Performance of CW Superconducting Cavity at ERL test Facility · Performance of CW Superconducting Cavity at ERL test Facility Feng QIU (KEK) LLRF 2013, F. QIU . 2 Main Content ...

1

Performance of CW Superconducting

Cavity at ERL test Facility

Feng QIU (KEK)

LLRF 2013, F. QIU

2

Main Content

LLRF 2013, F. QIU

Introduction

High Power Level RF system

Low Level RF system

Gain Scanning

Performance

Future Plan

Summary

3

Introduction

LLRF (2013), F. QIU

Compact ERL (cERL) is under construction as a test facility for the future 3-GeV

ERL project. It is a superconducting system and is operated in CW mode.

HLRF

3

Commissioning

of injector

started in April,

2013.

ERL Development Building

Construction Site@KEK

PF

PF-AR

Injector

Injector consists of 4 cavities: Buncher (NC),

Injector 1 (SC), Injector 2 (SC), Injector 3 (SC).

ML includes two 9-cell cavities (SC).

Layout of cERL

Two 9-cell

in ML (SC)

At present, the injector was constructed and the first beam commissioning was

performed from April to June of this year at cERL. The construction of the Main linac

would be carried out by October, 2013.

Entrance

HLRF (Power Source) At present, total 3 kinds of Power Sources are applied in cERL : 20-KW IOT, 25-

KW Klystron and 300 KW Klystron.

The 1.3 GHz 2-cell cavities are used as superconductive cavities.

LLRF (2013), F. QIU

0.1 % RMS, 0.1 deg. RMS for cERL

0.01%rms,0.01deg.rms for 3GeV-ERL

RF requirement

Buncher Injector1 Injector2 Injector3

Vector-sum

Controlling

300 kW Kly.25 kW Kly.

20-kW IOT 25-kW Kly. 300-kW Kly.

2-cell SC

IOT

1.3 GHz RF

Phase Shifter

0.4 MeV

~ 5 MeV

7 MV/m for Injector Cavities

ML

16 kW SSA 30 kW IOT 9-cell SC

Two 9-cell Cav. (SC)

NC SC SC SC

Dump

Will start operation at this Oct.

4

HLRF (Power Distribution System)

LLRF (2013), F. QIU

Rather narrow

space and

complicated

waveguide.

Vector-sum controlling

for Inj.2 and Inj. 3.

Outside shield

Inside shield

5

6

Low Level RF System

FPGA

Down-convertor

IQ Mod.

FPGA boards FPGA boards

Thermostatic Chamber

LLRF Cabinet

ADC: LTC 2208

Xillinx Virtex5 FPGA

ADC & DAC Interface

Digital I/O

uTCA Digital Board

Digital Board type Feature

ADC LTC2208 16 bits, 130 MHz (Max.)

DAC AD9783 16 bits, 500 MHz (Max.)

FPGA Virtex 5 FX 550 MHz (Max.), includes a Power PC with

Linux, EPICS is installed on the Linux.

16 bits, 130 MHz

LLRF (2013), F. QIU

7

Low Level RF System

EPCIS (Experimental Physics and

Industrial Control System) is installed

inside Micro TCA and is used as the DAQ

(data acquisition) system.

CSS (Control System Studio) is in charge

of the user interface programming.

We use Piezo+Motor for the tuner

controlling .

CSS Interface (user interface)

LLRF (2013), F. QIU

Schematic Diagram

Pr Pf

RF switch

Interlock

Klystron

PreAmp

IQ M

od

ulato

r

I

Q

baseband

1300 MHz

LO 1310MHz

16bit 80MS/s

MO 1300 MHz

Digital feedback board DIO(RF OFF)

IF =

10

MH

z (

MO

/12

8)

Do

wn

con

ver

tor

IQ

Correction

ADCIQ

Vector Sum

I

Q

+-

I_Set Table

-+

Q_Set Table

Delay

++

I_FF Table

Delay

++

Q_FF Table

I

Q

IQ

IQ

ADC

Correction

DAC

DAC

Limit

cossin

sincosA

ADC

ADC

DAC

DAC

FPGA

Power PC

EPICS IOC

CLK

Gb Ethernet

IQ

modula

tor

CAV2

1300MHz

Wave Forms Set Parameters

Linux

80MHz

IF 10 MHz I

Q

AMC Digital Feedback Card

DIO (RF OFF)

LTC2208AD9783

Dow

n

Convert

or

Klystron

Digital Filter

Digital Filter

)(cos)(sin

)(sin)(cos)(

tt

tttA

I

Q

CAV1

CAV2

1300 MHz

LO=1310 MHzAnalog filter

Amp.

