00 0 Changes to the PS RF system 04 April 2014 H. Damerau Many thanks for discussions and input to L. Arnaudon, D. Cotte, S. Hancock, R. Maillet, M. Morvillo,

111

Changes to the PS RF system

04 April 2014

H. Damerau

Many thanks for discussions and input toL. Arnaudon, D. Cotte, S. Hancock, R. Maillet, M. Morvillo, M. Paoluzzi, D. Perrelet, S. Rains, C.

Rossi, S. Totos

BE/OP Shutdown Courses 2014

2Overview

• Introduction• RF systems for manipulations

• Global beam control system modifications• Injection bucket control• RF controls renovation

• 10 MHz RF system• Voltage control and 1-turn delay feedback loops• Tuning group restructuring• 10 MHz matrix control, spare cavity selection and

linked timing

• High-frequency and wide-band RF systems• 20, 40, 80 MHz and 200 MHz• Wideband cavity for longitudinal damper

• Summary

3Overview




linked timing


• Summary

4Introduction

® Beams from single bunch to 72 bunches, flexible longitudinal pattern

® Intensity range from about 109 to 3 · 1013 particles per cycle

® Major bunch shortening along the cycle, from 180 ns to 4 ns (45 times!)

® After LS1: BCS, BCMS, PBC, etc.

What has changed for the RF systems after LS1?

What to do with the modified RF systems? Steven’s lecture

PSProtons

from PSB• Protons and ions

to SPS (and LHC)

• Protons to AD and nTOF target

Pb54+ (future: Ar11+, Xe39+) from LEIR

Protons to fixed target experiments

5RF Systems to perform manipulations

to SPS

PS

24 (+1) cavities from 2.8 to 200 MHz

Acceleration

2.8 – 10 MHz

200 MHzLongitudinal blow-up and 200 MHz structure for SPS 13/20 MHz

80 MHz

40 MHz

RF Manipulations

RF Manipulations

6Overview




linked timing


• Summary

7

• Bunches from PSB must be placed into the correct buckets

• Batch compression works only for even number of bunches

® Bucket number control during both transfers PSB PS

1 turn

Injection bucket selectionPA.DCNBINJ

8

DDS

h128

Inj. Bucket selectio

n

MHS

DDS

MHS h = 1

… …

RF

dir

ectl

y to

cavit

ies

(

on

e D

DS

per

cavi

ty)

f rev, c

lose

d

loop

Tagged clock

fRF, inj. = 9 ·

436.568 MHz

MHS hPL = 9

1. Sync. on h = 1,

fix bucket #

2. Lock f- loop on

inj. synth.

Shifted trains to

PSB, 1st inj.

Bucket number control

Synchronizing PSB and PS – 1st injection

Generate synchronous h1, h4 and h8 for PSB, while locking f-loop on h9

R.

Garo

by,

Mu

lti-

harm

on

ic R

F

Sou

rce f

or

the A

nti

-pro

ton

P

rod

ucti

on

Beam

of

AD

, C

ER

N

PS

/RF

/Note

97

-10

Df (phase loop)

Df (injection synchro.)

4 ms/div

1. 2. Inj.

9

DDS

h128

Inj. Bucket selectio

n

MHS

DDS

MHS h = 1

… …

f rev, c

lose

d

loop

Tagged clock

MHS hPL = 9

1.Sync. on h =

1, fix bucket

#Shifted

trains to PSB, 2nd inj.

Generate synchronous h1, h4 and h8 for PSB, while locking f-loop on h9

Bucket number control R

F d

irectl

y to

cavit

ies

(

on

e D

DS

per

cavi

ty)

Synchronizing PSB and PS – 2nd injection

R.

Garo

by,

Mu

lti-

harm

on

ic R

F

Sou

rce f

or

the A

nti

-pro

ton

P

rod

ucti

on

Beam

of

AD

, C

ER

N

PS

/RF

/Note

97

-10

10Overview




linked timing


• Summary

11Major renovation of LLRF controls

• Profit from global controls renovation (ACCOR) to move front-ends for PS beam controls to central building

Before The plan…

12Major renovation of LLRF controls

During… Back on…

® Close collaboration between BE/CO and BE/RF

• Profit from global controls renovation (ACCOR) to move front-ends for PS beam controls to central building

E. S

aid

and

inst

alla

tion

team

13Migration of controls devices

Devices Obsolete GM class

New class

Timings PTIM-V LTIM

Functions GFAS CVORB

Digital bit controls

DIGIO CGDIO_B (bit)

Digital selectors DIGIO CGDIO_E (enum)

Digital controls DIGCTL CGDIO_C (cont.)

