DESIGN OF A MULTI-BUNCH BPM FOR THE NEXT LINEAR COLLIDER1
Andrew Young, Douglas McCormick, Marc Ross, Stephen R. Smith
SLAC, Stanford, CA. USA
H. Hayano, T. Naito, N. Terunuma, S. Araki
KEK, Ibaraki, Japan
Abstract. The Next Linear Collider (NLC) will collide 180-bunch trains of electrons and
positrons with bunch spacing of 1.4 ns. The small spot size (ay < 3 nm ) at the interaction point
requires precise control of emittance, which in turn requires the alignment of individual bunches
in the train to within a fraction of a micron. Multi-bunch beam position monitors (BPMs) are to
determine the bunch-to-bunch misalignment on each machine pulse. High bandwidth kickers
will then be programmed to bring the train into better alignment on the next machine cycle. A
prototype multi-bunch BPM system with bandwidth (350 MHz) sufficient to distinguish adjacent
bunches has been built at SLAC. It is based on 5 G sample/s digitization of analog sum and
difference channels. Calibration tone injection and logging of the single bunch impulse response
provide the kernel for deconvolution of bunch-by-bunch position from the sum and difference
waveforms. These multi-bunch BPMs have been tested in the Accelerator Test Facility at KEK
and in the PEP-II ring at SLAC. The results of these measurements are presented in this paper.
OVERVIEW
The multi-bunch (MB) BPMs were designed to operate over a wide range of
conditions (Table 1) allowing for testing to be performed at SLAC and KEK. The MB
BPMs are used by the sub-train feedback, which applies a shaped pulse to a set of
stripline kickers to straighten out a bunch train. These are qualitatively different from
the quad (Q) and feedback (FB) BPMs due to their high bandwidth and relatively
relaxed stability requirements. The primary requirement on the MB BPMs is a bunch
train that generates a BPM signal, which is straight.
TABLE 1. BPM Specifications_______________________________
Parameters______________Value___________Comments____
Resolution
300 nm rms
For bunch-bunch
at 0.6xl010 eVbunch
displacements freq. Below
300 MHz
Position range
+2mm
Bunch spacing
1.4ns
No. of bunches
1-190
1.4ns
Beam current
Ixl0 9 -1.4xxl0 10
Particles per bunch
No. of BPMs
278
1
Work supported by the Department of Energy, contract DE-AC03-76SF00515
CP648, Beam Instrumentation Workshop 2002: Tenth Workshop, edited by G. A. Smith and T. Russo
© 2002 American Institute of Physics 0-7354-0103-9/02/$19.00
508
Implementation
Implementation
Figure
diagram of
of the
the multi-bunch
multi-bunch front-end
front-end prototype
prototype
Figure 11 shows
shows aa simplified
simplified block
block diagram
chassis.
directional couplers,
couplers, sum
sum and
and difference
difference hybrid,
hybrid,
chassis. The
The BPM
BPM chassis contains directional
bandpass
sold-state amplifies.
amplifies. The
The BPM
BPMchassis
chassistakes
takes
bandpass filters
filters for noise
noise rejection, and sold-state
the
BPM buttons
buttons that
that can
can be
be translated
translatedinto
intoXX
the four
four inputs
inputs ("Up"
(“Up” "Down")
“Down”) from the BPM
and
The BPM
BPM signal
signal isisthen
thenamplified
amplified
and Y
Y positions
positions and takes the sum and difference. The
in
outside the
the radiation
radiationarea.
area.The
Thefront-end
front-end
in order
order to
to run
run itit on a long cable to a digitizer outside
has
into the
the inputs
inputs of
of the
the sum
sum and
and
has aa feature
feature where
where a single tone can be injected into
difference
hybrid
for
calibration.
Thus
allowing
the
operators
to
perform
a
transfer
difference hybrid
allowing the operators to perform a transfer
function
function measurement.
measurement.
Figure 1.
1.
Figure
Block
Block diagram
diagram of
of the
the BPM
BPM front-electronics
front-electronics
To meet
meet the
the bandwidth
bandwidth and
To
and performance
performance requirements,
requirements, aa four-tap
four-tap directional
directional
coupler
that
is
shown
in
figure
2
and
allows
one
to
inject
a
single-tone
coupler that is shown in figure 2 and allows one to inject a single-toneinto
intothe
theHybrid
Hybrid
and perform
perform aa transfer
transfer function
function calibration.
and
calibration. The
The four-tap
four-tap configuration
configuration allows
allowsthe
the
bandwidth to
to be
be broadband.
broadband. This
bandwidth
This has
has the
the effect
effect of
of aa constant
constant coupling
couplingratio
ratioacross
acrossthe
the
frequency band.
band. This
This allows
allows one
frequency
one to
to remove
remove the
the phase
phasevariation
variationininthe
thehybrid
hybridand
andcable
cable
losses
and
insertion
losses
of
other
components.
