Bunch-by-bunch feedback and diagnostics in ESRF 2017-05-17 · Bunch-by-bunch feedback and diagnostics in ESRF Demonstration of iGp12 and LNFE D. Teytelman Dimtel, Inc., San Jose,

Bunch-by-bunch feedback and diagnostics in ESRFDemonstration of iGp12 and LNFE

D. Teytelman

Dimtel, Inc., San Jose, CA, USA

April 26, 2017

(Dimtel) BxB feedback and diagnostics in ESRF 2017-04-26 1 / 31

Outline

1 iGp12 introduction

2 Setup and planning

3 Single bunch studies

4 Multibunch measurements at zero chromaticity

5 Studies at nominal chromaticity


iGp12 Highlights

A 500+ MHz processingchannel.Finite Impulse Response (FIR)bunch-by-bunch filtering forfeedback.Control and diagnostics viaEPICS soft IOC on Linux.External triggers, fiducialsynchronization, low-speedADCs/DACs, general-purposedigital I/O.


Front/Back-end Unit

3 front-end channels.1–1.5 GHz front-end detectionfrequency.2-cycle comb generator.1–1.5 GHz back-endfrequency.Integrated control via iGp:

I LO phase shifters;I Attenuators;I Temperature measurement

and stabilization.


Front/Back-end Unit



and stabilization.


Front/Back-end Unit



and stabilization.


Front/Back-end Unit



and stabilization.


Front/Back-end Unit



and stabilization.


Front/Back-end Block Diagram

ΣΣ

iGp GPIODigital controlinterface

Two cycle comb filter MixerVariableattenuator

amplifierTo the power

To the ADC

Low-pass filterAmplifier

Phase shifter 3× frf

2× frf

Bandpass filterAmplifierVariable attenuator

Back-end section

Mixer

×

×

frequency multiplierStep recovery diode

3× multiplier

2× multiplier

Step recovery diodefrequency multiplier

From the DAC

frf

hybridFrom BPM

Front-end section

Front-end channel (×3)


iGp12 Specifications

Design goals:I Reliability;I Maintainability;I Ease of use;I Diagnostics.

FPGA based processing:I Flexible;I Field upgradable.

SpecificationsBunch spacing ≥ 1.9 ns

Harmonic number 32–5120ADC resolution 12 bitsDAC resolution 12 bits

ADC bandwidth 1.35 GHzFeedback filter 32-tap FIRDownsampling 1-256DAQ memory 12 MS

Triggers 2


Waveform Panel

Updates at 2 HzUses data from allbunches over manyturns (12672 forESRF).Four waveforms:

I Mean;I RMS;I Bunch with largest

RMS;I Averaged spectrum

of all bunches.


Waveform Panel




of all bunches.


Waveform Panel




of all bunches.


Waveform Panel




of all bunches.


Feedback in Action

With feedback off wesee vertical oscillation(28.6 dB peak);When feedback isturned on, verticalmotion is suppressedto the noise floor;


Feedback in Action

With feedback off wesee vertical oscillation(28.6 dB peak);When feedback isturned on, verticalmotion is suppressedto the noise floor;


Outline







Day by Day SummarySunday (2017-04-23):

I Unpacked the hardware;I Network and signal connections;I Passive monitoring with ESRF front-end;I Dimtel LNFE setup.

Monday (2017-04-24):I Adjusted timing to the BPM hybrid to optimize common mode

rejection;I Parasitically timed the front-end;I Using very low amplitude single bunch excitation timed the

back-end;I Set up multibunch feedback;I Checked notch monitoring, tune tracking, single bunch transfer

function.Tuesday (2017-04-25):

I MDT shift from 8:00 to 19:00;I Covered 9 out of 10 items in the plan.


















Setup

Baseband processor (iGp12)and 352 MHz low noisefront-end (LNFE);Dimtel BPMH-20-2G hybridprocesses 4 button signals togenerate ∆Y , ∆X , and Σ;Differential DAC output drivestwo power amplifiers directly.


Setup



Setup



The Plan

! Single bunch timing and calibration @ 8 mA, high chromaticity;! Single bunch limits vs. chromaticity, gaps open and closed;

! Feedback operation: zero chromaticity, uniform fill;! Characterization of growth rates;! Bunch cleaning (zero chromaticity);! Parasitic tune monitoring.

! Standard 7/8 fill, high chromaticity! Tune monitoring tools! Bunch cleaning

! Injection studies;% Operate with the current (ESRF) front end.


