Fast BLM acquisition system

Fast BLM acquisition system

Fast BLM acquisition system, B.Dehning 1

Bernd DehningCERN BE/BI

Plots are taken from:

Tobias Baer, Henrik Janson, Maria Hempel, Elena

Castro, Christoph Kurfuerst

BI-TB, 03. 04. 2014

Content

The diamond detector CERN installations Specification

Signal versus time Arrival time histogram

Acquisition systems

BI-TB, 03. 04. 2014 Fast BLM acquisition system, B.Dehning 2

IC Diamond PM (ACEM) Response time: 100 ns < ns ns Pulse duration: 100 us 5 ns few ns Dynamic 9 orders 9 orders 3 orders

Radiation tolerance 100 MGy several 10 MGy 100 kGy Volume 1l 0.1 cm3 100 cm3


Beam loss sensors

MGy

Signal degradation factor 10

50 ns bunch spacing

Risetime: 2.34nsFalltime: 10.34nsAmplitude: 397mVBackground noise: 9.9mVTemporal width: 6.06ns

CVD detector superior in terms of response time,

pulse duration and dynamic range

Leakage current

Max 20 pAOften it is the case that the acquisition chain is limited by the sensor; diamond sensor (CVD) exploitation requires high performance electronics

BLMED Installation Overview

Acc IP Position Detector

LHC 2 BLMED.04L2.B1I10_TCTVB.4L2 CIVIDEC

LHC 3 BLMED.06L3.B1I10_TCP.6L3.B1 CIVIDEC

LHC 3 BLMED.06R3.B2E10_TCP.6R3.B2 CIVIDEC

LHC 6 BLMED.04L6.B2I10_TCSG.4L6.B2 CIVIDEC

LHC 6 BLMED.04R6.B1E10_TCSG.4R6.B1 CIVIDEC

LHC 7 BLMED.06L7.B1E10_TCHSS.6L7 CIVIDEC

LHC 7 BLMED.06R7.B2I10_TCHSS.6R7 CIVIDEC

LHC 8 BLMED.04R8.B2C10_TCTVB.4R8 CIVIDEC

SPS 4 MSE 41876 CIVIDEC

SPS 6 TPSG 61773 CIVIDEC

LHC 2 BLMED.04L2.B1C10_TDI.4L2.B1 BCM1F4LHC

LHC 8 BLMED.04R8.B2C10_TDI.4R8.B2 BCM1F4LHC

LHC 4 BLMED.05L4.B1C10_BGI BCM1F4LHC

LHC 4 BLMED.05R4.B2C10_BGI BCM1F4LHC

LHC 5 BLMED.04L5.B1I10_TCTVA.4L5.B1 BCM1F4LHC

LHC 5 BLMED.04R5.B2I10_TCTVA.4R5.B2 BCM1F4LHC

• Following detectors are foreseen to be installed during LS1 and LS2:

• 1 detector for Fast Servo spill in the SPS TT20 (LS1)• 1 detector for HiLumi in the LHC (LS1)• 8 detectors for Booster dump (LS2)• 10 (12) Detectors for PS (LS2)

BI-TB, 03. 04. 2014 4Fast BLM acquisition system, B.Dehning

Location of Diamond Detectors at CERN Accelerators

31.10.2012 Status and Application of Diamond BLMs; B.Dehning 5

Location of Diamond Detectors at CERN Accelerators


Optical diamond detector

(BCM1F4LHC)

• All detector signals arrive at the same position

• Easy access of data acquisition system

• Lower dynamic range



Tunnel set ups

CIVIDEC system

CMS DESY/Zeuten system

Optical fibre

Technical Specification of Fast DAQ

Analogue bandwidth DC – 500 MHz V In min 10 mVpp V In max 10 Vpp 7V RMSAcquisition: Resolution 8/10/12 bit Solution specificSampling rate 1Gs/s (2.5ns – 1ns)Sampling length 10ms -1.0s @ 1Gs/s per channel selectable

Buffer @ 8 bit10k – 1G (points) = 240kbit – 24Gbit (4Gbyte) 4 Channels

Buffer@ 12 bit 10k – 1G (points) = 360kbit – 36Gbit ( 3 Channels Trigger system: Signal input Edge (Threshold set with 10bit DAC) Min. 5mVExt. Input 2 Edge TTL signalSoftware trigger On event via data transmissionHistogram: Resolution 1.55671875ns (57024 bins) Bunch length / 16Trigger threshold min. 5mV 10 bit DACCounter 20-24 bit per bin Update rate 1s Scaler: Trigger threshold min. 5mV 10 bit DACCounter 20-24 bit Update rate 1s Controller or FPGA software: Source code User accessible for modifications Preferable Drivers & Control software Driver Open source Linux PreferableControl Linux Preferable


Specification

Sampling frequency, bandwidth 500 MHz analogue bandwidth relates to a rise time of 0.7 ns (10 – 90%) Rise time observation 2.3 ns, limited by 1GHz sampling frequency and

cable, rise time contribution from analogue part of acquisition almost negligible


PS SPS LHCDynamic [orders of magnitude ] 5-6 5-6 5-6

Sampling frequency [GS/s] 1 1 1

Number of value to be stored per channel

2E6 22E6 1E9

Synchronisation with machine events

y y y

Signal versus time acquisitions


September, 18th 2012 12Tobias Baer

Diamond Measurement 4 batches

4·36 bunchesLHC injection gap

36 bunches

SPS injection gaps


Diamond Measurement 1 batch

36 bunches50ns spacing between bunches

All bunches contribute to the beam losses, as expected for a macro particle interaction.

LHC Machine CommitteeAugust, 2nd 2012 14

Event Sequence

Start of B1 losses in IR7.