P I

P IDigital Filter

Digital Filter

Xillinx Virtex 5 FPGA

Low Level RF System

1),1()1()()(: nynxnyIIR

FFPIController :

systemEPICS

LLRF (2013), F. QIU 8

9

Gain scanning (Definition of Gain) Gain-scanning: Scanning different proportional gain KP and integral Gain KI to

find out the optimal gains.

The scanning experiment was carried out at low RF field.

Definition of the KP and KI

Suppose on resonance

∫KI

KP

Cavity

FF

error

Delay

Disturbance

Δω

s

KIKPsK )(

sTes

sH

5.0

5.0)(

Set value

DACADC

Gain from ADC

to DAC (analog

part) is

normalized to be

0 dB, remaining

gains is only

contributed by

PI controller

LLRF (2013), F. QIU

10

Gain Scanning (Delay measurement) In order to acquire some priori information about the maximum gain, we have evaluated

the loop delay at first due to there is a relationship between the loop delay and the

maximum gains.

Loop delay is measured by exciting the system with square wave in the DAC output.

About 1 us loop delay

DAC ADC

Gain margin for Inj .1 ( suppose delay = 1 µs )

Inj .2 &3

Inj .1

LLRF (2013), F. QIU

Simulation Gain margin: Inj1=230, Inj. 2&3=100

Loop delay Inj. 1 Gain Margin

Operated in the pulse mode

Bondary

Bondary

11

Gain scanning (Critical gains) The Critical gain has measured by the KI=0, KP Scanning.

If the proportional gain is larger than the critical gain, the loop would be oscillated.

The simulation and the

measurement are in

agreement very well.

Stb Bun. Inj. 1 Inj. 2 Inj. 3

QL 1.1e4 1.2e6 5.8e5 4.8e5

f0.5 [kHz] 58 0.54 1.12 1.35

Inj. 1: Critical Gain=250

Inj. 2&3: Critical Gain=90

Gain Inj. 1 Inj. 2&3

Gain Margin (Sim. ) 230 100

Critical Gain (Meas. ) 250 90

Meas.

LLRF (2013), F. QIU

12

Gain scanning (Buncher)

High gain is not available for Buncher cavity (NC) due to its large bandwidth

(QL=1.1e4) .

Opt. KP

Opt. KI

Optimal Gains:

KPopt=0.8, KIopt=1.2e5.

Sensitive to KI

ΔA/A [%] Δθ [deg.]

ΔA/A [%] Δθ [deg.]

It is clear to see that the integral Gain

KI is dominant gain for the Buncher

cavity because of the limitation of high

proportional gain KP (not available ).


QL 1.1e4 1.2e6 5.8e5 4.8e5

f0.5 [kHz] 58 0.54 1.12 1.35

KI=const. KP scanning

KP=const. KI scanning

LLRF (2013), F. QIU

13

Gain scanning (Inj. 1)

KI=const. KP scanning

KP= const., KI scanning

The dominant gain in Inj. 1 is

proportional gain (KP), very

common in SC cavity controlling.

High gain controlling can be

realized due to its narrow

bandwidth.

Opt. KP

QL=1.2e6

Sensitive to KP

ΔA/A [%]

ΔA/A [%]

Δθ [deg.]

Δθ [deg.]

Opt. KI

Optimal Gains:

KPopt=84, KIopt=1.0e5.

High gain is available for Inj .1 cavities (SC).


QL 1.1e4 1.2e6 5.8e5 4.8e5

f0.5 [kHz] 58 0.54 1.12 1.35

LLRF (2013), F. QIU

14

Gain scanning (Inj. 2&3)

Both KI and KP have

an effect for Inj. 2&3.

KI is also significant

due to there is an 300 Hz

component in the HVPS.

KI=const., KP scanning

KP= const., KI scanning

Opt. KI

ΔA/A [%]

ΔA/A [%]

Δθ [deg.]