The Plan






The Plan






The Plan






The Plan






Outline







Calibration

−150 −100 −50 0 50 100 150130

140

150

160

170

180

190

200

210

220

Vertical bump (µm)

AD

C c

ou

nts

pe

r m

A

Vertical calibration factor 0.36 counts/µm/mA

Configured 9 dB of attenuationto handle apparent orbit offset;Calibration factor is0.36 counts/µm/mA;Centering the beam requires a−476 µm bump;At this offset can supportsingle bunch currents up to12 mA;Much higher sensitivities arefeasible with better pickup (β)and orbit offset trimming.


Calibration

−150 −100 −50 0 50 100 150130

140

150

160

170

180

190

200

210

220

Vertical bump (µm)

AD

C c

ou

nts

pe

r m

A




Calibration

−150 −100 −50 0 50 100 150130

140

150

160

170

180

190

200

210

220

Vertical bump (µm)

AD

C c

ou

nts

pe

r m

A




Calibration

−150 −100 −50 0 50 100 150130

140

150

160

170

180

190

200

210

220

Vertical bump (µm)

AD

C c

ou

nts

pe

r m

A




Calibration

−150 −100 −50 0 50 100 150130

140

150

160

170

180

190

200

210

220

Vertical bump (µm)

AD

C c

ou

nts

pe

r m

A




Maximum Single Bunch Current vs. Chromaticity

0 2 4 6 8 10 120

2

4

6

8

10

12

Vertical chromaticity

Ma

xim

um

bu

nch

cu

rre

nt

(mA

)

0 2 4 6 8 10 121

1.5

2

2.5

3

3.5


Ra

tio

Feedback off

Feedback on

A bit of feedback tuning ateach point;Could definitely benefit frommore tuning time;Roughly consistent with theresults from the existingsystem;Modifying feedback duringinjection might help extend thelimit.



0 2 4 6 8 10 120

2

4

6

8

10

12


Ma

xim

um

bu

nch

cu

rre

nt

(mA

)

0 2 4 6 8 10 121

1.5

2

2.5

3

3.5


Ra

tio

Feedback off

Feedback on




0 2 4 6 8 10 120

2

4

6

8

10

12


Ma

xim

um

bu

nch

cu

rre

nt

(mA

)

0 2 4 6 8 10 121

1.5

2

2.5

3

3.5


Ra

tio

Feedback off

Feedback on




0 2 4 6 8 10 120

2

4

6

8

10

12


Ma

xim

um

bu

nch

cu

rre

nt

(mA

)

0 2 4 6 8 10 121

1.5

2

2.5

3

3.5


Ra

tio

Feedback off

Feedback on



Single Bunch Grow/Damp Measurement

0 100 200 300 400 500 600 70020

40

60

80

100

120

140

160

180

Time (µs)

Am

plit

ud

e (

µm

)

Single bunch grow/damp at 2~mA, zero chromaticity

0 100 200 300 400 500 600 700−150

−140

−130

−120

−110

−100

−90

−80

Time (µs)

Ph

ase

(d

eg

ree

s)

Oscillation phase relative to 149~kHz

Zero chromaticity, 2 mA;Feedback is turned off for180 µs;Growth looks linear;Reactive feedback setup —closed loop induces frequencyshift.



0 100 200 300 400 500 600 70020

40

60

80

100

120

140

160

180

Time (µs)

Am

plit

ud

e (

µm

)


0 100 200 300 400 500 600 700−150

−140

−130

−120

−110

−100

−90

−80

Time (µs)

Ph

ase

(d

eg

ree

s)





0 100 200 300 400 500 600 70020

40

60

80

100

120

140

160

180

Time (µs)

Am

plit

ud

e (

µm

)


0 100 200 300 400 500 600 700−150

−140

−130

−120

−110

−100

−90

−80

Time (µs)

Ph

ase

(d

eg

ree

s)





0 100 200 300 400 500 600 70020

40

60

80

100

120

140

160

180

Time (µs)

Am

plit

ud

e (

µm

)


0 100 200 300 400 500 600 700−150

−140

−130

−120

−110

−100

−90

−80

Time (µs)

Ph

ase

(d

eg

ree

s)




Outline







Grow/damp Measurements

Grow/damp at 100 mA, 8 msgrowth time;Only resistive wall modes;Damping rates non-uniform —low frequency response of theamplifier?Fits look very clean, textbookexponential transients;Filled to 200 mA, not limited byinstabilities;Growth time too long, whenfeedback turns on there is notenough gain to damp themotion.









