RF Commissioning

Andy Butterworth BE/RFThanks to: J. Tuckmantel, T. Linnecar, E. Montesinos, J.

Molendijk, O. Brunner, P. Baudrenghien, L. Arnaudon, P. Maesen, F. Dubouchet, D. Valuch, D. Van Winkle, C.

Rivetta and many others

Outline• Introduction• Equipment checkout• Warm commissioning• Cavity conditioning• LLRF commissioning• Data acquisition and diagnostic tools• Controls• Summary

Introduction• Based on experience with LHC and LEP SC RF

commissioning• Assumptions:

– Power plant derived from 800MHz IOT amplifier system currently being specified for the SPS Landau cavities

– Low Level RF similar to the LHC 400MHz system– Controls derived from LHC and SPS RF and integrated into

the LHC control system

Introduction: LHC and SPS systems• LHC acceleration system: 16 SC cavities @ 400MHz

– each driven by 300kW klystron• Power system (slow controls) using PLCs

– high voltage power supplies, klystrons, RF power distribution, cryostats and ancillary equipment

• Low Level system (Cavity servo controller and Beam Control)– digital, FPGA at 80 MS/s, DSP for turn-by-turn (11 kS/s)– in parallel with fast analog for transient beamloading– 2 VME crates per cavity

• SPS Landau cavity system: 2 travelling wave cavities @800 MHz– each driven by 4 x 60kW IOT amplifiers (currently in

procurement)

Commissioning: Equipment check-out• RF and control cables: individual testing

– reflection (short circuit) for cable identification and initial validation

– transmission for phase sensitive signals– reflection tests with termination to assess

damage, faulty connectors etc. – > 100 cables per cavity in LHC...

• Control racks equipped and tested in lab– after installation, signals test up to PLC control

level• Data exchange with vacuum, power

converters, cryogenics• Interlock system: individual test and

validation, access, radioprotection

D. Valuch

Warm commissioning: amplifier• Initial commissioning of power amplifiers done during

reception testsTechnical Specification for

the Power Amplifier for the SPS 801 MHz RF system

12.8 Site acceptance tests•48 hour full power operational test, during which the frequency, phase, linearity, output power, stability and efficiency will be monitored.•Measurement of gain.•Output voltage and current.•Output voltage ripple at specified frequency ranges.•Power factor at full output.•Harmonic injection into the mains supply at selected levels including full output.•Time taken to switch to zero output on receipt of the ‘fast switch-off’ command.•Energy deposited into a short-circuited output.•Effectiveness of internal protection circuits.•Temperature rise during ‘soak’ operation at full power of transformers, inductors and semi-conductors.•Compliance with electromagnetic noise requirements (includes RF, x-rays).•Compliance with acoustic noise requirements.•Purity of IOT RF spectrum.•Thermal run.•Reproducibility check.•Harmonic analysis.•Inrush current measurement.•Power factor correction test.•Mains regulation tests.•Line regulation tests.•High voltage DC test to be carried out at an over-voltage to be agreed.•Power Amplifier RF tests.•etc...

Warm commissioning• ... of complete power system with RF

– waveguides short-circuited to isolate cavities

• HV and RF interlocks: individual tests– water flows, WG arc detectors etc

• Bring amplifier slowly to full RF power– calibration of power measurements (directional couplers)

and attenuators for signal distribution– test and adjust circulator and load

• Long-term power test (100 hours)

If SPS type 800MHz power plant is used, most of the procedures will already be well defined

Low power measurements

• ... on cold cavities• to confirm measurements

from test stand– loaded Q– tuning range

• using resonant frequency and bandwidth measurements with network analyser– drive signal injected via coaxial

transition on waveguide– return signal from cavity field

measurement antenna

We can now remove the short circuit and proceed with high power testing...