Pentek synthesizer

DIGCTL V346

10 MHz matrix RFMAT MATRF

‘Tra

nsp

are

nt’

m

igra

tion

® 67 CVORB function channels® ~220 physical timing channels, ~1240 timings in total

(multi-pulses)® ~35 Further devices: digital controls, RF synthesizer,

etc.® Kontron PC front-end for MIL1553 loops (all cavities)

on/off/reset® About 10 kilometers of new cables, but in about 250

smaller pieces

14

® Should be sufficient for the future® Should be sufficient for the near future?

Controls upgrade from TG8 to CTRV:® Each function with restarts now 16 instead of 8 restart

timings

Twice more complicated RF manipulations possible:® Sequence of 16 phase loop harmonics® Sequence of 20-25 RF harmonics® RF manipulations twice as complicated as before LS1

(2 BCMS)

Ready for new RF manipulations

15Overview




linked timing


• Summary

16

AVC

1TFB

h h200

Final Amplifier, 10 MHz

Cavity, Fast Wideband FB

DAC

ADC

DAC

Gap ReturnDrive

H

- Fast wide-band feedback around amplifier (internal) Gain limited by delay

- 1-turn delay feedback High gain at n frev

- Slow voltage control loop (AVC) Gain control at fRF

Vprog (digital)

PS 10 MHz feedback overview

D. Perrelet

17Voltage control (AVC) loop

• Pre-LS1 hardware required analog voltage program• Regulation characteristics not optimum for fast voltage

jumps

RF

fro

m

beam

con

trol

Vp

rog

(an

alo

g)

Analog voltage control:1. AVC loop

functionality2. Interlocks

Loop filter

® Upgrade to fully digital implementation of AVC loop® Migrate interlock part to separate surveillance hardware

P. M

aese

n, P

S/R

F/N

ote

94-2

5

Pre-LS1

18Digital voltage control (AVC) loop

® Harmonic number functions PA.GSHA/B/C required for all beams

RF

fro

m

beam

con

trol

Vp

rog

(dig

ital)

Digital PID

Non-I/Q detector

• Digital voltage control loops integrated into 1-turn delay feedback HW

• Separate surveillance module to assure hardware safety

M. H

aase

D. P

erre

let

Veto

h (digital)

19Principle of the 1-turn delay feedback

® Comb filter for high gain at frev harmonics® Delay circuit to correct total feedback delay to a full

turn® Additional notch filter to cancel feedback gain at fRF

• Classical feedback limited by unavoidable delay• BUT: Impedance reduction of cavities only needed at frev

harmonics

D. Boussard, G. Lambert, PAC83, pp. 2239-2241

20

• No need for:• Multiple clocks, avoiding double sampling at 4 fRF and 80

frev

• Wide-range clock phase locked loops

• External delay cables

• Increased resolution of signal processing from 10 to 14 bits

• Ready for proton beams at any harmonic number• No need to start from h = 8 (limitation in old

system)• Compatible with all LHC-type beams

• Integrated electronic delay generation

• Possibility to raise feedback gain by firmware improvements

• Include a digital AVC in the firmware to replace analog hardware

New 1-turn delay feedback

21Flexible feedback board development

D. P

erre

let

• Versatile board:• 4 ADC/DAC channels with powerful FPGA• Delay line chains to complete delay of 1-turn• CVORB and fast serial ports for connections• Further/future applications: PS transverse feedback,

coupled-bunch feedback, 1-turn delay feedback for 20/40/80 MHz systems and transverse dampers of PSB, AD, LEIR

22

• Four new VME crates installed in building 359 (one per tuning group)

• Example for tuning group B (cavities C56, C66, C76 and C81):

Installation

All 10 MHz cavities equipped:

AVC loop commissioning ongoing, then 1-turn delay feedback

D. P

erre

let

23Overview




linked timing


• Summary

24

• 10 MHz require a tuning current of up to 3000 A to cover 2.8…10 MHz

• Multiple cavities in series to reduce tuning power supplies:® 3+1 tuning power supplies since 1984® Two groups of 2 cavities and one big group (2 + 2 +

6)® For RF manipulations voltage limited to 40 kV per

group• Change tuning group

configuration to 3 + 3 + 4® Adapt to present needs® 60 kV per group

® Rewire 3 kA cables!

