spx llrf r...feb 06, 2012 · spx0 = r&d system proof of principle (1 sector, 2 cavities)...

SPX LLRF R&D

Feb. 6, 2012ASD Seminar

ANL: Tim Berenc, Hengjie Ma, Ned Arnold, Frank Lenkszus, Tom Fors, Bill Yoder

LBNL: Larry Doolittle, Gang Huang, John Byrd,Jim Greer, Kerri Campbell

Outline

Intro LLRF Receiver (prototype results) Deflecting Cavity Behavior (static and dynamic) LLRF Controller (benchtop performance tests) Storage Ring RF modification plans Summary

Advanced Photon Source Upgrade (APS‐U) project

2

3

PlannedSPX Location

SPX LLRF R&D ‐ 2/6/2012 ASD Seminar

APS-U SPX System – Zholents’ Transverse RF Chirp Concept1

4

Goal: provide ~2 psec (presently 50‐100 psec) X‐ray pulses at 6.5 MHz rep. rate for time‐resolved studies

Ideally, second cavity exactly cancels effect of first cavity

1 A. Zholents et al., NIM A 425, 385 (1999)

Correlation between vertical distribution of x‐rays and time‐distribution of electrons that generated them. Pulse can be sliced or compressed with an asymmetric cut crystal.

RF at ~2815 MHz(8th harmonic of Storage Ring RF)


5

SPX0 = R&D System Proof of Principle (1 Sector, 2 cavities) Cavities counter‐phased (180deg) to demonstrate tolerance requirements Cavities run in‐phase to create a chirped beam around entire ring to have a look at short pulse x‐rays


CommonMode

Differential

SPX0 System Performance Requirements 3

6

SPX LLRF R&D ‐ 2/6/2012 ASD Seminar 3 SPX0 PRD, ICMS# APS_1423800 (1/17/12)

Residual kick

Residual tilt

7


Low Level Radio Frequency (LLRF) System

CavFieldProbe

Klystron

Controller

Receiver/ Detector

Setpoint

LLRF

cavRFcav tV cos

Primary responsibility is to regulate the cavity field

tVtV RFQRFI sincos

Polar Coordinates Cartesian Coordinates

22QIcav VVV

QIcav jVV arg

8


Receiver You can’t regulate any better than your receiver

ADCReference

Cavity Signal

ADC

LO CLK

DigitalDownconversion

Analog Front EndDigital Downconversion

cos

sin IF

IF

IF

RF

RF

I

Q

I

Q

Digital Receiver

Ref. PhaseLock Loop

Regulate to a Designated Phase Reference

• Don’t let the LO assume the role of the phase reference• Phase is a Differential Measurement• Mixers preserve phase information, x’s and ÷’s preserve timing• In theory, common mode LO and clock noise cancels

9

SPX Study Meeting – LLRF T. Berenc 7/27/2010

1221

12 2

22

N

q

ADC Quantization Noise

Variance

N = # of ADC bits

CNSNR 76.102.6

C = dB carrier is below full scale

[dB]22/ so fN

2log1076.102.6 sfCNdB

Receiver

10


ADC Quantization Noise

dB ra

d2/Hz

)( fSphase noise variance

2log1076.102.6 sfCN

0

max

0

2 )(f

dffS

14 bit ADC

dB rad2

deg0032.0 rms

16 bit ADC

dB rad2

rms

852 972

(single sideband spectrum)

deg0008.0

Receiver

11


ADC Aperture Jitter

Phase noise

rms aperture jitter

[dB]

)cos())(cos( IFIFIF tt

jitter

fIF=30 MHz

fIF=90 MHz

fIF=150 MHz)log(20 jitterIFjitterSNR

Receiver

12


dB ra

d2/Hz

)( fSphase noise variance

0

max

0

2 )(f

dffS

14 bit ADC, 0.5psec rms

rms30 MHz IF

rms

(single sideband spectrum)