High power tests: cavity conditioning• Run cavity up to full power and voltage while observing

vacuum and cavity field emission• Vacuum

– in practice, we are looking at the vacuum in the main input coupler

– difficult to see outgassing in cavity (pumped by the cold surfaces)

• Field emission in the cavity– shows as He pressure excursions RF trips– in LEP, feedback on X-ray emission was possible (total

cryomodule voltage of 40MV)– radiation not detectable in LHC (single cavity, 2MV)

• The input couplers are equipped with a DC bias voltage to suppress multipacting during normal operation: this is switched OFF during conditioning

Cavity conditioning method• Pulsed FM modulated RF power is applied to the cavity in a

controlled way with vacuum feedback• Two loops

– Fast vacuum feedback– Slow computer controlled loop to generate AM envelope and increase

field and power as conditioning progresses (pulse to pulse at 50Hz)

rate of rise rate of fall

flat-top

Total Envelope <= 90sRep-rate 20ms pulse width: 10,20,50us thru 1,2,5,10ms,CW

RF Power0 thru 300kW

Programmable Conditioning Envelope (Not to scale)

J. Molendijk

Cavity conditioning method

Increase Power using a fixed pulsewidth until Pmax is reached.

Increase Pulse width and proceed with the Power as above.

Finish with CW and proceed with the Power as above.

A

B

C

J. Molendijk

Conditioning: Implementation• Conditioning system is fully integrated in LHC Low Level RF

Cavity Controller

Tuner Loop

Switch & Limit

Conditioning DDS

VME CPU

J. MolendijkCavity Controller Tuner/Klystron crate

J.C. Molendijk - CWRF2008 142008-03-27

The Conditioning DDS Module

500MHz Synthesizer

(DDS clock)

4 Channel DDSRF Summing & Gain Control

RF Generator 1

RF Generator 2

Vacuum Loop CPLD

J.C. Molendijk - CWRF2008 15 2008-03-27

Conditioning DDS – I/Q plot of Dual FM sweepsIcFwd

-2500

-2000

-1500

-1000

-500

0

500

1000

1500

-3500 -3000 -2500 -2000 -1500 -1000 -500 0 500 1000 1500 2000 2500 3000 3500

I

Q

Actual Data Obtained from Forward Current, I/Q memory in the Tuner loop module

LHC SC RF Conditioning GUI

Envelope & Power

Vacuum Loop

Dual FM Generator & Status

Pulse

Global

Cavity V gapCavity P Fwd

Cont

rols

J. Molendijk/F. Dubouchet

Conditioning: Summary• Typical conditioning time to full power and voltage

for an LHC cavity was a few days to 1 week• Highly automated, but still requires regular human

supervision to adjust parameters• Integrated conditioning system in LLRF hardware has

proved very efficient, and allows conditioning of multiple cavities in parallel

• Main power coupler DC bias switched on only after conditioning

Cavity Servo Controller. Simplified Block Diagram

Technology: DSP

CPLD or FPGA (40 or 80 MHz) Analog RF

Signals: Digital:

Analog:

Digital I/Q pair:

Analog I/Q pair:

X

Digital IQ

Demod

SUM

Tuner Processor

X

Digital IQ

Demod

Dir. Coupler

Single-Cell Superconducting

Cavity

Fwd

Rev

X

Digital IQ

Demod

DAC

Digital RF feedback (FPGA)

1 kHz

60 dB

From long. Damper

Voltage function

I0

Q0

dpdV

Set Point Generation

DA

C

Phase Equalizer

ADCDAC

SUM

Vcav

SUM

SUM

Analog RF feedback

1 kHz

20 dB40 dB

1-Turn FeedforwardWideband

PUDAC ADC

An

alog IQ

Dem

odulator

Ic fwd

X

Digital IQ

Demod

Ic rev

TUNER LOOP

SET POINT

RF FEEDBACK

RF MODULATOR300 kW Klystron

Circ

Ig fwd

Analog IQ

D

emodulator

QI

ANALOG DEMOD

Klystron Polar Loop (1 kHz BW)

AD

C

Gain CntrlSUM

Baseband Network Analyzer

DA

C

noise

X

Digital IQ

Demod

X

Digital IQ

Demod

1-Turn Feedback

Tuner Control

Ic fwd

CONDITIONING DDS SWITCH/PROTECTION

SWITCH

Master F RF

Analog IQ

M

odulator

RF Phase Shifter

Phase Shift

Var G

ain RF

A

mpifier

-

LHC cavity Low Level

Low Level RF commissioning• Tuner loop

– tuner phase adjustment to set the cavity on tune when the loop is closed

• RF feedback– phase alignment of digital and analog feedback branches– adjustment of feedback gain and phase before closing loop– measure closed loop response: important for

• beam loading response• cavity impedance seen by beam• bandwidth of cavity voltage control

– measure phase noise

• Amplifier phase/amplitude loop– adjust gain and phase setpoints– adjust dynamic loop responses

Lots of work with a network analyser, but new tools are at hand...