Change of tuning group configuration

Below PS ring (inside)

25

To C66

Connections to C56

Rewiring 3 kA cables: before

To C66

To C51

To C51

Connection box below C10-56

V. D

esq

uie

ns

26

To C66

Connections to C56

Rewiring 3 kA cables: after

To C66

To C51

To C51

Connection box below C10-56

V. D

esq

uie

ns

27

hA: 36, 46

hB: 51, 56, 66, 76, 81, 91

hC: 86, 96

Tuning-Groups:

Group 4: 11, test cavity

→ All cavities of group tuned to same frequency

→ Hard-wired structure of tuning groups

→ 40 kV in three groups

Frequency: Fixed tuning circuits (1984-2013)

28

hA: 36, 46, 51hB: 56, 66,

76, 81

hC: 86, 91, 96

Tuning-Groups:

Group 4: 11, test cavity

Frequency: Fixed tuning circuits (2013-)

→ All cavities of group tuned to same frequency

→ Hard-wired structure of tuning groups

→ 60 kV in three groups

29Tuning group change: status and consequences

Behavior of tuning groups verified before and after the change

First six cavities in the ring now pulsing

• But:

• All cavities in voltage program group should be in same tuning group

® Now violated for almost all cycles® Need to adapt timing trees ( Steven’s presentation)® Need to rebuilt each cycle before it can be executed!

• Major effort of reprogramming all RF functions

Please do not just put pre-LS1 cycles in the super-cycle!

Absolutely ‘non-transparent’

30Overview




linked timing


• Summary

3131Voltage program generation: hardware matrix

Pre-LS1

® Dangerous single point of failure

Fu

ncti

on

s an

d

tim

ing

s p

er

gro

up

Fu

ncti

on

s an

d

tim

ing

s p

er

cavi

ty

• Hardware matrix has served from the mid 1990’s until LS1

• Complicated hardware with embedded micro-processor

6 analog functions + 12 timings in 11 analog function + 22 timings out

• Few spare boards in unknown state left

32

Global program

Modifier grp. 1

Modifier grp. 2

Modifier grp. 3

Modifier grp. 4

Modifier grp. 5

Modifier grp. 6

×

×

×

×

×

×0…100%

0…200 kV

C1 1

C3 6

C4 6

C5 1

C5 6

C6 6

C7 6

C8 1

C8 6

C9 1

C9 6Voltage programs to

cavities:

Global red. ×0…100%

Mapping from groups to cavities

• voltage programs• gap relay timings

Voltage program generation

3333New software based 10 MHz matrix

New implementation guidelines:• Distribution of digital voltage program data to each

cavity® CVORB function generator channel and CTRV timings per

cavity • No specific central hardware as single point of failure

® Only simple electronics to distribute serial data streams• Move relevant parts of matrix to software virtual

matrix

1. Combine functions and restart timings to so-called real-time function per voltage program group

2. Copy real time functions per group to functions per cavity

® Major implementation effort by P. Pera Mira and G. Kruk

H.

D.,

S.

Han

cock,

CE

RN

-AT

S-N

ote

-20

13

-02

1 T

EC

H

34Real-time function generation

• Voltage program modifier functions per group (with restarts) difficult to map

1. Calculate real-time functions PA.GS…RT, based on functions and timings

2. Copy only real-time (RT) functions to the functions per cavity

Function with

restarts

Real-time function• Functions per group drive

no hardware anymore® Possibility to virtualize

later

( )

35

Global program

Modifier grp. 1

Modifier grp. 2

Modifier grp. 3

Modifier grp. 4

Modifier grp. 5

Modifier grp. 6

×

×

×

×

×

×

0…100%

0…200 kV

C1 1

C3 6

C4 6

C5 1

C5 6

C6 6

C7 6

C8 1

C8 6

C9 1

C9 6Voltage programs to

cavities:

Global red. ×0…100%

Modifier grp. 7

Modifier grp. 8

×

×

Mapping from groups to cavities

• voltage programs• gap relay timings

• Hardware switching of functions and timings migrated to software

• InCA MakeRules to copy settings from groups to cavities

• Integrated spare cavity selection mechanism for C11

® Virtual matrix

Voltage program generation

36

• More flexibility thanks to renovated generation of voltage programs: so-called ‘matrix’

• New hardware to generate digital voltage program data for each cavity

® 8 logical groups of cavities® Matrix functionality implemented in software® Commissioned and essentially ready for start-up

Upgraded distribution of voltage programs

37Application program

• New application (D. Cotte, R. Maillet)

® Finally little changes from the operations point of view® Integrated spare cavity selection, also based on InCA

MakeRules

38Spare cavity selection

• Spare cavity C11 replaces any other 10 MHz cavity, needs:• Voltage program like any other cavity• Harmonic number and relative phase special

• Previous implementation: hardware multiplexers for GFAS functions

• MakeRules now copy relevant functions and timings on all LSA cycles

PA.GSHART

PA.GSHBRT

PA.GSHCRT

PA.GSHDRTA, B or C

PA.GSRPART

PA.GSRPBRT

PA.GSRPCRT

PA.GSRPDRTA, B or C

Tuning group of cavity to be replaced by C11

® No dedicated hardware involved anymore® Flexible, could even drive C11 with its own functions and

timings

Harmonic Relative phase

39Overview




linked timing


• Summary

40Glue logic for the control: linked timing

→ Timing tree structure assures coherence between harmonic number (tuning group) and voltage programs (matrix)