Δθ [deg.]

Opt. KP Optimal Gains:

KPopt=41, KIopt=1.1e5.


QL 1.1e4 1.2e6 5.8e5 4.8e5

f0.5 [kHz] 58 0.54 1.12 1.35

High gain is available for Inj .2 (SC) and Inj .3 (SC).

LLRF (2013), F. QIU

15

Gain scanning (Conclusion) Conclusions:

The proportional gain KP plays

an much more important roles in

SC cavity and it is usually located

in the ½ to ⅓ of the critical gains.

The integral gain KI is

significant in NC cavity due to the

limitation of the critical gains.

Inj .1 Inj .2 & 3

The performance would be best in the optimal

gain case.

The amplitude and phase stability of Inj. 1

and Inj. 2&3 can be 0.01% RMS and 0.02 deg.

RMS, respectively.

Gain Bun. Inj. 1 Inj. 2 Inj .3

Prop. Gain (KP) 0.8 84 41

Int. Gain (KI) 1.2e5 1.0e5 1.1e5

Critical Gain 3 230 90

f0.5 [kHz] 58 0.54 1.12 1.35

Vector-sum

About ½ to ⅓

Comparison btw the optimal gains and other gains.

LLRF (2013), F. QIU

0 1000 2000 3000 40001.555

1.56

1.565

1.57

1.575x 10

4

Channel

Am

plitude

ADC1: 0.14111%rms

0 1000 2000 3000 4000-107.5

-107

-106.5

-106

-105.5

-105

-104.5

-104

Channel

Phas

e(d

eg.

)

ADC1: 0.51122deg.rms

0 1000 2000 3000 40001.568

1.57

1.572

1.574

1.576

1.578

1.58

1.582x 10

4

Channel

Am

plitude

ADC2: 0.11541%rms

0 1000 2000 3000 4000-95.5

-95

-94.5

-94

-93.5

-93

-92.5

Channel

Phas

e(d

eg.

)


0 1000 2000 3000 40001.03

1.035

1.04

1.045

1.05

1.055x 10

4

Channel

Am

plitude

ADC3: 0.28656%rms

0 1000 2000 3000 40004

5

6

7

8

9

Channel

Phas

e(d

eg.

)


0 1000 2000 3000 40001.024

1.025

1.026

1.027

1.028

1.029

1.03x 10

4

Channel

Am

plitude

ADC4: 0.07367%rms

0 1000 2000 3000 4000

97.9

98

98.1

98.2

98.3

Channel

Phas

e(d

eg.

)


0 1000 2000 3000 40001.5592

1.5594

1.5596

1.5598

1.56

1.5602

1.5604

1.5606x 10

4

Channel

Am

plitude

FIL: 0.011756%rms

0 1000 2000 3000 4000-100.05

-100

-99.95

-99.9

-99.85

Channel

Phas

e(d

eg.

)

FIL: 0.022096deg.rms

0 1000 2000 3000 40001.76

1.78

1.8

1.82

1.84

1.86

1.88

1.9x 10

4

Channel

Am

plitude

DAC Amplitude

0 1000 2000 3000 400045

50

55

60

65

Channel

Phas

e(d

eg.

)

DAC Phase

FB2-P6000I200-7MV-rate799-WL130k-20130614-1.csv

16

0 1000 2000 3000 40001.097

1.098

1.099

1.1

1.101

1.102

1.103

1.104x 10

4

Channel

Am

plitude

ADC1: 0.080143%rms

0 1000 2000 3000 4000-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

Channel

Phas

e(d

eg.

)


0 1000 2000 3000 40001.545

1.55

1.555x 10

4

Channel

Am

plitude

ADC2: 0.084565%rms

0 1000 2000 3000 40000.8

0.9

1

1.1

1.2

1.3

1.4

1.5

Channel

Phas

e(d

eg.

)


0 1000 2000 3000 40009720

9740

9760

9780

9800

9820

Channel

Am

plitude

ADC3: 0.10138%rms

0 1000 2000 3000 4000-98.2

-98.1

-98

-97.9

-97.8

-97.7

-97.6

Channel

Phas

e(d

eg.

