font: keeping the lc collidingscipp.ucsc.edu/seminars/experimental/files/seminar... · 2004. 2....

FONT: Keeping the LC Colliding

(a beam based feedback system for the Linear Collider)

SCIPP Seminar 26th Jan ’04

Gavin NesomUC Santa Cruz

Gavin Nesom, SCIPP SEMINAR 26/01/04

• Daresbury Laboratory:Alexander Kalinine, Roy Barlow (elec. eng.), Mike Dufau (des.) Susan Smith, Rob Smith, Mike Dykes, Mike Poole

• Oxford:Colin Perry (elec. eng.) + technicians Gerald Myatt (retd. faculty) Simon Jolly, Gavin Nesom (grad students emeritii)

• Queen Mary:Philip Burrows (faculty), Glen White (RA), Tony Hartin (prog.)Stephen Molloy, Shah Hussain (grad. students)

• SLAC:Joe Frisch, Tom Markiewicz, Marc Ross, Chris Adolphsen, Keith Jobe, Doug McCormick, Janice Nelson, Tonee Smith, Mark Woodley + technical support

FONT Group


Talk Outline

• Why do we need beam feedback?• NLCTA; layout, and beam conditions

• FONT Hardware– Magnets, Beam Postion Monitor– Electronics– Feedback schematic

• Font1 Results• Font2; Upgrades, Additions, and Results• Font3 goes to ATF?

Thanks to Phil Burrows / Steve Molloy for a large fraction of the slide contents


Vibrations and Beam Instability

Colliding beams of NLC will not be stable– Component misalignment– Ground Motion– Cultural Noise

Correct using series feedback systems(Joe Frisch, Tom Himel, Andrei Seryi, Steve Smith, and others)

– Slow: drift issues– Medium: SLC like inter-bunch corrections – Mechanical: Inertial stabilisation– Intra-train: FONT; nanosecond timescale


Luminosity Loss at the I.P

• Relative offsets in final Quads due to fast ground motion leads to beam offsets of several σy (3 nm for NLC-H 500 GeV).

• Correct using beam-based feedback system near IP.


The Need for Fast Feedback


Measuring the Offset

• 50% drop in luminosity ⇒ 4σ offset ⇒ 100 µrad kick


Beam-based Feedback (FONT)

Intra-train beam feedback is last line of defence against ground motion

Key components:• Beam position monitor (BPM)• Signal processor• Fast driver amplifier • E.M. kicker• Fast feedback circuit


Requirements of the Fast Feedback

Daniel Schulte, Steve Smith (and others) have considered the problem: LCC-26 & LCC-56

Recover significant amount of lost luminosity:– sensitivity to nm variations in beam-beam offset

Correct offset within a small fraction of 266 ns bunch train

Dominant time factor should be distance to IP, NOT speed of feedback. Too fast for digital electronics.


FONT Luminosity Recovery (NLC ‘H’)

For small offsets( 80% of design luminosity

Much easier (and required) at TESLA: 2820 bunches X 337 ns


NLC - NLCTA Comparison

~1800190Bunches/train

60 MeV250 GeVBeam Energy

~170 ns266 nsTrain length

0.08ns1.4nsBunch spacing

500µm3nmBeam size

NLCTANLC


Charge Variation and Beam Size

(NLC design charge

variation


Why implement FONT at NLCTA?

• Timing issue is machine independent⇒ check design electronics acts within required time

• NLCTA has charge variation along the pulse⇒ system will work with flat beam if it copes with variation

• Test different feedback algorithms and other emerging issues

Using NLCTA for FONT Tests


FONT Beam Test on the NLCTA

Colliding beams Single beam

BPM

Kicker

BPMProcessor

KickerAmplifier

Round-tripDelay

(short cable delays, long beam flight time)

(short beam flight trip time, long cable delays)

NLC NLCTA

4.5m


NLCTA Layout

Gun Chicane Structures Beam DumpFONT region


FONT: beamline


NLCTA FONT Installation

Magnet assembly and X-band BPM installed onto NLCTA downstream of RF structures.

Feedback loop

Beam direction


FONT: kicker / magnet assembly

SLC dipole

and

post-damping

ring kicker


Magnet Assembly

Kicker

Dipole


FONT1: kicker driver amplifier

3kW amplifier:

Planar triode tubes

7.5 A, 350V

• ±5V input, ~1 kV output• Delay time = 10ns• 200ns usable output • Move 65Mev beam ±1mm


X-band Resonant Button BPM

• Dubbed a “Resonant Button” BPM• Initial readout w. diode detectors

• Pickups tuned to 11.424 GHz• Electric feedthrough• Standard Vacuum Parts


BPM Response

Estimate Q = 10– Short pulse response: time constant = 1.2ns– Network analyser measurement (also shows cavity resonance)


FONT1: BPM processor

Read each y pickoff signal:

Formed sum and difference,

mixed down from X-band to

baseband.

Charge normalisation:

1/sum performed w. AWG

(slow) with real-time

first-order correction


BPM RF Components

Attenuator

Mixer

X-Band Ref.

Difference Signal

Sum Signal

Isolator180 phase delay

Isolator

Mixer


Signal Processing

• Need to de-convolve charge contribution from the BPM signal

• Have sum signal, q, and inverted signal of a previous pulse, 1/q`.