• Old timing trees (LKTIM) incompatible with LTIM/CTRV timings

• Migration of X-motif application to Java needed

→ New trees again with node and parent lists (Mbno device ID)

→ Rules for timing and status calculation moved to InCA

… … … …

`

Pre-LS1

41Glue logic for the control: linked timing

• New application (D. Cotte, G. Kruk for the InCA part):

® Little changes from the operations point of view® Comfortable handling of large timing trees, commissioning

ongoing

42Overview




linked timing


• Summary

43Status of 20, 40, 80 and 200 MHz cavities

• 20 MHz cavities As before LS1

• 40 MHz C40-78 repaired, being commissioned

Pre-driver amplifiers water-cooled Otherwise as before LS1

• 80 MHz cavities Pre-driver amplifiers water-cooled

Otherwise as before LS1

• 200 MHz cavities Cavities as before LS1 Renovation of C201/206 amplifiers

All high frequency cavities will be available for the start-up

44Automatic tuning for 40 and 80 MHz systems

• Keep cavities at fixed frequency® Microprocessor-based system

after LS1

L. A

rnau

don,

S. T

otos

® Will also pilot switching of 80 MHz cavities for protons and ions

® Commissioning during/after start-up

PX.SSC-C200

Start SuperCycle

0V

24V

2µs

0µs

Motor DriverDigital Microstep Driver

DMD5056

R

Control Panel

Lim OutLim Out

Lim InLim In

RUNRUN

LCD 2*16 or7-segLED

CMD INCMD IN CMD OUTCMD OUT

AUTO

MANUAL

AUTO

MANUAL

I2C

AUTOTUNING

CONTROL

FESA

ETHERNET

RFSHUT OFF

&TIMING

VETO 2

AVCCTRLLOOP

DAC

ADC

AVC VETO

TEST PERTURBATION

CONTROLMOTORCONTROL

RF 0degAmplifiers

CAVITYPHASE DISCRIMINATOR

RABBITRCM4010

SERVOTUNER

AY04Pulse

Repeater

RESOLVERDECODER

M

45Renovation of 200 MHz amplifiers (C201/206)

Before

New amplifiers in BA3 (SPS)

Old amplifiers

• High power chain C202/C203/ C204/C205 completely renovated during the 2006/2007 shutdown

® C201/206 followed during LS1

S. R

ains

, Ch.

Ren

aud

New amplifiers

46Renovation of 200 MHz amplifiers (C201/206)

New amplifiers ready in building 151 (C201) • Amplifiers

® Special old types irradicated

® Interchangeable with SPS

• Interlock system® More evolved

then for C202-C205

• Power supplies® Reliability

® C201/206 renovated for the start-up all systems almost identical

® Strategy for operation after LS1:1. Run with four 200 MHz RF systems, including C201

and C2062. Keep two cavities as hot spares

New amplifiers

47Overview




linked timing


• Summary

48New cavity (#25) in the PS ring

M. P

aolu

zzi

• Wide-band (0.4 – 5.5 MHz, VRF = 5 kV) cavity based on Finemet material

• No acceleration, but damping of coupled-bunch oscillations SS026-cell cavity unit

Accelerating gap

Power amplifiers (solid state)

B E A M

F I N E M E T

G A P

• Cavity installed in SS02, start with amplifiers on 2 gaps® First installation of transistor power amplifiers close to

beam in PS

49Coupled-bunch oscillation damping• Bunches oscillate with different phases (and

amplitudes)Example of an n = 12 mode (Df ≈ 206)

• Mode number n defined by phase advance from bunch-to-bunch:

® Coupled-bunch oscillations show up as side-bands of nfrev or (h-n)frev

Df = 2p n/h

® Frequency range of new Finemet cavity allows to damp all modes

® Commissioning after start-up

frf

Kick

MeasureKicker

basebandPreferred detection

50Overview




linked timing


• Summary

51Summary

• Extensive modifications to the PS RF systems during LS1

® Most of them expected to be ‘transparent’, though the underlying design changes are significant

® Tuning group change not transparent: ® Rebuilt almost all cycles® Existing cycles must not be played without prior

modifications!

• Re-commissioning of RF systems progressing• Initial commissioning of new equipment, e.g.,

longitudinal damper will continue after start-up

® Steven’s lecture on 10/04/2014

What to do will all this (new) stuff? ® Smooth start-

up?® Smooth start-

up!

52

Thank you very much for your attention!

Documents

00 0 Changes to the PS RF system 04 April 2014 H. Damerau Many thanks for discussions and input to L. Arnaudon, D. Cotte, S. Hancock, R. Maillet, M. Morvillo,