)


0 1000 2000 3000 40001.34

1.341

1.342

1.343

1.344

1.345

1.346

1.347x 10

4

Channel

Am

plitude

ADC4: 0.059251%rms

0 1000 2000 3000 4000128.6

128.7

128.8

128.9

129

129.1

Channel

Phas

e(d

eg.

)


0 1000 2000 3000 40001.0985

1.099

1.0995

1.1

1.1005

1.101

1.1015

1.102x 10

4

Channel

Am

plitude

FIL: 0.046782%rms

0 1000 2000 3000 4000-0.2

-0.1

0

0.1

0.2

Channel

Phas

e(d

eg.

)

FIL: 0.061938deg.rms

0 1000 2000 3000 40001.44

1.445

1.45

1.455

1.46

1.465

1.47x 10

4

Channel

Am

plitude

DAC Amplitude

0 1000 2000 3000 4000-0.5

0

0.5

1

1.5

2

2.5

Channel

Phas

e(d

eg.

)

DAC Phase

FBon-2013-0531-FB0-P20I40-IIR500k-deci799-1.csv

0 1000 2000 3000 40001.568

1.569

1.57

1.571

1.572

1.573

1.574x 10

4

Channel

Am

plitude

ADC1: 0.049341%rms

0 1000 2000 3000 400052.8

52.9

53

53.1

53.2

Channel

Phas

e(d

eg.

)


0 1000 2000 3000 40001.52

1.54

1.56

1.58

1.6

1.62x 10

4

Channel

Am

plitude

ADC2: 0.63453%rms

0 1000 2000 3000 400070

75

80

85

Channel

Phas

e(d

eg.

)

ADC2: 2.0584deg.rms

0 1000 2000 3000 40001.46

1.48

1.5

1.52

1.54

1.56x 10

4

Channel

Am

plitude

ADC3: 0.64748%rms

0 1000 2000 3000 400072

74

76

78

80

82

84

86

Channel

Phas

e(d

eg.

)

ADC3: 2.0586deg.rms

0 1000 2000 3000 40001.353

1.354

1.355

1.356

1.357

1.358

1.359

1.36x 10

4

Channel

Am

plitude

ADC4: 0.054301%rms

0 1000 2000 3000 4000151.8

151.9

152

152.1

152.2

152.3

Channel

Phas

e(d

eg.

)


0 1000 2000 3000 40001.5694

1.5696

1.5698

1.57

1.5702

1.5704

1.5706

1.5708x 10

4

Channel

Am

plitude

FIL: 0.010388%rms

0 1000 2000 3000 400052.9

52.95

53

53.05

53.1

Channel

Phas

e(d

eg.

)

FIL: 0.01807deg.rms

0 1000 2000 3000 40002.05

2.1

2.15

2.2

2.25x 10

4

Channel

Am

plitude

DAC Amplitude

0 1000 2000 3000 40000

2

4

6

8

10

12

14

Channel

Phas

e(d

eg.

)

DAC Phase

FB1-P13500I200-WL130k-deci799-20130620-21h48m-3.csv

FB0 (Buncher)

FB1 (Inj. 1)

Amp: 0.01% rms , Phase 0.02 deg. rms

Amp 0.05% rms, Phase 0.06 deg. rms

Performance (RF field) FB2 (Vector-sum of Inj. 2 and Inj. 3)

Amp: 0.01% rms, Phase 0.022 deg. rms

ΔA/A

[RMS]

Δφ

[RMS]

Loop

Delay

Bun. 0.05% 0.06 deg. 1.1 μs

Inj. 1 0.01% 0.02 deg. 1.1 μs

Inj. 2&3 0.01% 0.02 deg. 1.1 μs

Our Goal:

0.1% for amplitude and 0.1 deg. for phase.

300 Hz fluc. from HV

Power Supply

No 300 Hz fluc.

Selections of KP and KI are based on the gain scanning experiment!

LLRF (2013), F. QIU

100 kS/s

100 kS/s

17

Performance (Screen Monitor) The beam momentum is measured by screen monitor and determined by the peak

point of the projection of the screen.

Momentum was determined by the peak point

of the projection of the screen.

Dispersion @ screen

monitor = 0.82m

Resolution = 53.4 m/pixel

(P/P=6.5e-5)

Screen Monitor

Attention: Vector-sum error would influence

the beam momentum jitter greatly! Thus the

phase error btw inj. 2 and inj. 3 should be

optimized at first!

