33RD INTERNATIONAL COSMIC RAY CONFERENCE, RIO DE JANEIRO 2013THE ASTROPARTICLE PHYSICS CONFERENCE

Status of the new Sum-Trigger system for the MAGIC telescopesJ.R. GARCIA∗,1,3, F. DAZZI2, D. HAFNER1, D.HERRANZ4 , M. LOPEZ4, M. MARIOTTI2, R. MIRZOYAN1 , D.NAKAJIMA1 , T. SCHWEIZER1, M. TESHIMA1.1 Max Planck Institut fur Physik.2 University of Padova, INFN Sezione di Padova.3 Instituto Astrofisico de Canarias.4 Universidad Complutense de Madrid.∗ Presenting author

jezabel@mpp.mpg.de

Abstract: MAGIC is a stereoscopic system of two 17 m diameter Imaging Air Cherenckov Telescopes (IACTs)for γ-ray astronomy. Lowering the energy threshold of IACTs is crucial for the observation of Pulsars, high redshiftAGNs and GRBs. A lower threshold compared to conventional digital trigger can be achieved by means of a novelconcept, the so called Sum-Trigger, based on the analogue sum of a patch of pixels. The Sum-Trigger principlehas been proven experimentally in 2007 by decreasing the energy threshold of the first MAGIC telescope from55 GeV down to 25 GeV. The first VHE detection of the Crab Pulsar was achieved due to this low threshold. Afterthe upgrade of the MAGIC I and MAGIC II, a new Sum-Trigger system will be installed in both telescopes inSummer 2013. The expected trigger threshold in stereo mode is about 25÷ 30 GeV. It is a an improvement overthe existing threshold (about 50 GeV) of the digital trigger. We will report about the current status of the project.

Keywords: Magic, Hardware, trigger, analogue, Sum-Trigger, Crab.

1 IntroductionOne of the main attempt of γ-ray Astrophysics detectordevelopment is to cover the energy range from a few GeVto 80 GeV, since this region is difficult to access, for both,space-based and ground-based instruments. There are strongscientific motivations to extend the IACT technique below80 GeV. The main goal is to study transient objects suchas high redshift AGNs and GRBs, which are not easilydetectable by satellites due to the low flux. In addition,lowering the threshold also favourites both the study of EBLand the measurements of the spectra of VHE Pulsars [2].

Therefore, the development of new strategies to decreasethe trigger thresholds and increase the sensitivity in this gapis crucial. In 2008, MAGIC already succeeded in this senseby developing a prototype analogue Sum-Trigger (singletelescope) that significantly lowered the trigger thresholddown to 25 GeV. This trigger system achieved excellentresults, since it allowed the first detection of very highenergy pulsed gamma radiation from the Crab Pulsar [2].

Using the experience acquired during the constructionand operation of the first prototype, a new analogue trigger,so-called Sum-Trigger-II, was designed and developed. Thispaper aims to explain the concept, development, and statusof this system.

1.1 Description of the MAGIC experimentThe MAGIC telescopes are located at a height of 2200 ma.s.l. on the Observatorio del Roque de los Muchachos(28◦N, 18◦W) on Canary Islands. It is a stereo systemcomposed of two 17 m diameter (f/D ∼1) IACTs for VHEγ-rays observation. Both telescopes are built using thesame light-weight carbon-fibre structure, which allows for arapid repositioning time necessary for observations of shortphenomena, such as GRBs. The camera of each telescope iscomposed of 1039, 0.1◦ hexagonal pixels (PMTs), it has a3.5◦ field of view and a trigger area of 1.25◦ radius. In each

pixel the signals from the PMTs are optically transmitted tothe counting house where the trigger and the digitization ofthe signals take place.

Since 3 years, regular observations are performed instereoscopic mode using the standard trigger system, com-posed of three stages. The level-zero trigger sets the dis-criminator threshold in real time for each pixel in the triggerregion, excluding the electronic noise and part of the NightSky Background (NSB). Each telescope separately has alevel-one digital trigger with the 3 next neighbour (3NN)topology, it means that it only selects close compact eventsthat include at least 3 next neighbouring pixels. Only eventsthat trigger both telescopes are recorded. The stereo trigger,level-three, makes a tight time coincidence between bothtelescopes.

