Bunch Numbering

Bunch Numbering

P. Baudrenghien AB/RF

for the LHC/RF team

March 11, 2007 LHCCWG meeting 2

Numerology For each ring:

Convention 1: The 400 MHz RF defines 35640 buckets, spaced by one RF period, and numbered from 1 to 35640

Convention 2: Bucket 1 is the first bucket after the 3 s long abort gap (defined from bucket 34442 to 35640)

For 25 ns operation the bunches will occupy buckets 1, 11, 21 etc. with gaps occurring every PS or SPS kicker gap. For 43 bunch operation the bunches will occupy buckets 1, 811, 1621, etc. (see LHC-OP-ES-0003 rev 1.0 for the different schemes).


For 25ns operation the bunches will occupy buckets 1, 11, 21 etc. with gaps occurring every PS or SPS kicker gap. (see Figure 1 above reproduced from LHC-OP-ES-0003 rev 1.0).


For 43 bunch operation the bunches will occupy buckets 1, 811, 1621, etc. (see Figure 3 above reproduced from LHC-OP-ES-0003 rev 1.0).


Revolution Frequency or OrbitFor each ring:The revolution frequency (Frev or orbit) is a train of pulses, with one 5 ns long pulse per turn. This pulse “points” to bucket 1. The revolution frequency is obtained by dividing the RF by 35640

At a given place in the machine, and at a given beam energy (that is fixed RF frequency) the delay between the pulse and the passage of a bunch in bucket 1 will be fixed from run to run

Drift during the ramp: Signals remain in phase with the corresponding beam in IP4 (RF cavities). During the acceleration, the revolution frequency increases by 2.2 ppm for protons (868 Hz @ 400 MHz) and 14 ppm for Pb (5.5 kHz @ 400 MHz). [See LHC Radio-Frequency Swing, P. Baudrenghien, 14 th Leade meeting, Dec 15th, 2003]. At a given place, but varying energy (frequency) the Frev-bunch delay will drift during the acceleration ramp due to the difference between signal transmission delay and the beam time of flight. For protons we have 6.5 ps/km, for ions 41.25 ps/km. Not a problem.


Bunch ClockFor each ring:

The Bunch Clock is a square wave obtained by dividing the RF by 10. The divider is synchronized on the Revolution Frequency (see page 8)

At a given place in the machine, and at a given beam energy (RF frequency) the delay between the edge of the Bunch Clock and the passage of a bunch will be fixed from run to run

Drift during the ramp: During the proton ramp the Bunch Clock frequency increases by 86.8 Hz (protons) and 550 Hz (Pb). At a given place, but varying energy (frequency) the edge will drift with respect to the bunch. (Same figures as for the Revolution Frequency pulses.)


Collision PatternRing 1 w.r.t. ring 2:

While for each ring bucket 1 and its revolution frequency are locked together, the two bunch patterns can be “rotated” with respect to each other so as to move or set the collision points This is usually done before the injection process with collision (and crossing) points defined before filling. This is the preferred method as it keeps crossing points fixed from injection to physicsIt can also be done after injection by changing the phase of one RF system (and thereby the corresponding beam) with respect to the other (Machine Development)

Convention 3: bunches in bucket 1 of the two rings collide in IP1 (or any other IP. To be agreed now…)


Clock GenerationRing 1 Ring 2

1/10 Divider

RF IN OUT

Sync

400 MHz RF 2

F rev 2 (orbit)

Bunch Clock 2

F rev 2 (orbit)

400 MHz RF 2400 MHz RF 2

1/10 Divider

RF INOUT

Sync

400 MHz RF 1

F rev 1 (orbit)

Bunch Clock 1

F rev 1 (orbit)

400 MHz RF 1

Fiber optic links to CCR (AB/CO+EXP), CMS

400 MHz RF1

VTUVTU(square wave)

(square wave)

(pulse train)(pulse train)

Fiber optic links to CCR (AB/CO+EXP), CMSSR4

(square wave)

