WBS 1.7: sPHENIX Minimum Bias Detector - PowerPoint Presentation

Slide1

WBS 1.7:

sPHENIX Minimum Bias Detector (MBD)

Mickey Chiu

MBD Group

BNL: MC, M. Lenz, R. PisaniRIKEN: Yuji GotoColumbia/Nevis: C.Y. Chi, B. Sippach

Lehigh U:

Rosi

Reed

PHENIX BBC in sPHENIX

Higher Magnetic FieldsGain drop with higher B-fieldMaintain good MB eff even when placed out further in zUpdate of ElectronicsMaintain Single PMT time resolution dt ~ 50 ps

in offline, and 120

ps

in the trigger

.

Provide clean min-bias triggering w/ vertex selection (+/- 10 cm)

Provide Centrality determination

Provide Start-timeProvide Reaction Plane

Challenges in

sPHENIX

Placement in sPHENIX

Considering z = 144, 200, 250, and 300 cm, ie

, from original PHENIX location to near flux return

z

(cm)

η

min

η

max

B

Z

(T)

Rel. Gain

AuAu

MB

Eff

(%)

pp

MB

Eff

(%)

144

3.0

3.9

1.110.0190392003.334.230.750.1589362503.564.450.500.588343003.744.630.320.98732

B-field Study

5

Study of PMT Gain vs (Field

S

trength, Angle, PMT HV)

w/ Paul

Gianotti

,

Camillia FayyaziField Strength: Up to 0.735 T but in old documentation it says safer to stop at 0.4 TEventually stopped at 0.5 TField orientation at proposed locations of BBC range from 2-5 degreesRan at 0 and 5 degreesPiLas Laser

Magnet region

DRS4

Readout

Hall

Proble

B-field Study Results

BBC in PHENIX runs at 30 (inner) or 40 (outer) pC/MIPNeed MIP peak to calibrate gain

Need to confirm expected max particles in a tube (was ~30 in PHENIX) to get required dynamic range

Single particle peak

PHENIX@144 cm

sPHENIX@300 cm

Eff -3%

AuAu

, -7% pp

sPHENIX@250 cm

Eff -2%

AuAu

, -5% pp

PHENIX

AuAu

200 GeV

7

Readout Design

sPHENIX

GTM

sPHENIX

DAQ

sPHENIX

BBC LL1

sPHENIXGL1

L1 Trig,

Modebits

, CLK

Use

sPHENIX

EMCAL digitizers, and build a transition board to convert single-ended PMT signal to 100 ohm differential that stretches PMT signal to be recorded by 60 MHz ADC

Two possibilities to satisfy the trigger and timing requirements. Both options require splitting the signal. One is a shaped pulse for energy measurement. The other

is a square ~1V discriminator signal (for timing at trigger level and

precisiong

timing in offline).

o

r is a ramp and hold on a comparator (TAC) that is input to the digitizer

sPHENIX

Digitizers (2 or 4 ADC boards)

Disc/

Shaper

BoardBBCPMTsIn WBS 1.5 + 1.6 ScopeIn WBS 1.7 ScopeL1 FEM Board

PMT Resolution Study

8

Used 20

ps

piLAS

laser studies going well

Got 43

ps time resolution for 33 p.e. at lower than nominal gainLaser

BBC

Testing sPHENIX Digitizer Timing

NEW ADC

Input a signal from Agilent generator, that is passively split, and look at time difference between the two “identical” signals from the NEW ADCs to determine the timing resolution of the NEW ADCs.

Signal generated is shown on the right, and is intended to get close to matching what will be generated by the new disc/shaper board.

We need to know the final shape to use!

Shape tested was only ~2 samples during rise to check if this would be good enough

sPHENIX

Digitizers Timing Res (Fall 2017)12/19/17

10

We created signal shape templates and did a full chi-2 minimization (MINUIT) fit to get the start time for each of the two pulses from the split cable

This can only be done in offline, and is the best possible resolution one could ever get

For the resolution at L1, the resolution depends on the algorithm and still needs to be studied

We put the two split (identical) signals into the same board and also into different ADC boards so that we could see if there was a contribution from the clock propagation

Getting sub 1

ps from same ADC board, and ~13 ps (= 18.6/√2) from different boards. Clock propagation jitter at 13 ps level?Some anomalous systematic timing difference (at Δt = 0.01 ns) in some events that needs to be studied

Same Board

Different ADC Boards

Anomalous times?