• Use 1/q` to normalise the position signal, and use q to add a second order correction : y/q` (1- dq/q`) = y/q` (2 - q/q`)

• 3 multiplier chips are required for this scheme

1/q`y

2X

X X

+-q

1/q`

2- q/q`

y / q`y/q` (2 – q /q`)


Charge Normalisation


BPM Calibrations

• BPM measurement resolution ~100um good enough for FONT


A Promising Start


FONT1: Charge normalisation/feedback

12

34


Kicker and Feedback Circuit

• The feedback must take account of previous correction

• The circuit design takes care of this with a delay line

y signal gain control Amp.

Delay loop

sum

Beam


Feedback Loop


FONT1: Schematic Overview (recap)

BPM Electronics

Mix to D.C Charge

Norm.

Gain Control Feedback

Circuit

Kicker

Amplifier

Total Diff.

Beam


FONT1: expected latency

• Time of flight kicker – BPM: 14ns• Signal return time BPM – kicker: 18ns

Irreducible latency: 32ns

• BPM cables + processor: 5ns• Preamplifier: 5ns• Charge normalisation/FB circuit: 11ns• Amplifier: 10ns• Kicker fill time: 2ns

Electronics latency: 33ns

• Total latency expected: 65ns


FONT1: results

10/1 position correction of 65 MeV e- beam

achievedlatency of 67 ns

system tested in feed forward and feedback modes


FONT2: outline

Goals of improved FONT2 setup:

• Additional 2 BPMs: independent position monitoring• Second kicker added: allows solid state amplifiers• Shorter distance between kickers and FB BPM:

irreducible latency now c. 16 ns• Improved BPM processor:

real-time charge normalisation using log amps (slow)• Expect total latency c. 53 ns:

allows 170/53 = 3.2 passes through system• Added ‘beam flattener’ to remove static beam profile• Automated DAQ including digitisers and dipole control


FONT2: expected latency

• Time of flight kicker – BPM: 6ns• Signal return time BPM – kicker: 10ns

Irreducible latency: 16ns

• BPM processor: 18ns• FB circuit: 4ns• Amplifier: 12ns• Kicker fill time: 3ns

Electronics latency: 37ns

• Total latency expected: 53ns


FONT2: new beamline configuration

Dipole and kickers

New BPMs


FONT2: new front-end IF processor


FONT2: new front-end IF processor

2y on beamline: 6y+6x outside tunnel:

14 channels: 2y on beamline, 6y + 6x outside


FONT2: synchronous demodulator PCBs

6 boards outside tunnel:

(y1-y2)/(y1+y2) w. log amps


FONT2: new solid state amplifiers

Total drive same as last year

Amplifier pair w. shielding:

FB signal into amp (0.9c):


FONT2 run Nov/Dec 2003

• Nov 7-9: commissioned 3 BPMs and new electronicsresolution measured using ‘triplets’: 15 microns

• Opportunistic runs Nov 24 – Dec 13 (10 shifts)operating conditions difficult due to 8-pack + high gradient structure tests

• Dec 3: commissioned amplifiers, kicker, FB circuit + DAQ. Full system run with and with feedback

• Dec 9: commissioned beam flattener (w/o feedback)

• Dec 13: ran full system with beam flattener


FONT2: first results Nov 2003

All 3 BPMs see beam w. dipole scans

σy =15 microns

BPM1

BPM2

BPM3

charge

Time (ns)


FONT2 initial results: beam flattener

Performance

bandwidth limited:

80 MHz (AWG)

30 MHz (amp)

Flattener corrects

to average beam

position: removes

‘static’ structure

Feedback off for illustration:


FONT2: Latest Results


Ideas for further development work

• e+e- background studies in SLAC A-line

• Investigate other feedback systems

• World’s smallest emittance e- beam is at KEK/ATF– Scaling: 1 µm at ATF (1 GeV) ~ 1 nm at LC (1000 GeV)– Beam-based feedback could be scale model for LC


Development of Improved Feedbacks

• Beam angle-jitter:

correction best done near IP with RF crab cavity(needed anyway): system needs design + prototyping

• Ideally, feedback on luminosity:

bunch-by-bunch luminosity measurement would allow intra-train luminosity feedback


Comparison of ATF with NLCTA

NLCTA ATF

Train length 170 ns 300 nsBunch spacing 0.08 ns 2.8 nsBeam size (y) 500 µm 5 µmJitter (y) 100 µm 1 µmBeam energy 65 MeV 1.3 GeV

ATF has ‘right’ bunch spacing and train length, andthe beam is smaller and more stable than at NLCTAØ much better place for fast feedback prototypes


Possible future developments for FONT at ATF

3 suggestions:

1. Stabilisation of extracted bunchtrain at 1 micron level:low-power (< 100W), high stability amplifierstripline BPM w. ~ 1 micron resolution

these are exactly what are needed for the LC!

2. Stabilisation of extracted bunchtrain at 100 nm level:requires special BPM and signal processing

useful for nanoBPM project

3. Test of intra-train beam-beam scanning system:high-stability ramped kicker drive amplifier

very useful for LC


Closing Thoughts

FONT 1&2 have provided a successful proof of principal• Close loop on beam based feedback in

font: keeping the lc collidingscipp.ucsc.edu/seminars/experimental/files/seminar... · 2004. 2....

Documents