Also
the
calibration
will
allow
losses and insertion losses of other components. Also the calibration will allowone
onetoto
remove any
any non-linearities
non-linearities in
remove
in the
the digitizer
digitizer that
that will
will be
bediscussed
discussedlater
laterininthe
thepaper.
paper.
Figure 2.
Figure 2.
Stripline directional coupler
Stripline directional coupler
Other RF custom components such as amplifiers and switches were bought from
Other RF custom components such as amplifiers and switches were bought from
manufactures and then integrated into a circuit board. Special care was taken to lower
manufactures
and then integrated into a circuit board. Special care was taken to lower
insertion losses by using low-loss dielectric materials. The most critical component is
insertion losses by using low-loss dielectric materials. The most critical component is
the sum and difference hybrid. Instead of a stripline 5/4λ sum-difference hybrid
the
suma and
difference
hybrid.
of a stripline
sum-difference
design,
toroid
design was
used. Instead
This improved
the sum 5/4K
to difference
isolationhybrid
from
design,
a
toroid
design
was
used.
This
improved
the
sum
to
difference
20dB to >30dB and the bandwidth improved from 150MHz to 300MHz. isolation from
20dB to >30dB and the bandwidth improved from 150MHz to 300MHz.
509
The digitizer that was chosen for this experiment is a Tektronix 684..This is a
5GSa/s digital scope that has a maximum record length of 15000 points. The
waveforms were read out using a MATLAB script that communicated through the
scope was
was triggered away from the injection aa trigger
trigger was
GPIB. To insure that the scope
of the
the beam
beam train.
train. Seven
Seven turns
turns were
were recorded
recorded and
and then
then analyzed.
analyzed.
timed to the middle of
RESULTS
RESULTS
Three Y-position BPMs chassis were installed in the KEK Accelerator Test Facility
(ATF). The toroid current was recorded with each data set. The data presented in this
10
paper is with toroid current of 3e
3e10
particles per bunch. Figure 3 shows the results of a
single-bunch beam
beam stimulus.
stimulus.
frequency spectrum of aa single-bunch
Windowed Freq domain ATF Single-bunch BPM sblrb1
1000
1500
2000
2500
3000
3500
4000
4500
5000
Freq (MHz)
Figure 3.
Frequency spectrum of a single bunch beam
An elaborate calibration procedure was used to normalize the front-end. The
normalization was needed because the multi-bunch data will have overlapping
bunches. Therefore, the single bunch data can be used to decimate the individual
multi-bunch data
data into charge
bunches of the multi-bunch data thus, transforming the multi-bunch
bunches.
and position of real bunches.
First the front end was calibrated using a single tone at 600MHz, the center
frequency of the hybrid and RF coupler. Four gain and phase coefficients were derived
(Up to sum, Up to difference,
difference, Down to sum, Down to difference).
difference). These coefficients
1).
were used in the single bunch data (see equation 1).
Cor sb=FFT(sb)*(FFT(tone)/FFT(600MHz))
Cor_sb=FFT(sb)*(FFT(tone)/FFT(600MHz))
(1)
(1)
The next calibration procedure was to inject a single bunch of charge into the
machine and measure the response of the front-end. This also generated four
coefficients over the complete bandwidth of the front-end. Because the single bunch
beam calibration data, was digitized at separate times aa phase error was injected into
510
signals are illustrated in
in Figure
Figure 4.
4. This
This phase
phase error
error was
was
the signals. The corrected signals
corrected by writing aa MATLAB script that ensured the digitizers started digitizing at
time.
the same time.
ATF BPM red/blue side sblrb.
545
550
time (ris)
Figure 4.
data
Single bunch raw beam data
Using the corrected single bunch data, an inverse matrix was generated that was
multi-bunch data
data (see
(see equation
equation 22 and
and 3)
3)
used to correct the multi-bunch
sb11n
U   sb
 D  =  sb
   ?h21
−1
Of/21
sb12
Σ 
12 
∗  meas 

?h
sb22 
∆ meas 
(2)
(2)
Of/™
 Σ cor  U  1I 1I 
∆  =  D  ∗ 11 −-1
1
 cor    
(3)
(3)
computing the corrected sum
sum and difference signals.
signals.
Two procedures were used in computing
first procedure
procedure was
was to compute
compute the corrected
corrected sum
sum and
and difference
difference using
using aa
The first
multibunch data
data by
by windowing the data
MATLAB script that simplifies the raw multibunch
around the turn frequency and
and then taking an FFT of
of the
the data.
data. The
The second
second procedure,
procedure,
around
around the turn frequency
frequency and
and then
then convert
convert the
the data
data to
to
was to window the data around
baseband. Then
Then data
data was
was decimated as
as aa function of the bucket spacing.
spacing. The
baseband.
multi-bunch raw
raw data
data is
is shown
shown in
in Figure
Figure 5.