ADC Aperture Jitter

32

log10)log(20

s

jitterIFf

90 MHz IFdeg023.0deg008.0

Receiver

13


Receiver You can’t regulate any better than your receiver

ADCReference

Cavity Signal

ADC

LO CLK

DigitalDownconversion

Analog Front EndDigital Downconversion

cos

sin IF

IF

IF

RF

RF

I

Q

I

Q

Digital Receiver

Ref. PhaseLock Loop

Regulate to a Designated Phase Reference

• Don’t let the LO assume the role of the phase reference• Phase is a Differential Measurement• Mixers preserve phase information, x’s and ÷’s preserve timing• In theory, common mode LO and clock noise cancels

July 2011 SPX Workshop

14

CLK

IF

Receiver – Digital Receiver LLRF4 Board – Differential Phase Noise

15


drift

Receiver - CW Drift Compensation4

4 “Signal Processing for High Precision Phase Measurements”, G. Huang, L. Doolittle, J. Staples, R. Wilcox, J. Byrd , Proceedings of BIW10

16

0.09 ~ 35 dB~ 0.023 rms

~ 0.118 rms

5

Receiver - CW Drift Compensation Demonstration


LLRF 2011 Workshop17

Receiver - Intermodulation Distortion

18

33

221 iiioo vavavaav

For simplicity assume a Taylor series approximation (in general should use Volterra series)

For a 3‐tone input signal: o : RF carrier: Lower side‐band cal‐toneLSB

: Upper side‐band cal‐toneUSB

oUSBLSBo

tVVVVVVVVVVVaVav LSBUSBLSBoLSBoUSBLSBUSBoLSBLSBLSBo cos43

41

21 2223222

31

tVVVVVVVVVVVaVa USBLSBUSBoUSBoLSBUSBUSBoLSBUSBUSB cos43

41

21 2223222

31

tVVVVVVVVVVVVaVa oUSBoLSBoUSBoLSBoUSBoLSBoo cos23

41

21 223222

31



19

Marki T3-03MQP 5 mixer measurements, Sine vs. Square Drive



5 ”T3 Mixer Primer, A Mixer for the 21st Century”, Ferenc Marki, Christopher Marki, http://www.markimicrowave.com/3436/T3_Mixer_Primer.aspx

20


Marki T3-03MQP mixer measurements, Sine vs. Square Drive


21


Marki T3-03MQP mixer measurements, Sine vs. Square Drive


Estimated‐120dBrad^2/Hz at ‐40dBm input (~10dB < F.S.)

~10dB < F.S.

> 20dB improvement

LBNL/SLAC AFE Ver.1ANL 2‐ChannelDown‐Converter Prototype

23

~10dB < F.S.

LBNL/SLAC AFE Ver.1

ANL 2‐ChannelDown‐Converter Prototype

24

ANL 2‐ChannelDown‐Converter Prototype

+LLRF4

confirmed LBNL/SLAC AFE= ‐120dBrad2/Hz at ‐40dBm

Limited by LLRF4 ADC’swith ANL AFE

25

Receiver - comparison (Susceptibility to Interference)


26



CavFieldProbe

Klystron

Controller

Receiver/ Detector

Setpoint

LLRF


Get to know the plant you are controlling – the cavity (deflecting not accelerating) ….

27

Longitudinal voltage

Vertical deflecting voltage

yVyV mZ )(

o

mt

VjV

)0(/ yo

180 90 0 90 1800.5

0

0.5

Cavity Transverse VoltageCavity Longitudinal Voltage

Phase (deg)

Vol

tage

Magnetic FieldElectric Field

Time


Deflecting Cavity - Beam Loading

28

Dipole loss factor:

Circuit definition R/Q:CLR

Vt

yqq oeq 2

2

C

yqV ot

22

21

tloss CVkqU

22

2 24)(

yQR

UyV

qUk o

rZloss

8.17

2

2

UV

QR

r

t



Deflecting Cavity - Beam Loading5,6

29

180 90 0 90 1800.5

0

0.5

Cavity Transverse VoltageCavity Longitudinal VoltageBeam Induced Longitudinal VoltageBeam Induced Transverse Voltage

Phase (deg)

Vol

tage

)0( y)0( y

)(2 sjtooo ecjyi


5 Berenc, “An Equivalent Circuit Model ..”, ICMS# APS_14059786 Decker, “..Tilt Monitor”, DIAG-TN-2010-10, ICMS# APS_1417048