0 5 10 15 20 25 30 35 400

2

4

6

µm

apr2517/123303 Data, Fit and Error for Mode #987

Data

Fit

Error

0 5 10 15 20 25 30 35 400

10

20

µm

Mode #988

Data

Fit

Error

0 5 10 15 20 25 30 35 400

10

20

30

µm

Mode #989

Data

Fit

Error

0 5 10 15 20 25 30 35 400

10

20

30

µm

Mode #990

Data

Fit

Error

0 5 10 15 20 25 30 35 400

100

200

Time (ms)

µm

Mode #991

Data

Fit

Error







0 1 2 3 4 5 60

50

100

150

200

250

300

Time (ms)

Am

plit

ud

e (

µm

)



Spectral Notch Tune Monitoring

130 132 134 136 138 140 14210

−3

10−2

10−1

Frequency (kHz)

Ma

gn

itu

de

(A

DC

co

un

ts)

Zero chromaticity, uniform fill, 100 mA

In closed loop operation,feedback signals show a notchat the betatron frequency;Beam response is resonant atthe tune frequency;Attenuation of detection noiseby the feedback is proportionalto the loop gain;Transfer gain from noise to thefeedback input is 1

1+L(ω)

Maximum attenuation at theresonance, thus a notch.



∑ Error

Transverse position

DisturbancesDetection noise

Feedback Beam


1+L(ω)




∑ Error

Transverse position


Feedback Beam


1+L(ω)




∑ Error

Transverse position


Feedback Beam


1+L(ω)




∑ Error

Transverse position


Feedback Beam


1+L(ω)



Tune Monitoring Example

Tune tracking duringinjection from 10 to200 mA;Done by the iGp12 at2 Hz update rate:

I Exponentialaveraging of 5sweeps (2.5 s timeconstant);

I Minimum search in120–150 kHz range.

Network port changearound −4 minutemark.














All Mode Scan: Technique

0 5 10 15 20 250

0.5

1

1.5

2

2.5

3x 10

4

Time (ms)

Am

plit

ud

e (

arb

. u

nits)

Mode 1

Mode −2

Mode −1

Performed at 15.4 mA (underthe threshold of instability);Each mode is excited to asmall amplitude underfeedback control;In a transient measurementexcitation and feedback areturned off;Capturing 21 ms of beammotion twice a second,16.5 minutes to scan allmodes;27 GiB data set.



0 5 10 15 20 250

0.5

1

1.5

2

2.5

3x 10

4

Time (ms)

Am

plit

ud

e (

arb

. u

nits)

Mode 1

Mode −2

Mode −1




0 5 10 15 20 250

0.5

1

1.5

2

2.5

3x 10

4

Time (ms)

Am

plit

ud

e (

arb

. u

nits)

Mode 1

Mode −2

Mode −1




0 5 10 15 20 250

0.5

1

1.5

2

2.5

3x 10

4

Time (ms)

Am

plit

ud

e (

arb

. u

nits)

Mode 1

Mode −2

Mode −1




0 5 10 15 20 250

0.5

1

1.5

2

2.5

3x 10

4

Time (ms)

Am

plit

ud

e (

arb

. u

nits)

Mode 1

Mode −2

Mode −1



All Mode Scan: Results

0 100 200 300 400 500−250

−200

−150

−100

−50

0

Mode number

Gro

wth

rate

(s−

1)

Upper sideband

Lower sideband

Automated processing extractsgrowth or damping rates;Clear resistive wall signature;A band of higher order modesaround mode −365(129 + N × 352 MHz);A smaller HOM band around−298 (105 + N × 352 MHz);Radiation damping rate118 s−1 (8.5 ms dampingtime).



0 100 200 300 400 500−250

−200

−150

−100

−50

0

Mode number

Gro

wth

rate

(s−

1)

Upper sideband

Lower sideband




0 100 200 300 400 500−250

−200

−150

−100

−50

0

Mode number

Gro

wth

rate

(s−

1)

Upper sideband

Lower sideband




0 100 200 300 400 500−250

−200

−150

−100

−50

0

Mode number

Gro

wth

rate

(s−

1)

Upper sideband

Lower sideband




0 100 200 300 400 500−250

−200

−150

−100

−50

0

Mode number

Gro

wth

rate

(s−

1)

Upper sideband

Lower sideband



Outline







Bunch Cleaning

Started with scraper at 1.5,tested at 3.0, then 4.5;Cleaning works fine at 1.5 and3.0, somewhat touchy at 4.5;Need a bit of parameteroptimization, most likely cansupport reliable cleaning atscraper gap of 3.5–4.5.