5. In
In this
this figure,
figure, both the sum and
windowed multi-bunch
signals are
are displayed.
displayed. Examination
Examination of
of this
this data
data determined
determined that
that there
there is
difference signals
bunches in
in the accelerator.
accelerator. The data shows
shows that there is greater than–30dB
than-30dB of
eighteen bunches
isolation between
between the
the sum
sum and
and difference
difference ports.
ports.
isolation
511
ATF BPM multibunch at turn 4 mb71150
200
Figure
Figure 5.
5.
Figure
5.
250
time (ns)
300
Multi-bunch
Multi-bunch raw
raw beam
beam data
data
Multi-bunch
raw
beam
data
Using
the
first
procedure
Using the
the first
first procedure
procedure and
and Equation
Equation 222 and
and 333 the
the corrected
corrected
multi-bunch
data
is
Using
and
Equation
and
the
corrected multi-bunch
multi-bunch data
data is
is
computed.
computed. Figure
Figure 666 shows
shows the
the envelope
envelope of
of the
the corrected
corrected multi-bunch
multi-bunch sum
sum
data and
and
the
computed.
Figure
shows
the
envelope
of
the
corrected
multi-bunch
sum data
data
and the
the
raw
sum
signal
The
overall
eveelope
matches
raw sum
sum signal
signal The
The overall
overall eveelope
eveelope matches
matches the
the structure
structure of
of the
the raw
raw BPM
BPM signal.
signal.
raw
the
structure
of
the
raw
BPM
signal.
Using
this
along
with
the
corrected
Using this
this signal
signal along
along with
with the
the corrected
corrected difference
difference signal
signal the
the y-position
y-position and
and
Using
signal
difference
signal
the
y-position
and
resolution
resolution can
can be
be computed.
computed.
resolution
can
be
computed.
corrected data vs real data at turn 4
MO
420
tjme(ns)
Figure
Figure 6.
6.
Figure
6.
1;60
Multi-bunch
Multi-bunch corrected
corrected sum
sum and
and raw
raw sum
sum
data
Multi-bunch
corrected
sum
and
raw
sum data
data
Figure
77 shows
the
beam
position
Figure 7
shows the
the average
average beam
beam position
position over
over the
the 18
18
bunches
using
the
first
Figure
shows
average
over
the
18 bunches
bunches using
using the
the first
first
procedure.
procedure. As
As one
one can
can see
see the
the position
position and
and resolution
resolution begins
begins
to
get
large
at
the edge
edge of
of
procedure.
As
one
can
see
the
position
and
resolution
begins to
to get
get large
large at
at the
the
edge
of
the
train.
the train.
train. This
This effect
effect is
is due
due to
to the
the corrected
corrected sum
sum
signal
increasing and
and
the
corrected
the
This
effect
is
due
to
the
corrected
sum signal
signal increasing
increasing
and the
the corrected
corrected
difference
difference signal
signal is
is approaching
approaching zero.
zero.
difference
signal
is
approaching
zero.
512
T— avgVpos over thel 8 launches at turn 4
*— resolution wer the train
-20
,30
200
450
1250
time;(ns|
Figure 7.
Figure
7.
Multi-bunch
Y-position
over
seven
turns
data
Multi-bunch Y-position
Y-position over
overseven
seventurns
turnsdata
data
Multi-bunch
and Equation
data
Using the second procedure and
Equation 22 and
and 333 the
the corrected
corrected multi-bunch
multi-bunchdata
dataisis
is
corrected multi-bunch
sum
data
down-converted
computed.
multi-bunch sum
sum data
data down-converted
down-converted toto
to
computed. Figure 8
8 shows the corrected
filtered at
at
turn
4.
Also
illustrated
the
raw
sum
signal
baseband
and low-pass
at turn
turn 4.
4. Also
Also illustrated
illustratedisisisthe
theraw
rawsum
sumsignal
signal
baseband and
low-pass filtered
corrected data vs real data at turn 4
80
90
100
110
120
130
140
150
time(ns)
8.
Figure 8.
Figure
Multi-bunch
Multi-bunch baseband
baseband corrected
corrected
sum
and
raw
sum
data
Multi-bunch
baseband
corrected sum
sumand
andraw
rawsum
sumdata
data
maximum signal
By sampling
sampling the
the maximum
the sum
sum and
and difference
difference ports,
ports, the
the position
position of
of
By
signal on
on the
the
sum
and
difference
ports,
the
position
of
of
turns
can
be
calculated.
The
equation
for
the
Y-position
is
the
beam
as
a
function
calculated.