222

tan21/8

scav

B

r

mo

cav

B

o

tg P

Pf

ffQPP

QQR

VP

2221

2 tof

dco eIi

0 yVt0 yVt

)(2 sjtooo ecjyi

BBtB IVP cosˆ21

mAIDC 100 MVVcav 5.0




31For example: see P. Wilson, SLAC-PUB-2884, p. 26, prob 4.4

ttittitItI RFQRFIRFrT sin)(cos)(cos)(

ZRFQZRFIZRFrcav ttvttvtVtV sin)(cos)(cos)(

Deflecting Cavity – Dynamic Behavior

In‐general I/Q modulations of IT each cause both I/Q modulations of Vcav


32

*

* )()(21)()(

NsjZ

NsjZ

sGsG RFRFqqii

*

* )()(2

)()(N

sjZN

sjZjsGsG RFRFqiiq


)tan1(2

tan1cos)()( 222

2

Z

ZZqqii ss

sRsGsG

)tan1(2sin)()( 222

Z

Zqiiq ss

RsGsG

)tan1(2

)tan1()()( 222

2

Z

Z

ssssGsG

)tan1(2tan)()( 222

Z

Z

ssssGsG

Polar Coordinates Cartesian Coordinates

RFj

RFpolar jZejZN Z ZjCartesian eN

classical “Pedersen/Boussard Equations” Modern I/Q Equations

tIttIttItI RFrIRFrIRFrT sin)(cos)(cos)( ttittitItI RFQRFIRFrT sin)(cos)(cos)(


33


Polar Coordinates Cartesian Coordinatesfor discussion “NO DETUNING”

Modern I/Q Equations

)(sGiq

)(sGii

)(sGqi

)(sGqq

GTiiI

GTqiI

s

YsGsG GG 1)()(

0)()( sGsG GG

Y = Beam Loading Factor which can be negative for deflecting cavities. Amp/Phase control can be lost when beam offset drives the cavities to full field (there is no drive carrier). This doesn’t happen for I/Q control.

for discussion “NO DETUNING”

0)()( sGsG Giq

Gqi

s

RsGsG Gqq

Gii )()(

)()(

)()()()(

)()(

sIsI

sGsGsGsG

sVsV

GQ

GI

Gqq

Gqi

Giq

Gii

GQ

GI

Use vector projection techniques to find the transfer functions from generator current and beam current modulations. Generator Current in‐phase modulation is shown here. Total of 15 transfer functions describe the cavity.


34


m

)(sG GNiq

)(sCq

)(sCi

)(sL

)(sGGNqi

)(sG GNqq

)(sGGNii

ny

)(sGBNq

)(sGBNiy

nzy

)(sNi

)(sNq

)(sG BNi

)(sGBNqy

setI

setQ

drivernoiseI kly

noiseI

drivernoiseQ kly

noiseQ

detnoiseI

detnoiseQ

ns)(sGBN

i

)(sGBNqn

)(sGBNi

)(sGBNq

IV

QV

35




36


Lorentz Force

Detuning

Klystron

+

+

m

Controller Cavity

+

-

KlystronController+

-

Microphonics

I-componentLoop

+ )(sCq

)(sCi

)(sL

)(sG GNqq

)(sGGNii

)(sNi

)(sNq

setI

setQ

drivernoiseI kly

noiseI

drivernoiseQ kly

noiseQ

detnoiseI

detnoiseQ

Q-componentLoop

Cavity

IV

QV

No cross‐coupling when cavity is tuned to resonance,except through Lorentz Force Detuning (minimal if stiff cavity)

Pursue R&D benchtop tests with a cavity emulator to quantify the LLRF system contribution to the system error budgets.

37



CavFieldProbe

Klystron

Controller

Receiver/ Detector

Setpoint

LLRF


The Controller….