(square wave)

All signals are generated in SR4

They are transmitted to CMS (point 5) and BT (dumps) directly

And they are transmitted to the other experiments and BI via CCR (optical splitter 1:4)

Injection pulses are sent directly to the kickers (not shown above)


Reference Clock Generation

The Ref Bunch Clock is a stable, constant frequency square wave at 40.078966 MHz (7 TeV proton) obtained by asynchronous division (1:10) of a 400 MHz Reference (400.789658 MHz). This 400 MHz Reference is generated by a commercial Low Noise Frequency Synthesizer

On the flat top (7 TeV), but before physics starts, the 400 MHz RF1 and RF2 (and the corresponding beams) are rephased onto this 400 MHz Reference. The Bunch Clocks 1 and 2 follow automatically

But the Ref Bunch Clock can still be off by 1 to 9 periods @ 400 MHz. To fix that the Ref Bunch Clock will be slowly phase shifted by the proper number of 400 MHz periods to bring it in phase with the bunch

When physics starts, the delay between the edge of the Ref Bunch Clock and the passage of a bunch will be thus fixed from run to run

Common (Rephasing)

7 TeV Synth

400 MHz Ref 1/10 Divider

VTU

Ref Bunch Clock

Synchro Module

Bunch Clock1

Ph

ase

shift

er

TDC

Cogging

Cogging of Ref 40 MHz to put it in phase with the bunch at 7 TeV: phase shift by 0

to 9 full RF periods at 400 MHz

To Beam Control system: Rephase both rings onto

this reference before physics

Bunch Clock 2

SR4

400 MHz Ref

Fiber optic links to CCR (AB/CO+EXP), CMS

(square wave)

(square wave)

400.789658 MHz


Availability of clocks during runssee also edms 628536 v1, TTC System Upgrade, S. Baron, Sept 29, 2005

1/35640 and 1/10 dividers are resynchronized at each filling, once, before the first SPS-LHC transferDuring resynchronization the output signal disappears for ~1 msSignal presence and validity is then guaranteed, from shortly before filling until beam dumpAfter beam dump it is not excluded that the RF manipulate its equipment (switch a crate off/on) and signal may disappear. Experiments should make sure that no manual reset is needed in that casePresence of signals is not guaranteed outside physics run due to maintenance needed on the RF equipment, or if basic control facilities are lacking. That applies in particular to shutdown periodsThe AB/RF group is responsible for the transmission up to the Fiber Optic receivers installed in the experiments (including on-call service). In return the experiments will make for an easy access to this equipment and help implement remote acquisition of the status of the Fiber Optic RX (optical power, presence and frequency of detected RF signal) so that AB/OP and AB/RF can quickly diagnose the transmission


Installation planning

RF Synchronization equipment being installed/commissioned in SR4

Expect all running by mid-April


RF Synchro: Functionalities

Synchronization of the SPS-LHC bunch into bucket transfer

A train of pulses at the Fiducial Frequency Fc (= FrevLHC/7 = FrevSPS/27 = 1.6 kHz) is sent to the SPS for synchronization

By displacing this train by x RF periods the transfer will end-up in bucket+x -> Mechanism to select the injection bucket. Transparent to the SPS

We have only one “hardware” bucket selector for both rings

It is re-synchronized on two different Frev for the 2 rings -> Mechanism to rotate one ring with respect to the other (before injection)

Generation of beam synchronous signals: 40 MHz bunch clocks, revolution frequencies, injection pulses


DDS1 DDS2F RF Prog 1 F RF Prog 2

WB

sw

itch

WB

sw

itch

Inj. F rev

F injRing 1

Ring 2

Ring 1

Ring 2

Pha

se s

hift

er

1/(7 h) divider

1/h Divider

RF IN

OUT

1/h DividerRF IN

OUT

SyncOUT

RF IN

Sync

Bucket nbr.