0

Fire M

KD.B

2

8/8 turns

IR1

IR6

IR7

IR5

IR3

IR4

IR2 IR8

3/8

No IR5

bb ki

ckLo

sses

in IR

71/8

Pacman structure of beam losses.


Event Sequence

Losses due to uncaptured beam in abort gap duringMKD rise time observable in IR7.

0

Fire M

KD.B

2

8/81/8

IR1

IR6

IR7

IR5

IR3

IR8

IR4

IR2

3/8

No IR5

bb ki

ckLo

sses

in IR

76/8

No IR2

, IR1,

IR8 b

b kick

Dump l

osse

s


Event Sequence

Losses due to uncaptured beam in abort gap duringMKD.B1 rise time observable in IR7.

0

Fire M

KD.B

2

8/81/8

IR1

IR6

IR7

IR5

IR3

IR8

IR4

IR2

3/8

No IR5

bb ki

ckLo

sses

in IR

76/8

No IR2

, IR1,

IR8 b

b kick

Dump l

osse

sB2

dum

ped

Fire M

KD.B

1

10/8

Dump l

osse

s

Bucket counting on consecutive turns


MKI UFO dump which illustrates the timing. The data is acquired by the IR7 diamond BLM for B1. The yellow line shows the losses during the last turn incl. the spike due to the MKD rise time. As reference in red the losses around the abort gap in the previous turn. One can see that the losses which are observable in IP7 occur about 875ns after the beginning of the abort gap and about 1.8us after the last bunch.

SPS 5 ns bunch structure measureable with 150 m of Cu cable


CNGS - SPS extracted pulse

10 us 5 ns


Ring BLM Measurements

Spatial loss profileUFO Location: BSRT.05L4.B2

Temporal loss profile

On the following slides: Measurements with BMLED.06R7.B2I10_TCHSS.6R7.B2. 40dB signal amplification.

B1B2

UFO location

Diamond BLMB2 direction


Diamond Measurement Overview

1 turnLosses due to beam dump


Diamond Measurement 1 turn

1 turnLosses due to beam dump

Beam abort gap

4·36 bunches2·36 bunches

The data was taken during the EOF test of the beam-beam MD on 13.12.2012, where B2 was dumped first, which led to a coherent oscillation of B1 due to the missing LRBB deflections.

The frequency resolution is limited, because the losses were acquired for only ~200 turns.


Bunch by bunch tune measurement

The duration of recording (1s) determines

the required memory size

(IEC team, T. Pieloni)

Arrival time histogram referenced to revolution period


LMC 10.02.2010 B.Dehning 24

HERA Scraping Beam with Wires HERA

Bunch spacing 96 ns Samples taken every 21 ns

No coasting beam halo 3 samples with zero intensity

(left, top) Partially coasting halo

Inter bunch samples measure

intensity (left, middle) Only coasting halo No observation of bunch structure

(left bottom) Strong coasting component at

begin of fill (centre top) Bunched halo only minutes

later (centre, middle) With retracted wire diffusion

of coasting component into

cleaned area (centre bottom),

SPS

Effect caused by RF system

noise, faulty amplifier

Halo

Inte

nsi

ty

Inner wire

Outer wire

Outer wire

wire ~ 4 sig

wire ~ 4 sig

wire ~ 6 sig

begin fill

4 min later

wire retracted


Arrival time histogram IP5 CMS system

Allows details of bunch spacing

With 25 ns spacing and debris from

the IP spacing reduced to 12.5 s


Arrival time histogram with simulated and CVD signal

Bunch period

signal at input of

acquisition system

CVD detector

signal at input

acquisition system

Arrival time histogram during ramp


Bunch spacing reduced due to cross talk

Sub 25 ns resolution required


Arrival time histogram - Coasting beam and instabilities

By continues updated display observation of all bunches possible


Arrival time histogram with the CMS system

IP2, 5 and 8 losses from tails and experiment

IP4 losses from core of beam

Arrival time histogram & injection losses


Before injection



After injection



later after injection

Clean injection check should also be possible at the PS and SPS

New Diamond Detector Cascade

Schemes

• Detector signal split in two signal paths (direct and 40dB amplified)

• Monitoring of complete dynamic range possible

• No need to access tunnel to exchange amplifier for different users


tunnel side tunnel

System propositions • 8 - 10 - 12 bit

solution possible

• 3 – 4 channel per system needed

• Splitter and/or optical input

• can be separated form DAQ

• In case of a modular system more than 4 channel possible

• Parallel processing of histogram and loss versus time mode preferable


System Comparison


OASIS 250kbyte 8 bit 1-2 GSPS

No Yes No The FPGA can not be programmed

SystemImportant Technical Aspects BI/BL Man power Dead

lineComments

Storage per Ch

Resolution Rate FPGA CPU

Production

BI/BL 8 MByte 10 bit 1 GSPS Yes Yes Yes No Tender for FMC neededTender

1 1.25 GByte 10 bit1.25 GSPS No Yes No No Offer is without chassis and CPU

Tender 2 2 GByte 12 bit

1.6 GSPS Yes Yes No Yes

Modular system, Onsight consualtion

Tender 3 4 GByte 8 bit

1.25 GSPS No No No Yes QT GUI exist, Future option 12 bit

Tender 4 1.5 GByte 12 bit 1 GSPS No No No Yes Completely new development

Tender 5 250 MByte 12 bit 1 GSPS No No No Yes Future option 8 Gbyte



CMS system


IP4 abort gap recording

Cross talk from other beam dominant


Fast BLM acquisition system

Documents

application of diamond

acquisition chain

channelstrigger system

ns ns nspulse duration

detector signals

fast servo spill

ns bunch spacingrisetime

ps ls2 bitb