38


ControllerUSB interface to host computer

39

LLRF4 Receiver Chassis Prototype


Controller – Benchtop Tests

40

BPFx8

1/6

LO

Frequency Generation Chassis

351.94 MHz

REF

2815.52 MHz

2756.86 MHz58.66 MHz

Cavity Emulator


Controller – Benchtop Tests


41

LO

ch2

ch1

LLRF4

ch3

CLK

OUT 1

LO

LO

CavityEmulatorout in

tune

cal

ch4

LLRF4 + Analog Front End

Out-of-LoopPhase Noise Measurement

FFTAnalyzer

BPFx8

1/6

LO


351.94 MHz

REF

2815.52 MHz

2756.86 MHz

Divide by 2CLK

Prescaler

58.66 MHz


Controller – Single System Benchtop Test

42

Noise sources include an independent LO generator used for cavity emulator. This is not something that exists in the ‘real’ system.

What matters is the noise suppression capability in combination with our expected noise sources.

43

What matters is the noise suppression capability in combination with our expected noise sources.

44

ssK

sKsC Z

PZ

P1)(

cavset

E

)(sGqq)(sC

dse

dsqqE

cav

esGsC

)()(11

cav

cavqq s

sG

)(

sec8.1 d

2PK

sec/312 radecav

Controller – Noise Suppression Model

45SPX LLRF R&D ‐ 2/6/2012 ASD Seminar

LO

ch2

ch1

LLRF4

ch3

CLK

OUT 1

LO

LO


tune

cal

ch4


FFTAnalyzer

BPFx8

1/6

LO


351.94 MHz

REF

2815.52 MHz

2756.86 MHz

Divide by 2CLK

Prescaler

58.66 MHz

LO

ch2

ch1

LLRF4

ch3

CLK

OUT 1

LO

LO


tune

cal

ch4


Divide by 2

CLK Prescaler

Out-of-LoopPhase Noise Measurement

ResidualPhase Noise Test Set

Controller – 2 System Benchtop Test

Freq. Generation Chassis

LLRF Receiver #1

Cavity Emulators#1 and #2

46

ResidualPhase Noise Test Set

(measure noise between cavity emulators)

LLRF Receiver #2

Price paid for capability tosuppress low freq. noise

(i.e., beam loading & microphonics)

LLRF System contribution~ 20 fsec rms [0.1 Hz – 1MHz]

Controller – 2 System Benchtop Test

LLRF R&D Outlook (overly simplified)

47

Adding the calibration tone scheme into the cavity control gateware. Currently these are 2 separate code bases, parts need to be merged.

Preparing for real single cavity testing to begin ~ June 2012

Modifications to Storage Ring RF System …

CommonMode

Differential

SPX0 System Performance Requirements 7

48

SPX LLRF R&D ‐ 2/6/2012 ASD Seminar 7 SPX0 PRD, ICMS# APS_1423800 (1/17/12)

Residual kick

Residual tilt

49

Present Storage Ring Beam Jitter 8

22 5.17.065.1~

22 5.07.086.0

22 15.07.072.0

SPX LLRF R&D ‐ 2/6/2012 ASD Seminar8 Sereno et. al, “Storage Ring Phase Noise Studies …” AOP-TN-2012-001

50

Horizontal BPM Data

Feed Forward OFF

Feed Forward ON

AM & PM suppression at Both Stations

Experiment with 360Hz Feed‐Forward correction of Storage Ring Klystron High‐Voltage Power Supply (HVPS) induced noise

Proof of Principle Feed Forward Experiment


51

Adaptive Noise Cancellation Concept 9

9 Widrow et. al, “Adaptive Noise Cancelling: Principles & Applications” IEEE Vol. 63, No. 12, Dec. 1975

From [9]


52

Adaptive Noise Cancellation Concept 9

From [9]

9 Widrow et. al, “Adaptive Noise Cancelling: Principles & Applications” IEEE Vol. 63, No. 12, Dec. 1975

tNtN NQNI sincos

22 )()( QQII NYNY

53

Storage Ring RF AM/PM Noise Suppression Concept


Summary

54

Digital LLRF system shows promise of femto‐second level synchronization [0.1 Hz – 1MHz]with proper attention to common source distribution

Great design improvement demonstrated for Analog Front End with T3 mixers

New I/Q small‐signal baseband model developed for SRF Deflecting Cavities

Adaptive noise cancellation of Storage Ring main 352MHz RF system AM/PM noise is being pursued to reduce present beam jitter

spx llrf r...feb 06, 2012 · spx0 = r&d system proof of principle (1 sector, 2 cavities)...

Documents