Strategy for the Beam Interlock System - PowerPoint Presentation

Slide1

Strategy for theBeam Interlock SystemUpgrade

Jan UythovenWith input from B. Puccio, C. Martin, S. Gabourin, I Romera Ramirez, D. Wollmann, M. Zerlauth

MPE Technical Meeting, 10 November 2016

Ideas for the Future Beam Interlock SystemJan Uythoven

With input from B. Puccio, C. Martin, S. Gabourin, I Romera Ramirez, D. Wollmann, M. Zerlauth

6th HL-LHC Collaboration Meeting, 14 – 16 November 2016, Paris

OutlineDefinition of the Beam Interlock SystemHL-LHC Beam Interlock System

Why change it ?Functional ChangesTechnical ChangesPresent developments: BIS 1.23Time lineConclusionsJan Uythoven HL-LHC Collaboration Meeting 2016

3

Definition of the

Beam Interlock System, BIS

Jan Uythoven HL-LHC Collaboration Meeting 2016

4

BIS

Dumping system or

Extraction Kicker or

Beam Stopper or

Beam source….

Actuators

Beam ‘Permit’

Signals

Σ(User Permit = “TRUE” ) => Beam Operation is allowed

IF one User Permit = “FALSE” => Beam Operation is stopped

The BIS has to be highly reliable and available

Beam

Permit

only

in hardware,

redundant

VHDL codes

The BIS has to

be

fast

: transmit dump

request

< 2

turns

The BIS has to be connected to many different types of users systems,

Standard connection boxes with standardised interface

The BIS is a generic system: same hardware is used for all accelerators at CERN, implementation in the injector chain ongoing

Ring architecture (LHC) or Tree Architecture (Linac4,

PSBooster

)

Implementation in the LHC17 Beam Interlock Controllers (BIC)200 User systems connected

4 loops: 2 Counter Rotating loops for each beam

Beam Dumping System included in the beam permit

loopFast:All in hardwareSystem response time to

local

BIC:

some

µs

Fibre

optics

for transmission

Jan Uythoven HL-LHC Collaboration Meeting 2016

5

HL-LHC Beam Interlock SystemWhy change it ?Jan Uythoven HL-LHC Collaboration Meeting 2016

6

BIS 20

years

old

BIS

30

years

old

2023

Present operational experience is very good

In operation in the SPS since 2006, in the LHC since 2008

Reliable:

No dump request was missed !

Post Mortem check after every dump confirming full redundancy and response time

Available

Unavailable only 3x: once BIC module failure and 2x optical loop transmission

Power Supply failures are the most common, but don’t require immediate intervention due to redundancy

However, at the time of HL-LHC the BIS

electronicswill be obsolete

HL-LHC Beam Interlock SystemFunctional ChangesIt is a logical ‘AND’

It has to be reliable and availableSolution in hardware, redundancy etc.Separate monitoring from safety functionsSystem can be remotely tested, including redundancy….System is modular and scalableTo be used in the complete accelerator chainEither in ring architecture or tree architecture

Masking of input depending on Classification of channels, only some can be maskedMasking only with ‘Safe Beam’

It needs to be fast (< 2 turns)Standard User input connection

Jan Uythoven HL-LHC Collaboration Meeting 2016

7

UNCHANGED

POSSIBLE CHANGES

-

Possible improvement on

availability

(optical part)



See later

-

More flexibility

in masking and

link to SBF

V2 project

?



See

later

Discussion about fast link

 See laterMore flexibility, connection of more users to one BIC  See later

HL-LHC Beam Interlock SystemFunctional ChangesIt is a logical ‘AND’

It has to be reliable and availableSolution in hardware, redundancy etc.Separate monitoring from safety functionsSystem can be remotely tested, including redundancy….System is modular and scalableTo be used in the complete accelerator chainEither in ring architecture or tree architecture

Masking of input depending on Classification of channels, only some can be maskedMasking only with ‘Safe Beam’

It needs to be fast (< 2 turns)Standard User input connection

Jan Uythoven HL-LHC Collaboration Meeting 2016

8

UNCHANGED

POSSIBLE CHANGES

-

Possible improvement on

availability

(optical part)



See later

-

More flexibility

in masking and

link to SBF

V2 project

?



See

later

Discussion about fast link

 See laterMore flexibility, connection of more users to one BIC  See laterNo fundamental change

in functionality

Some ‘details’ possibly to be fine-tuned

Does it need to be faster ?

New HL-LHC failure modes

See presentations in parallel sessionsM. Valette, K.

Sjobaek, H. BurkhardtFaster failures than in the present LHCCrab cavities, Beam-Beam kick, Firing of quench heaters 

COMBINED FAILURES

Halo cleaning, tails, witness bunches, …

Damage by fast failures

versus

detection of these fast failures

There might be a

necessity to

be FASTER

Fast detection (diamond BLMs or …)

Fast transmission (BIS)

Faster dump

Jan Uythoven HL-LHC Collaboration Meeting 2016

9

Courtesy A.

Santamaria

Bunch

at primary collimator

with two crab cavities failing

Faster BIS ?Local BIC detection time: some µs

Transmission  Can one gain? Shortest path in 99.9 % of the cases: < 100 µs «Guaranteed» transmission time for 99.9999 % of the cases: < 200 µsCan «gain» if the redundant link is doubled up by a dedicated link for equipment close to the beam dumping system in IP6

Crab cavities in IP1Primary collimators in IP7Is it worth it?Isn’t there something fundamentally wrong if we start to search for some µs

Jan Uythoven HL-LHC Collaboration Meeting 2016

10

Effective

link

Redundant

link

11

< 100 µs

(99.9 %)

> 80 µs for BLMs

<

89

µs

89

µs

~ 3 LHC turns after failure detection

<

200 µs

(99.9999%

Gain with additional abort gaps but high price in

lumi

Allow

asynch

dumps for

very

rare, identified cases

Most potential:

Study of failure modes

Criticality and likelihood

Faster detectorsTails / halo and witness bunchesGain by having a detection close to IP6Delay determined by distanceThis would require a dedicated link to the retrigger systemThink at least twice !Faster BIS ?