TDC ADC

Contributions to Timing ResolutionNB: Contribution to trigger resolution depends on algorithm to extract time

sPHENIX

Digitizer

DAQ

Discriminator

Shaper

D/S Board

L1 FEM

Board

LL1

Board

+

+

=

43

ps

?

13

ps

Simple

dCFDUsing simple digital Constant Fraction Disc algorithm (Leading Edge with linear interpolation and threshold set at 50% of pulse height), we can get ~75 ps

resolution using a DRS4 Evaluation Board

Easily implemented in FPGA for L1 trigger

More complicated algorithms (FIR with sliding windows)

contemplated

This is with DRS4. Will need to reconfirm with final electronics chain

clip

delay

Test pulse

gate

disc

Test pulse

Charge

injector

threshold

gated

TVC pulse

shaper

shaper

splitter

3 poles

20ns (each)

shaper

disc

gated

pulse

shaper

splitter

shaper

Test pulse

Charge

injectorthresholdCalorimeter 64 channel 60 MHZ ADCgatediscTVC pulse3 poles 20ns (each)shaperPMTPMTOne threshold DAC per channelOne TP Charge injector DAC per board Even/odd channel enable for TP8 PMT channel per board128 channel, 16 boards totalBeam clock2mm HM cable2mm HM cableNew Beam Beam electronics Block DiagramL1 Trigger primitive

L1 Trigger Event

The TVC

shaped

pulse will have multiple samples from 60 MHZ ADC

CY

Chi, W.

Sippach

Conclusions

Studies to confirm use of BBC in sPHENIX going wellLoss of only 3% in AuAu MB Eff, ~15% in p+pZeroing in on z = 250 cm (closer-in option) as the preferred placement of BBC.

MB

Eff better (particularly for

p+p

) and gain drop can be counteracted by HV increase of just 100 V.

Moving toward full electronics chain

Nevis to produce prototype disc/shaper board next year

Backup Slides

Pseudorapidity Coverage

Multiplicity doesn’t drop too quickly with pseudorapidityBut how well can PMTs survive in B-fields?

z

(cm)

η

min

η

maxBZ (T)144

3.0

3.9

1.11

200

3.33

4.23

0.75

250

3.56

4.45

0.50

300

3.74

4.63

0.32

PHOBOS

PHOBOS

MBD Eff for 200 GeV

Au+Au and p+pThe Au+Au MB efficiency doesn’t drop much when moving the BBC back (according to pure Hijing

)

A little worse in

p+p

(as a percentage), goes from 39% to 32%

p+p

efficiency matters mostly for effect on cleaning up central arm triggers, but for high pt it’s not needed or wantedNot too relevant for relative luminosityThe purely MB triggered events are mostly uninteresting events anyway

Au+Aup+p

MB Efficiency Loss in

Au+AuRan 10K Hijing events (100K PYTHIA6 events), and selected coincidence events, >1 tubes on each side (>0 tubes on each side

)

All efficiency loss is in just the peripheral collisions, as expected, where there are only a few participants

You wouldn’t a large MB efficiency loss since the extreme peripheral is just a few % of the collisions

All

Hijing

BBC@144 cmBBC@200 cmBBC@250 cmBBC@300 cm

Properties of BBC Mesh Dynode PMTs

BBC used Hamamatsu R6178, similar to R5505. (R6178 discontinued and I couldn’t find the datasheets)In PHENIX, we ran BBC in ~0.2-0.3 T field, where gain drops 20% from 0-field runningThere’s possibly some room to compensate for lost gain in higher field in sPHENIX by increasing HVOne could possibly run at low gain in high B-field, but then compensate with CMOS amplifier (x100). Only concern would be if timing resolution can still be as good.

Field in sPHENIX

At flux return (z=300 cm), gain drops only by 10% from PHENIX (70%)At z=250, gain drops to ~50%Can compensate by increasing HV by ~100-200 VAt z=200, gain drops to 15%.HV increase probably not tenable (if we want to maintain safe running conditions)Could use CMOS amplifier. “Easy” to get gain of 8 without sacrificing signal quality

.

Fields angles are small, from 1-4 degrees (right figure)

@ r=10 cm

B

Z

B

R