LLRF embedded data acquisition • All LHC LLRF boards have on-

board signal recording memory

• 2 parallel sets of buffers:– Post-Mortem capture– User “Observation”

• 64 turns @ 40Ms/s 128kB/signal

• Revolution frequency tagging

RF feedback board

“Baseband network analyzer” (BBNWA)

• Loops excited with noise• Output signals recorded

digitally• Transfer function

estimate program calculates frequency domain transfer function

• “Fit” an idealized linear model to the measured data and calculate recommended adjustments

0 500 1000 1500 2000 2500 3000 3500 4000 4500-2000

0

2000

Sample #

Cou

nts

White Noise Kernel

0 0.5 1 1.5 2 2.5 3

x 105

-1

0

1

I-O

utpu

t (N

orm

.)

I-Output Signal

0 0.5 1 1.5 2 2.5 3

x 105

-1

0

1

Sample #

Q-O

utpu

t (N

orm

.)

Q-Output Signal

• LLRF boards also have embedded memory buffers for “excitation data”: can inject signals into the loops

D. Van Winkle

MATLAB tools for remote setting-up of LLRF developed with US-LARP collaborators (D. Van Winkle, C. Rivetta et al.)

RF feedback alignment in the comfort of your home

Page 22

Open Loop Alignment

-60 -40 -20 0 20 40 60-80

-70

-60

-50

Frequency (kHz)

Gai

n (d

B)

Measured Magnitude and Phase with Fit of Open Loop

Fit

Data

-60 -40 -20 0 20 40 60-400

-200

0

200

Frequency (kHz)

Pha

se (

degr

ees)

0.0005

0.001

0.0015

0.002

30

210

60

240

90

270

120

300

150

330

180 0

Polar Plot of measurement and Fit of Open Loop

-60 -40 -20 0 20 40 60-80

-70

-60

-50

Frequency (kHz)

Gai

n (d

B)

Measured Magnitude and Phase with Fit of Open Loop

Fit

Data

-60 -40 -20 0 20 40 60-400

-200

0

200

400

Frequency (kHz)

Pha

se (

degr

ees)

0.0005

0.001

0.0015

0.002

30

210

60

240

90

270

120

300

150

330

180 0

Polar Plot of measurement and Fit of Open Loop

Un-Aligned Open Loop Response and Model Fit (SM18)

D. Van Winkle

BBNWA contd.• Comparison of closed loop response measured with

instrument vs. embedded BBNWA measurements

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1-20

-15

-10

-5

0

5

10

Frequency (MHz)

Gai

n (d

B)

Closed Loop measure in SM18 (BBNA)

Network analyzer (Agilent) BBNWA

D. Van Winkle

Frequency (MHz)

Automated tuner setup• MATLAB script for Automatic setup of tuner loop

Tuner sweep to find resonance

Sweep across resonance with different phase shifter values

Converge to optimum phase

LLRF: Summary• It took almost 1 month to set up the Low-Level RF of

the first cavity • Once the procedures were well defined, the last few

cavities took about 1 day each

• New automated tools using MATLAB and the BBNWA feature of the LLRF hardware will save a lot of time

Many thanks to our US-LARP colleagues from SLAC

Controls and software• PLCs and Low-Level crates

interfaced via FESA (Front End Software Architecture)– a “FESA class” software module

is written for each type of equipment

– communication via the CERN Controls Middleware

• Application software:– LabView expert synoptic panels– Matlab scripts for more

sophisticated specialist applications

– LSA (standard machine operations software) manages settings and sequencing

– Logging and Post Mortem– Alarms (LASER)– ...

Summary• Power system commissioning procedures will be well

known if using existing power source (SPS type 800MHz)

• Controls and application software should be based on standard CERN controls infrastructure

• LHC low-level electronics has built-in conditioning and diagnostics facilities, and is already well integrated into the control system

• Powerful tools are being developed for LLRF setting-up which could equally be applied to the crab cavity system