Recirculation: looking from inside

Igor, Alessandro and AlexeySpecial thanks to Erik

12 GHz PETS testing at CLEX, CERN (as it was shown in 2008)

Operation mode #1 #2 #3 CLIC

Current, A <30 14 4 101

Pulse length, ns 140 <240

<1200

240

Bunch Frequency, GHz 12 12 3 12

PETS power (12 GHz), MW

<280

61 5 135

DBACRDL

TBTSCLEX

CTF3

CTF2

#1

#2

#3

• Different scenarios of the drive beam generation in the CTF3

• To compensate for the lack of current, the active TBTS PETS length was significantly increased: from the original 0.215 m to 1 m.

• In order to demonstrate the nominal CLIC power level and pulse length, it was decided to implement a different PETS configuration – PETS with external re-circulation.

Variable Splitter (coupling: 01)Variable

phase shifter

To the Load

PETS output

Drivebeam

PETS input

Round trip efficiency: 76%Round trip delay: 25 ns

0 100 200 300 4000

50

100

150

Time,ns

Pow

er, M

W

P 0.9

Calculated output RF pulse envelops in PETS with re-circulation. Circles – mode 2, diamonds – mode 3, boxes – the CLIC pulse by design. Solid line – PETS output, dashed line – to the load.

<30A

14 A

4 A

0 500 1000 15000

100

200

300

Time,ns

Pow

er, M

W

#1. The coupling and pulse length optimized to provide pulsed parameters comparable to the CLIC nominal values.

#2. Full re-circulation (coupling=1) and full pulse length for the mode 3.

Network analyzer

Rec. loop RF check configuration

#2 #1

#3

Measured transmission and reflection for the case of the full recirculation

PETS

Measured transmission and reflection for the case with NO recirculation

S12

S13

Transmissions:

Pis

tons

pos

ition

Pis

tons

pos

ition

b

nn

rc N

n

atjnN

kb

jk etAFFweSFFtFFwFFttU00

12 ))(()()(

PETS single bunch response (GDFIDL)

Measured spectrum of the recycling loop transmission

artificial RF phase delay

Number of round trips

The complete system single bunch response spectrum Multi-bunch part

Number of bunches

Distance between bunches (~RF phase)

Bunch amplitude (~q and F)

PETS with recirculation operational analysis

The method allows for the detailed reconstruction of re-circulation including: The tuning of the loop phase length (if needed) Manipulation of the current amplitude and RF phase along the bunch train Provides direct calculation of the reflection in the loop and the power extraction to accelerating structure

Analyzing the spectrum of the signal, we can add artificially an extra RF phase delay () to the original measurements and tune the total recirculation phase:

Resonant extraction

=0

=180

TBTS PETS single bunch responseGDFIDL simulation

Direct power productionPower build up in 49 bunches

single bunch signal circulating in a loop(full re-circulation)

Power build up for the full (200 ns) beam and full recirculation

Destructive extraction

forward reflected

Example of the full recirculation & rect. 200 ns DB current pulse

Full recirculation and RF phase errors vs. simple model (used by Erik Adli)

g=sgrt(0.752)t=25 ns As measured

=0

=/2

=

Excellent agreement!

Off regime for the On/Off mechanism

When we switch-off recirculation, the RF phase is needed to be shifted by /2 compared to the best phase advance used for the full recirculation. If not then residual recirculation will occur:

single bunch signal circulating in a loop

=/2I=10A

From the PETS

=0

NO recirculation

To the structure

Losses in attenuator ~ 10%

Back to the PETS

Power split=(1-S)Att

TP=S Att

g2 = S Loop =TP Loop/ Att

Measured:Loop=0.76Att=0.9

In 2010 we have measured Tp (blue).In 2011 we measured g2 and Power split (red).

Reference plane

Power balance (2011)

zoom

(1+S)-S1-S = Power split / 0.9S = g2 / 0.76

g2,

Pow

er s

plit

Following theory, both curves can be nicely fitted with cosine/sine functions:

2

2

5.1029.10cos76.0)(

xxg

2

5.1029.10sin9.0

xPowersplit

DB: 10 A x 250nsAtt: 16.6 mm

The attenuator settings s=16.55 mm (g22011 = 0.334). The recirculation is tuned in phase with rectangular dB pulse

PETS output

To structure

gr

gb

QV

L

L

V

QRFLIP

2

,)exp(1

4

/

0

2

0222

With measured DB current pulse (FF2=1)

Power production in the PETS (normalization):

1 A drive beam produces 0.306 MW RF power

R/Q= 2290 OhmBeta= 0.453 CQ= 7200L= 1 m

Vs. simple model

F2=0.415=3.73 mm

Simulated power (s=16.55 mm) scaled to the measured (s=16.60mm):

Fitted phase error for recirculation is 1.80

From PETS(#707)

F2=0.56=3.03 mm

F2=0.415

Power split(#701)

F2=0.415

F2=1.8

Reflected from 2nd attenuator

The whole system signals analysis. The measured spectra of the 2nd attenuator and accelerating structure were used

From PETS(#707)

(#710)

707

701

702

708

F2=0.415=3.73 mm

All signals tracking

These simulations allow to conclude on the relative calibration errors.For example, if one believe that #707 is properly calibrated, then power level in #701 is overestimated by 35% and in #710 by factor 4.3.

Spectral analysis vs. Adli’s model

g2 = 0.334measured

g2 = 0.355fitted

=1.80

F=1

The two approaches show a good agreement. The difference in recirculation gain is ~3%

10%

Rise time 20 nsFall time 30 ns

The drive beam BPM bandwidth issues

Drive beam amplitudeAnalysis stage by stage

20 degrees RF phase jump

+ frequency detuning by 1.1 MHz

Drive beam phase and frequency

The residual discrepancy of the pulses shapes can came from: Variation of the bunches form factor linearity of the detector

Relative single bunch form factor

0.19

‘simple’ sine-type modulation

707 701 710

702706

Fitted form-factor 0.906

X 0.6

X 3.9

All signals tracking

-50dB Directional coupler re-calibration