Aiming for the lowest possible threshold, as an alterna-tive to the current standard trigger (level-zero plus one), theSum-Trigger-II will be installed by end of this Summer.

1.2 Low energy observations with IACTsThe physics principles of cascades in extended air show-ers are the same in all the energy ranges, but the observa-tion of low energy cascades is more difficult due to the factthat less energy gets converted while the particles are cross-ing the atmosphere. The cascades reach their maximum de-velopment around a depth of 180-240 gr/cm2 (11÷ 12 kma.s.l.), very far away from the position of the IACTs thusa large part of the shower photon signal is absorbed in theatmosphere. Adding the fact that the energy threshold forthe Cherenkov photon production is higher in the upper lay-ers of atmosphere due to reduced air density (35÷ 40 MeVfor electrons), the number of emitted photons is further re-duced. On top, the higher is the altitude, the lower is the re-fraction index and thus, the smaller is the Cherenkov angle,and the generated photons are strongly collimated along thecharged particles trajectory. In the case of showers initiated

arX

iv:1

404.

4219

v1 [

astr

o-ph

.IM

] 1

6 A

pr 2

014

MAGIC: Sum-Trigger-II33RD INTERNATIONAL COSMIC RAY CONFERENCE, RIO DE JANEIRO 2013

by γ-rays, which have a modest transversal development,the main consequences are small shower projections (smallsize) and a limited number of triggered events at large im-pact parameters.

Monte Carlo simulations level of the morphology oflow energy events (∼10÷ 60 GeV) have been done at thecamera [3]. These studies revel that most of these events areconcentrated in a thin doughnut shaped region around thecamera center. Although the primary shower energy is verylow, the photons span on a wide region. Fig.1 shows threetypical gamma showers between about 13 GeV and 60 GeV.However, the photon density in most of the single pixels isvery low (≤ 10 photons, violet and blue pixels) and only fewof them contain enough signal to be detected (green, yellowand red). Usually, up to 25 GeV, 2 or 3 groups or islands ofsignals are observed. At higher energies these islands startto overlap and the ovoid like image of the Cherenkov lightis recovered.

Fig. 1: Distribution of photons (black points) at cameralevel for three different γ-ray energies, from left to right:12.9 GeV, 21.5 GeV & 58.6 GeV. Adapted from [3].

All these studies led to the construction of a new trig-ger system, the Sum-Trigger, particularly sensitive to lowenergy cascades.

2 The Sum-Trigger-II2.1 The conceptThe basic principle of the Sum-Trigger-II is to add up thesignals from a group of neighbouring pixels (macrocell)and then apply a threshold to the summed signal. Forthis purpose the trigger region has been covered withthree overlapping layers of macrocells, see Fig.2. MonteCarlo studies suggested that the optimum solution is to useroundish macrocells composed by 19 pixels (of size 0.1deg), then a macrocell contains most part of a low energyshower image.

One of the advantages of this technique is that it improvesthe signal to noise ratio (shower signal to NSB), since allpixels within a macrocell contribute to the trigger decision,i.e. it uses small photon signals below the single channelthreshold (below which the standard trigger is sensitive to).

Besides the simplicity of the idea, the Sum-Trigger-IIconstruction has required the development of fast, highprecision, analogue electronics. The PMTs in the camerasdo not have exactly identical properties, but there aredifferences in signal propagation times, gains and pulsewidths.In addition, these different signal delays will be worseneddue to small length differences in the 162 m long opticalfibres that transport the signals from the camera to thereadout electronics. The skew must be minimized in thewhole trigger area to assure a proper pile up among signals.

The pulse width itself is perfectly optimized for thetemporal evolution of a shower core started from a low