LLRF (2013), F. QIU

Momentum Jitter= 0.006% rms

Result of the screen monitor

Beam Energy 0 50 100 150 200 250 300 350 400 450 500

-0.3

-0.2

-0.1

0

0.1

0.2

time(s)

dP/P

(%)

FB2 LG:dP/P= 0.063607% rms HG:dP/P=0.0056858% rms

FB1HG&FB2LGFB1HG&FB2HG

-0.4 -0.2 0 0.2 0.40

20

40

60

80

Num

ber

dP/P (%)

FB2LG: dP/P= 0.063607% rms

-0.02 -0.01 0 0.01 0.020

100

200

300

400

500

Num

ber

dP/P (%)

FB2HG: dP/P= 0.0056858% rms

Performance (Beam energy)

18 LLRF (2013), F. QIU

19

Future Plan Whole cERL operation would be carried out in October of 2013. At that moment, we

will operate the main Linac 9-cell cavity.

There are total 9 modes in the 9-cell cavity, only the is selected as the accelerating

mode .The 8/9 mode which is closed to mode should be suppressed (otherwise, it

would influence the system stability).

The frequency of 8/9 mode of our cavity is about 1.6 MHz, we should remove it by

digital filter, the current 1st order IIR filter would only has about 24 dB suppression at 1.6

MHz with100 kHz bandwidth.

1,)1(1

)(:

),1()1()()(:

jejHFrequency

nynxnyTime

Obviously, the current 1st IIR

filter is not competent!

We should select other IIR

filters.

LLRF (2013), F. QIU

20

Future Plan Possible solutions:

a)Increasing the order of the filter (connecting the 1st order low pass IIR filter in

series).

b)Application of the notch filter (band stop filter).

IIR LP +IIR LP

IIR LP + Notch

Which would be better!

We will compare.

Notch

Step response shows the Notch

filter is attractive choice with

same bandwidth (100 kHz) Step response of “Notch+LP” and “LP+LP”

LP LP

LP

All with 100 kHz BW

LLRF (2013), F. QIU

21

Summary

Summary

Construction of the RF system for cERL-injector was finished.

Optimal gains has been determined in the operation.

The power supply caused 300 Hz fluctuation is suppressed by high

gain controlling.

RF field requirements is satisfied.

Very good beam momentum.

LLRF (2013), F. QIU

22

Question?

Thank you very much for your attending

LLRF (2013), F. QIU

23

Back up

LLRF (2013), F. QIU

Performance (300 Hz Fluctuation) The 300 Hz fluc. at Inj2&3 and Buncher cavity during CL/OL operation. This 300 Hz

fluctuation would influence the system performance.

The Inj. 1 LLRF system doesn’t not has evident dominant components.

24 Study at cERL (2013)

300 Hz Fluc.

300 Hz Fluc. Bun.

Inj.1

Inj.2&3

Amp. Pha. Pha. (FFT)

No dominant

25

Performance(300 Hz fluc. suppression) The Power supply is the main source of the 300 Hz component.

The RF fluctuation agrees well with the PS fluctuation (suppose 10 deg /HV%, then

the 20mV fluctuation in PS will lead to 10 deg×(100×25mv/15V) = 1.67 deg ).

Fluc. @ 300 Hz Buncher Inj2&3 (VS)

Open loop ΔA/A -43.5 [dB] -46 [dB]

Δθ 0.9 [deg.] 1.6 [deg.]

Closed loop

(KI=5500, KP=0)

ΔA/A -54 [dB] -56.5 [dB]

Δθ 0.3 [deg.] 0.5 [deg.]

300 KW Kly. High Voltage

Cavity input (OL)

0.5 deg. 1.6 deg. Closed-loop

Scope

Spectrum

Open-loop

Cavity Pick up (CL)

Clear to see that the

300 Hz component is

suppressed by CL

operation.

FFT

Study at cERL (2013), F. QIU

26

Gain scanning (300 Hz suppression)


The 300 Hz fluctuation would be suppressed by higher gains.