The
equation
for
the
Y-position
the beam as a function of turns can be calculated. The equation for the Y-position is
is
defined as
Ypos=R/2(∆/Σ) where,
the radius
radius of
of the
the beam
beam pipe,
pipe, R/2=3000microns,
R/2=3000microns,
defined
as Ypos=R/2(A/E)
where, R
R is
is the
the
radius
of
the
beam
pipe,
R/2=3000microns,
is the
the corrected
corrected difference
difference signal,
the
corrected
sum
signal
Figure
shows
that
A∆ is
signal, ΣE isis
is the
the corrected
corrected sum
sumsignal
signalFigure
Figure999shows
showsthat
that
15
microns
over
seven
turns
with
a
single
shot
resolution
of
the
beam
position
varies
over
seven
turns
with
a
single
shot
resolution
of
the beam position varies 15 microns over seven turns with a single shot resolution of333
microns.
microns.
513
f
I
\
\
>-— rej6ltitf6n over the train
Up|i i
•An-
£
|nj\
—— -{— -
*: a
rnicrons
f
—— avg Ypbs oxrerthe 18 lunches at turn 4 ,,:,>„;; , , > ^
40
-10
£
H^
-,,,„,.„,,;;,,,,,,
I
\
\
;
;:;
1
I
\
I
I
\
I
|,i,,,,i,.^.,^l^,.^i^,,,4.^.,4,^,.^
...,j;,,.,.,J,.,,,,,,,l.^
s
!
!.-
:.''
:•:
?•
•
i
i-
^
^i
i
300
350
400
i
100
50
Figure
9.
Figure9.
9.
Figure
1|0
200
;250
timers}
45
Multi-bunch
Y-position
Multi-bunchY-position
Y-position over
over seven
seven turns
turns data
data
Multi-bunch
irns Y^pos dwei' esaeh turri rnt)71ri150
1
2;
Figure10.
10.
Figure
10.
Figure
3
4
turns
;5
6
7
RMSY-Position
Y-Position over
over each
each turn
turn
RMS
Y-Position
over
each
turn
RMS
Figure10
10illustrates
illustratesthat
thatthe
the position
position resolution
resolution over
over each
each turn
turn and
and varies
varies from
from 5.45.4Figure
10
illustrates
that
the
position
resolution
over
each
turn
and
varies
from
5.4Figure
11.1
microns.
This
will
need
to
be
improved
to
meet
the
specification
for
the
NLC.
11.1
microns.
This
will
need
to
be
improved
to
meet
the
specification
for
the
NLC.
11.1 microns. This will need to be improved to meet the specification for the NLC.
SUMMARY
SUMMARY
SUMMARY
Amulti-bunch
multi-bunch BPM
BPM was
was built
built at
at SLAC
SLAC and
and tested
tested at
at KEK.
KEK. The
The BPM
BPM electronics
electronics
A
multi-bunch
BPM
was
built
at
SLAC
and
tested
at
KEK.
A
The
BPM
electronics
can
resolve
both
single
and
multi-bunch
fills.
The
data
clearly
indicates
that
the multimultican
resolve
both
single
and
multi-bunch
fills.
The
data
clearly
indicates
that
the
can resolve both single and multi-bunch fills. The data clearly indicates that the
multibunch
BPM
electronics
can
measure
the
beam
to
within
15
microns
and
6
microns
bunch
BPM
electronics
can
measure
the
beam
to
within
15
microns
and
6
microns
bunch BPM electronics can measure the beam to within 15 microns and 6 microns
overseven
seventurns.
turns. However,
However, the
the design
design goal
goal of
of measuring
measuring the
the beam
beam position
position within
within 11
1
over
seven
turns.
However,
the
design
goal
of
measuring
over
the
beam
position
within
micron
was
not
achieved.
The
goals
were
not
accomplished
due
to
possible
problems
micron
was
not
achieved.
The
goals
were
not
accomplished
due
to
possible
problems
micron was not achieved. The goals were not accomplished due to possible problems
514
in the bandwidth of the output filters. In the next design broadband filters will be used
to help with clock recovery and bunch spacing determination. More care will also be
needed in setting up a trigger such that every BPM signal will be insensitive to timing
jitters in the digitizer lead to the larger position resolution.
Another solution to solve this problem of position resolution is to use a higher
frequency 1428MHz and perform hardware downconversion thus operating at an IF
frequency of 200-MHz. The advantage of this solution is a better signal to noise ratio,
smaller components, and less sensitivity to phase noise in the digitizers.
ACKNOWLEDGMENTS
The authors wish to acknowledge the work that Boni Cordova-Grimaldi and Dave
Anderson did to insure the chassis were manufactured and tested properly. The authors
are extremely appreciative of the ATF collaboration at KEK for allowing the testing of
this device at their facility and especially thankful to the operators that were on staff
during the beam testing of these devices.
515