
6 ch.
DAC
Ground
motion
meas.
6 ch.
ADC
sine wave
generation
sown
sampling
factor 4
• DAC and ADC
• sampling rate: 44kS/sec
• resolution 24 Bit
•FFT:
• Length: 64kS
• no windowing
• 22 FFTs are averaged
(measurement time >1000sec)
In CTF3 this measurement would take about 100 days
VME crate with DABs
down
sampling
factor 1 to 256
FFT
USB
• Data transfer and
• Control of DABs and
DAQs
LabView program
Storage
Displa
y
Laptop (Windows XP)
VME-crate with DAB cards
Digitizers
Getting BPM resolutions below the nm
•
•
•
•
Aperture of BPM approx. 50 mm or more
Wide band electronics thermal noise limit: 10^-5 of aperture
Narrow band front-end gains factor 10…100
State of the art commercial BPM system (“Libera Brilliance”)
reaches 5nm/sqrt(Hz),
i.e. with 1000 s measurement time 150 pm rms noise.
• Different approach:
BBQ electronics: “Zoom in” getting high sensitivity for beam
oscillations, but loosing absolute information of DC = closed orbit
information.
Direct Diode Detection (3D) – the principle








Peak detection of position pick-up electrode signals (“collecting just the cream”)
fr content converted to the DC and removed by series capacitors
beam modulation moved to a low frequency range (as after the diodes modulation is on much longer pulses)
A GHz range before the diodes, after the diodes processing in the kHz range
Works with any position pick-up
Large sensitivity
Impossible to saturate (large fr suppression already at the detectors + large dynamic range)
Low frequency operation after the diodes
• High resolution ADCs available
• Signal conditioning / processing is easy (powerful components for low frequencies)
M.Gasior, BE-BI
Direct Diode Detection – The principle
Signals of both peak detectors
Detector signal difference
 = 0,  = 0.01
q = 0.1, Cf = Cpu
 =T
 = 100 T
4 bunches
 = 100 T
M.Gasior, BE-BI
Architecture of the Base Band Q (BBQ) Measurement System
Detector box (for one PU electrode)
 For CESRTA the system bandwidth is 10 Hz – 5 kHz
M.Gasior, BE-BI
Analog front-end box (2 channels)
Amplitude calibration
•
•
•
•
•
The BBQ electronics does not have a good absolute calibration for the
measurement of absolute beam oscillation amplitudes.
On the contrary the BBQ electronics is linear over many decades and
frequency independent within the bandwidth given by the electronic filters.
Disconnect orbit steerer from control system and get two wires for own
excitation…
1st attempt of calibration: DC mode: inject 1 A ( lab current source)of DC
current into steerer magnet, measure (calculate) orbit displacement at BBQpickup and use this relation for future low current AC excitations.
Done, but problem is unknown bandwidth of magnet and vacuum chamber
2nd attempt: AC-mode: inject AC modulation (0.5 A rms) at various
frequencies and measure resulting orbit oscillation with BPM system.
(For 0.5 A we had to borrow a HiFi amplifier and we just managed to be
above the sensitivity of the turn by turn CESR orbit system:
0.5 A at 20 Hz gave a 300 +-30 um peak-peak orbit oscillation.
DC
Calibration
AC
0.5 A@20 Hz
300 +-30 um
0.5 A@40 Hz
150+-30 um
0.5 A@80Hz
80+-30 um
Preliminary “fit by eye”;
data of this calibration
(CESR BPM rawdata)
on a file on a website,
which is presently not
accessible to us
- 1 mm at Q8W
Amplitude Calibration
Measured in parallel with turn by turn orbit
system:
measured amplitude: 300 um pp ~ 100 um rms
Simultaneous excitation with 6 different tones @ 17.2 mA rms each
meas. 20_40_80_160_320_640_final
-30
Magnitude [dBFS]
2 GeV electrons
Excitation at 20, 40, 80, 160, 320, 640 Hz, 17.2 mArms
-50
-70
-90
-110
0
100
200
300
400
Frequency [Hz]
500
600
700
Continous Calibration
• In the end we
decided to leave a
continuous 20 Hz
excitation of 20 mA
as reference
excitation through all
measurements.
• This corresponds to
a beam oscillation of
about 4 um rms at 20
Hz.
4 um reference
tone @ 20 Hz
1 nm line
Noise evaluation
40 pm
FFTs averaged sigma
1
18,07 pm
22
3,67 pm
Ratio: 4,92 <-> sqrt 22 = 4,69
18 pm in 47 s measurement time = 0.12 nm/sqrt(Hz)
Compare to Libera Brillance with 0.25 um @ 2KHz = 5nm/sqrt(Hz)
Is all we measure beam motion?