Bunch Cleaning



Bunch Cleaning



Single Bunch Transfer Function

120 125 130 135 140 145 150 155−70

−65

−60

−55

−50

−45

−40

Frequency (kHz)

Ma

gn

itu

de

(d

B)

Beam Transfer Function

120 125 130 135 140 145 150 155−200

−150

−100

−50

0

50

100

Frequency (kHz)

Ph

ase

(d

eg

)

Turn off feedback for bunch 40;Apply swept sinusoidalexcitation;Measure beam transferfunction;A simple-minded fit of aresonant response;Fit a linear combination of 3resonances;5 resonances;7 resonances;9 resonances;11 resonances.



120 125 130 135 140 145 150 155−70

−65

−60

−55

−50

−45

−40

Frequency (kHz)

Ma

gn

itu

de

(d

B)

Center frequency=136.01 kHz, τ=0.1 ms

120 125 130 135 140 145 150 155−200

−150

−100

−50

0

50

100

Frequency (kHz)

Ph

ase

(d

eg

)




120 125 130 135 140 145 150 155−70

−65

−60

−55

−50

−45

−40

Frequency (kHz)

Ma

gn

itu

de

(d

B)


120 125 130 135 140 145 150 155−200

−150

−100

−50

0

50

100

Frequency (kHz)

Ph

ase

(d

eg

)

∆f=[−1.5 1.7] kHz; τ=[0.3 0.1] ms




120 125 130 135 140 145 150 155−70

−65

−60

−55

−50

−45

−40

Frequency (kHz)

Ma

gn

itu

de

(d

B)


120 125 130 135 140 145 150 155−200

−150

−100

−50

0

50

100

Frequency (kHz)

Ph

ase

(d

eg

)

∆f=[−1.8 1.8 −3.6 3.3] kHz; τ=[0.5 0.4 0.5 0.2] ms




120 125 130 135 140 145 150 155−70

−65

−60

−55

−50

−45

−40

Frequency (kHz)

Ma

gn

itu

de

(d

B)


120 125 130 135 140 145 150 155−200

−150

−100

−50

0

50

100

Frequency (kHz)

Ph

ase

(d

eg

)

∆f=[−1.8 1.9 −3.7 3.6 −5.7 5.4] kHz; τ=[0.7 0.6 0.5 0.6 0.3 0.2] ms




120 125 130 135 140 145 150 155−70

−65

−60

−55

−50

−45

−40

Frequency (kHz)

Ma

gn

itu

de

(d

B)


120 125 130 135 140 145 150 155−200

−150

−100

−50

0

50

100

Frequency (kHz)

Ph

ase

(d

eg

)

∆f=[−1.8 1.9 −3.7 3.6 −5.8 5.5 −7.8 7.3] kHz; τ=[0.7 0.7 0.7 0.5 0.4 0.5 0.2 0.2] ms




120 125 130 135 140 145 150 155−70

−65

−60

−55

−50

−45

−40

Frequency (kHz)

Ma

gn

itu

de

(d

B)


120 125 130 135 140 145 150 155−200

−150

−100

−50

0

50

100

Frequency (kHz)

Ph

ase

(d

eg

)

∆f=[−1.8 1.8 −3.7 3.5 −5.9 5.4 −7.9 7.4] kHz; τ=[0.9 0.6 0.7 0.6 0.4 0.5 0.2 0.7] ms



Single Bunch Phase Tracking

CORDIC

Aφ → IQ

A

φ

sin

cosz−1

Integrator andrange limiter

I

Q

CORDIC

IQ → Aφ

A

φ

Beam

Full scale

Phase accumulator

Beam excitation

Drive frequency

Drive amplitudeDDS-based sinusoidal drive generator

Drive frequency modulation

Phase shift setpoint

I

Q

sincos

DDC+CICBeam response

A single bunch is excited with a sinusoidal excitation at lowamplitude (20–40 µm);Response is detected relative to the excitation to determine thephase shiftIn closed loop, phase tracker adjusts the excitation frequency tomaintain the correct phase shift value;Adjustable integration time, tracking range, loop gain.


Phase Tracking and Chromaticity Measurement

Close the loop at −45 s;Slow settling at low gain,faster as the gain israised;Chromaticity scan with10 Hz steps;Return step of 40 Hz istoo large, tracker locks atthe wrong point;Open and close the loopto re-lock.