KI=0 case

∫ki

kp

Cavity

Kly/IOT

DACADC

Loop delay

FF

Pf

SET Value

ADC

Pt

H(s)

K(s)

Disturbance (E.g. 300 Hz fluc.)

Δω Detuning

Cavity

Kly/IOT

DAC

Loop delay

FF

Pf

ADC

Pt

H(s)

Disturbance (E.g. 300 Hz fluc.)

Δω Detuning

Open loopClosed loop

No disturbance suppression! Disturbance suppression: H(s)/(1+K(s)H(s))

27

Performance(300 Hz fluc. suppression)


The 300 Hz component is suppressed by high gains.

The simulation agrees well with

the measured one

Suppression of 300 Hz component

Meas. vs. Simulation

About 20 dB suppression when

increasing KP by 9 times. Spectrum of pick up signal

Sim.

Meas.

~20 dB

28

Fluctuation at 300 Hz (Source)

According to current controlling parameter (KI=10, KP=0), the 300 Hz component

is suppressed by ~10 dB (~3 times), not enough.

10 dB suppression

@ 300 Hz (with KI=10, KP=0)


Fluc. @ 300 Hz Buncher Inj2&3 (VS)

Open loop ΔA/A -43.5 [dB] -46 [dB]

Δθ 0.9 [deg.] 1.6 [deg.]

Closed loop

(KI=10, KP=0)

ΔA/A -54 [dB] -56.5 [dB]

Δθ 0.3 [deg.] 0.5 [deg.]

29

Performance (Vector-sum controlling)


We have used the vector-sum controlling for Inj. 2 and Inj. 3 (see page 4&5 in this

report).

For vector-sum controlling, the measured vector-sum (M+N) which is seen by the

FPGA or DSP is different from the true accelerating voltage which is seen by the

beam (m+n).

The calibration (phase or amplitude) error would result of vector-sum error

Calibration factor

Vector seen by beamVector seen LLRF (Meas.)

m

n

MN1cP

2cP

Phase Calibration Error

neAmeANMV

nmV

cc jPjP

M

B

21

21

Open Loop

m

n

M1cP

2cPClosed loop

N

Detuning

P QSame point

Not same

30

Performance (Vector-sum controlling)


Suppose the detuning comply with 1 deg. RMS Gauss distribution, similar with the

measured result, then the 45 deg. Phase calibration error would result of 0.47% RMS .

. amplitude vector-sum error.

45 deg. calibration error

would result of 0.47% RMS

vector-sum error!

About 0.47% RMS amplitude error

Distribution histogram of the

detuning, similar with Gauss

distribution.

About 1 deg. RMS Zoom!

About 45 deg.

31

Gain scanning (definition) Gain scanning: determine the optimal controlling gains (@ 2MV).

Definition of the integral and proportional gains . I. FPGA input parameter KP and KI.

II. Digital Gain Kp and Ki.

III. Analog Gain kp and ki.

IV. Real Gains: ASet/(ASet-AMeas.)


Gains Integral Proportional

FPGA KI KP

Dig. Ki=KI/218 Kp=KP/27

Ana. ki=Ki/TS(1) kp=Kp

Real ≈ ki*Gop(2)

≈ kp*Gop

1 .TS is FPGA sampling clock period (TS= 1/162.5e6 in cERL LLRF system)

2. Gop is the open-loop gain (Gains from FF to SEL(Fil) during the open-loop operation. For the Inj1 and Inj2&3, Gop ≈ 1 (0 dB).)

KI&KP (FPGA) vs. Ki&Kp (dig.)

Ki (dig.) vs. ki (ana.)

₋₁ z

Ki(digital)

1/s

∫

ki(analog)

kiKi

1/8 1/16

1/8

₋₁ z

∑ 1/32768

I/Q in

KP(FPGA)

KI(FPGA)

I/Q out

FF

Kp (Digital)

Ki (Digital)

ki&kp (ana.) vs. real gain ASet/(ASet-AMeas.)

1/s

∫

ki(analog)

ki

kp(analog)

kp

Cavity

Kly/IOT

DAC

Loop delay

FF

Gop

Performance of CW Superconducting Cavity at ERL test Facility · Performance of CW Superconducting Cavity at ERL test Facility Feng QIU (KEK) LLRF 2013, F. QIU . 2 Main Content ...

Documents