Do we need to connect New HL Users ?

New magnet circuits will go through PIC / WIC systems which function as concentratorsNew devices : some free inputs are availableCrab cavities, Wire compensator ( WIC), eLens

, crystal collimators, …Request is more from injectors where old concentrators like the MUGEF cease to exist

Make new concentrators. To be done by clients but can there be a generic solution ?More flexibility in the present layout can be consideredCheck configuration against data base at each injection / cycle etc.

Consider changing user input from current source to e.g. RS485

Longer distances

Jan Uythoven HL-LHC Collaboration Meeting 2016

12

Beam

Interlock Controller

7x

Maskable

(SBF)

7x

Unmaskable

3

x Single

Beam

4

x

Both

Beams

3x Single Beam

4x Both Beams

17 x in LHC

238 user inputs

Communication within BIS crate presently via VME busTechnology developed in the ‘70s, military standard, still widely used but old technology

Still supported by BE-CO, but until when? LS4 YES – LS5 ? – …Modern standard chosen by laboratories like ESS or DESY is µTCA or consider PXIAdapt to new CERN standard –

whatever this

will beMore compact and faster (not an argument for BIS)

Do we need the high speed communication links on the bus?

However, we want to safely connect our 200 users

with solid connectors…

HL-LHC Beam Interlock System

Technical Changes

13

Manager board

FPGA chip

(Monitoring part)

CPLD chip

(Matrix A

)

CPLD chip

(Matrix B)

Present

user interface

BIS 1.23Study and tests of Optical SystemThe optical loops around the LHC are ‘powered’ by the optical daughterboard called CIBO

It is a custom made optical transceiver realised with an Eled optical transmitter and a photodiode optical receiverJan Uythoven HL-LHC Collaboration Meeting 2016

14

CIBO

The drawbacks of the current CIBO are:

Very low power margin

Typical output power -19

dBm

Typical

receive sensitivity -28

dBm

LHC max fiber attenuation 6.5 dB (UA27

to UJ33)

Raw margin: 2.5 dB

!

High gain drift versus temperature

(no

feedback compensation circuit for

Eled

)

- 0.16 dB / °C

No monitoring of:

Real output power

Real receiver sensitivity LHC fibers attenuation LHC power

margin ?

CIBO  SFPJan Uythoven HL-LHC Collaboration Meeting 2016

15SFP daughterboard

CIBO daughterboard

100 % pin compatible

100% functionality compatible

The

S

mall

F

orm-factor

P

luggable (

SFP

) has

been studied for the BIS over the last few years and is a good candidate to replace the present

CIBO

High output power (-15

dBm

to -3

dBm

)Allow for monitoring of the optical powerPrecision presently being studied in the lab, code being written

Both the SFP as the CIBO (eLed) are made for high bandwidth data transmission and don’t work properly with a DC signalPresently for eLed: if beam NOT OK, “no signal”, badly definedStudy and testing of having frequencies

for beam OK (different per beam) and another frequency for beam NOT OK: well definedCompatible with present hardware of the beam dumping system

Will be tested with TSU

SFP Real Life TestingJan Uythoven HL-LHC Collaboration Meeting 2016

16Use the LHC during two years (2017 – 2018) to validate SFP solution under real conditions

We don’t touch the active BIS loops

We add new hardware:

17 new BIC crates (Elma VME crate)

2 new fibers all around the LHC

(not touching the present spares)

We link the two BICs by the active

BIC “local Permit” signals

Up to LS2, the behavior of the active loops and the test loops will be constantly compared

Foresee to integrate a TSU in the test loop for 2018

One could consider passing a lot of ‘extra’ information on the SFP loop

Is there a need ?

Separation of Safety and Monitoring

Time Line

Ongoing & future work and tests

Jan Uythoven HL-LHC Collaboration Meeting 2016

17

Define BIS 2

Studies in the lab

Install test LHC / PS / ??

Test in machines

Produce series

Install HL-LHC & …

Operation for HL-LHC & …

BIS 1.23

BIS 2.0

Study

eLed

alternative

Produce electronics etc.

Study SFP in the lab

Install SFP loop in LHC

Implement SFP monitoring

SFP test at LHC performance

Include TSU in SFP loop

If required and tests successful:

Install BIS 1.23 in LHC

Operation BIS 1.23

2023

ConclusionsFor HL-LHC the present BIS will need to be replaced by BIS 2.0 The prime reason is outdated components by 2026

Foreseen to take place in LS3, including injector chainFunctional requirements BIS 2.0 seem identical to BIS 1.0Dedicated fast link under discussion, but limited gain / necessityProfit of new system by considering some hardware changes:Optical system

 BIS 1.23 projectWeakest point present system, only concerns availability

Presently working on using SFP technology, test @ LHC following this EYETSStay with present VME bus? Long term support by BE-CO?Update User connectivity? More flexibility

These changes don’t affect the functionality

Two projects started: BIS 1.23 and BIS 2.0

Possibly BIS 1.23 implementation after LS2. If required !

Jan Uythoven HL-LHC Collaboration Meeting 2016

18