Fast Phase Tracking

Time (turns)

Fra

ctio

na

l fr

eq

ue

ncy

Fast tune tracking (200 turns decimation), 4096 turn FFT

1 2 3 4 5 6 7 8 9

x 104

0.38

0.381

0.382

0.383

0.384

0.385

0.386

0.387

0.388

0.389

0.39

Decimation factor in phasetracker controls trackingbandwidth;200 turns decimaton, 1.77 kHzmeasurement bandwidth;180 Hz closed loop trackingbandwidth;Use time-domaindownconversion to betterresolve tune modulation;Spectrum shows lines at 10and 50 hertz.


Fast Phase Tracking

Time (turns)

Fra

ctio

na

l fr

eq

ue

ncy


1 2 3 4 5 6 7 8 9

x 104

0.38

0.381

0.382

0.383

0.384

0.385

0.386

0.387

0.388

0.389

0.39



Fast Phase Tracking

Time (turns)

Fra

ctio

na

l fr

eq

ue

ncy


1 2 3 4 5 6 7 8 9

x 104

0.38

0.381

0.382

0.383

0.384

0.385

0.386

0.387

0.388

0.389

0.39



Fast Phase Tracking

0 50 100 150 200 250 3000.8

0.9

1

1.1

1.2

Time (ms)

Am

plit

ude (

AD

C c

ounts

)

Fast tune tracking: amplitude

0 50 100 150 200 250 300−600

−400

−200

0

200

Time (ms)

Phase (

deg)

Phase

0 50 100 150 200 250 3000.3826

0.3828

0.383

0.3832

0.3834

0.3836

Time (ms)

Fra

ctional fr

equency

Tracking frequency vs. time



Fast Phase Tracking

100

101

102

103

10−3

10−2

10−1

100

101

102

103

104

105

Frequency (Hz)

Ma

gn

itu

de

(a

rb.

un

its)

FFT of tune frequency signal



Tune Variation

0 50 100 150 200 250 3000.8

0.9

1

1.1

1.2

Time (ms)

Am

plit

ud

e (

AD

C c

ou

nts

)

Fast tune tracking: amplitude

0 50 100 150 200 250 300−600

−400

−200

0

200

Time (ms)

Ph

ase

(d

eg

)

Phase

0 50 100 150 200 250 3000.3826

0.3828

0.383

0.3832

0.3834

0.3836

Time (ms)

Fra

ctio

na

l fr

eq

ue

ncy

Tracking frequency vs. time

Tune moves around by 5 × 10−4;Fitting complex phase spacetrajectories fails due to tunemodulation;Modal scan runs at 2 Hz andaliases the modulation;Grow/damp transient shows tuneshifts around 100 Hz (2.8 × 10−4)at 100 mA;At 15 mA expect only 4.2 × 10−5,completely obscured by thisbaseline modulation.


Tune Variation

0 5 10 15 20 252.2

2.25

2.3

2.35

2.4

2.45

2.5

2.55x 10

4

Time (ms)

Magnitude (

arb

. units)

Mode −1

0 5 10 15 20 25120

140

160

180

200

220

Time (ms)

Phase (

degre

es)



Tune Variation

0 200 400 600 800 10000.3885

0.3886

0.3887

0.3888

0.3889

0.389

0.3891

Mode number

Fra

ctional tu

ne



Tune Variation



Tune Variation

0 200 400 600 800 10000.3885

0.3886

0.3887

0.3888

0.3889

0.389

0.3891

Mode number

Fra

ctional tu

ne



Injection Transient

−500 0 500−250

−200

−150

−100

−50

0

50

100

150

Bunch number (0 − single)

∆Y

(µ

m)

First turn displacements during injection

Before

After

Captured beam transient dueto the injection kickers firing;Automatic extraction of thedifference orbit;Converted to physical unitsusing bunch-by-bunchcurrents and measuredcalibration factor.


Injection Transient

−500 0 500−250

−200

−150

−100

−50

0

50

100

150


∆Y

(µ

m)


Before

After



Injection Transient

−500 0 500−250

−200

−150

−100

−50

0

50

100

150


∆Y

(µ

m)


Before

After



Summary

Successfully operated Dimtel bunch-by-bunch system in theESRF;Many diagnostic features have been demonstrated;In the demo setup feedback pushes vertical emittance from7.7 to 8 pm.With some balancing and optimization much lower noise floor iseasily achievable.I’d like to thank everyone who helped to make this a successfultest!


Summary



Summary



Summary



Summary



Bunch-by-bunch feedback and diagnostics in ESRF 2017-05-17 · Bunch-by-bunch feedback and diagnostics in ESRF Demonstration of iGp12 and LNFE D. Teytelman Dimtel, Inc., San Jose,

Documents