There is more signal with
the synchrotron off
What is this pedestal?
PM to AM demodulation of Rf-noise
• With the BBQ detection principle the
system is sensitive to hase jitter of the RF,
i.e. jitter of the revolution time.
Vibration Sensors, Setup 1
Quadrupole Q8W
Geophones
Sensitivity: 2000 V/(m/s)
Frequency range: 1/30 – 80 Hz
Accelerometers
Sensitivity: 1 V/(m/s^2)
Frequency range: 0,1 – 200 Hz
18/06/2009
Mechanical Measurement Lab
Magnus Sylte
EDMS 1004462
Acquisition System
and Analysis Parameters
MKII
Version:
Max sampling frequency: 384 KHz
A/D converter: 24 bit
Dynamic range: 10mV-50V
Analysis Parameters:
Measurement length: 1024s and 1440s
Window: Hanning
Sampling rate: 1024Hz
Block length: 64s
Average: Linear
Overlap: 66,7%
18/06/2009
Mechanical Measurement Lab
Magnus Sylte
EDMS 1004462
Average Spectra on ground and on top of quadrupole
compared to beam motion
Average FFT Geophones
Average FFT BPM
Ground Vibration
Quadrupole Vibration
18/06/2009
Mechanical Measurement Lab
Magnus Sylte
EDMS 1004462
Quad (black)
Beam
Does not really fit, but there are more than 100 quads on different supports…
First vertical bending modes
Transfer function
Coherence
Phase
18/06/2009
Mechanical Measurement Lab
Magnus Sylte
EDMS 1004462
First lateral bending mode
Transfer function
Coherence
Phase
18/06/2009
Mechanical Measurement Lab
Magnus Sylte
EDMS 1004462
Vibration Sensors, Setup 2
Accelerometer 2
Accelerometer 1
08/07/2009
Mechanical Measurement Lab
Magnus Sylte
EDMS 1004462
Accelerometer 1
Average FFT
100n
m
10n
Accelerometer 1
Geophone on the floor
1n
100p
10p
1p
100f
0
10
08/07/2009
20
30
40
50
60
70
80
90
Mechanical Measurement Lab
Magnus Sylte
100
Hz
EDMS 1004462
Accelerometer 2
Average FFT
100n
m
10n
Accelerometer 2
Geophone on the floor
1n
100p
10p
1p
100f
0
10
08/07/2009
20
30
40
50
60
70
80
90
Mechanical Measurement Lab
Magnus Sylte
100
Hz
EDMS 1004462
Comparison BPM vibration and beam spectrum
Side product:
modified BBQ electronic with higher bandwidth:
perfect tune-monitor with 60 db signal/noise ratio
Perspectives for the future
1 nm line
Feedback Off
Feedback ON
Conclusions and Perspectives (1/2)
• An electron beam (tens of um beam size) can be used to sense
disturbances (vibrations) down to the sub-nm level
- using an optimized BBQ electronics
- using about 10^9 samples in 17 minutes measurement time
• At CESR-TA the accelerator itself has disturbances that add up to
- smaller values than 1 nm rms above 80 Hz
- to much larger values below 80 Hz
• We need to clarify what fraction of this is PM-AM demodulation of
RF-phase jitter
• Hence in the interesting region 10 Hz – 100 Hz no full demonstration
of the stability of a CLIC quadrupole proto-type can be made,
but:
- in this frequency range the gain of a stabilization loop can be
measured at 10-100nm levels
- a special optics can be designed in order to get higher sensitivity
using the amplification of the lattice.
• Due to its availability CESR-TA stays interesting as CLIC test-bed
Conclusions and Perspectives (2/2)
• Other European light-sources have been contacted
(Diamond, Soleil, PSI, ESRF, PetraIII) for similar experiment.
 so far no “enthusiasm” to install a new quadrupole into their
machine
 PSI will receive us in late autumn for the same measurement
and proposes to stabilize an existing SLS quadrupole
• CESR-TA would hopefully except to have a CLIC quad installed
 limited experimental program at low frequencies
 (for all machines): not the real CLIC quad: higher aperture, lower
strength… still a lot of information to be gained by these experiments
• Presently we study the CESR optics in order to calculate the amplitude
factor between quad motion and visible amplitude in the BPM.
This factor can be much bigger than 1!
• Such an experience is only possible at all due to a major improvement
in BPM sensitivity for beam oscillations from
5 nm/sqrt(Hz) (Libera Brillance) to 0,12 nm/sqrt(Hz) (BBQ)
• Several details of BBQ noise performance, frequency dependence,
beam current dependence and dependence of beam emittance to be
measured