Shi

fte

d F

inj t

o S

PS

Del

aye

d F

c to

SP

S

SP

S e

xtr.

war

nin

g

Inj. Pulse generator

RF IN

OUT

Inje

ctio

n p

uls

e

Injection VETO

Sync

Ring 1

Ring 2

Ring 2

Ring 1

Next

Shut

Open

Inj. Phase

WB Demux

Inj.

Rin

g 1

Inj.

Rin

g 2

Fiber optic links to AB/BT

Fiber optic links to/from SPS FC

Ring 2

Ring 1

F rev prog1

F rev prog2

VTU

DUAL FREQ PRGM

BUCKET SELECTOR

INJ PULSE GENE

VTUVTU

This divider will adjust the relative position of ring 2 vs ring 1. It must be adjusted only once at start-up

This sets the RF frequencies before and at injection

By changing the programmable delay in the bucket selector one chooses the LHC bucket for transfer. It will be adjusted continuously by OP to change the collision point or for multiple-batch injections.

This module generates the pulses sent to the injection kickers

LHC LLRF RF Synchronization


Commisioning Procedure Label Bucket 1 Beam 1

RF sets all its dividers (VTUs) to offset 0 (ring 1)

Inject a single bunch pilot from SPS in ring 1

By definition: This pilot is in bucket 1

Get the pilot to make a few turns (OP)

Adjust delay in the Frev sent to the beam dump. Observe our signal at the beam dump location. Compare to bunch position. Adjust. (BT). The abort gap ends just before the passage of pilot (Convention 2)

Adjust delay in the 40 MHz/Frev clock received by the experiments (EXP). Best done by various EXP as RF has only 1 signal for all EXP but CMS


Commissioning Procedure Label Bucket 1 Beam 2

RF sets all its dividers (VTUs) to offset 0 (ring 2)

Inject a single bunch pilot from SPS in ring 2

By definition: This pilot is in bucket 1

Get the pilot to make a few turns (OP)

Adjust delay in the Frev sent to the beam dump. Observe our signal at the beam dump location. Compare to bunch position. Adjust. (BT). The abort gap ends just before the passage of pilot (Convention 2)

Adjust delay in the 40 MHz/Frev clock received by the experiments (EXP).


Problem: Convention 3 is NOT respected. Bunches in bucket 1 must collide in IP1 -> we must rotate ring 2 with respect to ring 1

BI or EXP measures the time difference of passages of the 2 pilots (ring 1 vs ring 2) In IP1 (or another IP)

The RF adjusts the VTU ring 2 thereby changing the collision point of the 2 pilots on next injection -> repeat until collision in IP (within +-1.25 ns). This adjustment does not change the position of the bunch ring 2 with respect to the Frev (orbit). (When triggering on Frev the bunch does not move).

Conclusion: The RF will adjust the relative positions of the 2 rings but we need a measurement from BI or EXP. This measurement must come from a PU that sees both beams.


FAQQ: Bunches in bucket 1 of both rings meet in IP1 and IP5. Can this be changed ?

A: No…but if we want to have single bunch collisions in another IP we should inject in a different bucket of ring 2 (for example) using the bucket selector (page 13). For example for collision in IP2: Inject pilot in bucket 1 ring 1, and pilot in bucket 1 + 35640/4 = 8911 ring 2 . Reminder: For its Phase Loop to work properly the RF must be told in what bucket the bunch is injected via the Bunch Mask Pattern.

By delaying the ring 2 injection bucket by ¼ turn we displace the collision point by 1 octant


Q: Should we count in 400 MHz bucket or in 40 MHz bunch clock?

A: The RF Bucket Selector counts in RF periods but most hardware is designed for a minimum 25 ns spacing between bunches. So we can use both indexing:

bucket 1 -> bunch clock period 1



….


or

Q: How about the 43 bunch scheme. Will bunches be numbered 1 to 43?

A: They should best be numbered in buckets (1, 811, 1621,…) or in bunch clocks (1, 82, 163, …)

11

10

bbc

Documents

Bunch Numbering