12

34

56

7

89

1011

12

13

1415

1617

18

19

2021

2223

2425

26

27

28

2930

3132

3334

35

36

37

3839

4041

4243

4445

46

47

48

49

5051

5253

5455

5657

58

59

60

61

6263

6465

6667

6869

7071

72

73

74

75

76

7778

7980

8182

8384

8586

87

88

89

90

91

9293

9495

9697

9899

100101

102103

104

105

106

107

108

109

110111

112113

114115

116117

118119

120121

122

123

124

125

126

127

128129

130131

132133

134135

136137

138139

140141

142

143

144

145

146

147

148

149150

151152

153154

155156

157158

159160

161162

163

164

165

166

167

168

169

170171

172173

174175

176177

178179

180181

182183

184185

186

187

188

189

190

191

192

193

194195

196197

198199

200201

202203

204205

206207

208209

210

211

212

213

214

215

216

217

218219

220221

222223

224225

226227

228229

230231

232233

234235

236

237

238

239

240

241

242

243

244

245246

247248

249250

251252

253254

255256

257258

259260

261262

263

264

265

266

267

268

269

270

271

272273

274275

276277

278279

280281

282283

284285

286287

288289

290291

292

293

294

295

296

297

298

299

300

301

302303

304305

306307

308309

310311

312313

314315

316317

318319

320321

322

323

324

325

326

327

328

329

330

331

332333

334335

336337

338339

340341

342343

344345

346347

348349

350351

352353

354

355

356

357

358

359

360

361

362

363

364

365366

367368

369370

371372

373374

375376

377378

379380

381382

383384

385386

387

388

389

390

391

392

393

394

395

396

397

398399

400401

402403

404405

406407

408409

410411

412413

414415

416417

418419

420421

422

423

424

425

426

427

428

429

430

431

432

433

434435

436437

438439

440441

442443

444445

446447

448449

450451

452453

454455

456457

458

459

460

461

462

463

464

465

466

467

468

469

470471

472473

474475

476477

478479

480481

482483

484485

486487

488489

490491

492493

494495

496

497

498

499

500

501

502

503

504

505

506

507

508

509510

511512

513514

515516

517518

519520

521522

523524

525526

527528

529530

531532

533534

535

536

537

538

539

540

541

542

543

544

545

546

547

548549

550551

552553

554555

556557

558559

560561

562563

564565

566567

568569

570571

572573

574575

576

577

578

579

580

581

582

583

584

585

586

587

588

589

590591

592593

594595

596597

598599

600601

602603

604605

606607

608609

610611

612613

614615

616617

618

619

620

621

622

623

624

625

626

627

628

629

630

631

632633

634635

636637

638639

640641

642643

644645

646647

648649

650651

652653

654655

656657

658659

660661

662

663

664

665

666

667

668

669

670

671

672

673

674

675

676

677678

679680

681682

683684

685686

687688

689690

691692

693694

695696

697698

699700

701702

703704

705706

707

708

709

710

711

712

713

714

715

716

717

718

719

720

721

722723

724725

726727

728729

730731

732733

734735

736737

738739

740741

742743

744745

746747

748749

750751

752753

754

755

756

757

758

759

760

761

762

763

764

765

766

767

768

769

770771

772773

774775

776777

778779

780781

782783

784785

786787

788789

790791

792793

794795

796797

798799

800801

802

803

804

805

806

807

808

809

810

811

812

813

814

815

816

817

818819

820821

822823

824825

826827

828829

830831

832833

834835

836837

838839

840841

842843

844845

846847

848849

850

851

852

853

854

855

856

857

858

859

860

861

862

863

864

865

866867

868869

870871

872873

874875

876877

878879

880881

882883

884885

886887

888889

890891

892893

894895

896897

898

899

900

901

902

903

904

905

906

907

908

909

910

911

912

913

914915

916917

918919

920921

922923

924925

926

927928

929930

931932

933934

935936

937938

939

940

941

942

943

944

945

946

947

948

949

950

951

952

953954

955956

957958

959960

961962

963964

965

966967

968969

970971

972973

974975

976977

978

979

980

981

982

983

984

985

986

987

988

989

990

991

992993

994995

996997

998999

10001001

10021003

10041005

10061007

1008

1009

1010

1011

1012

1013

1014

1015

10161017

10181019

10201021

10221023

10241025

10261027

10281029

10301031

1032

1033

1034

1035

1036

1037

1038

103928

3015

12

27 42

45

47

18

3

9

36

53

317

1

2349

55

22

24 37

3911

35

25

8

46

412

43

50

14

19

51

4

32

3416

33 44

21

2048

10

2954

13

17

385

40 52

26

6

Fig. 2: Layout of the trigger region of Sum-Trigger-II,where each color represents a different layer. 19 pixelhexagonal macrocell shape was selected to guarantee botha symmetrical overlap and an angular symmetry.

energy gamma event. If it is too small then the signal ofthe shower does not sum up at the same time, if it is tooslow then one integrates too much night sky noise. Theoptimal pulse’s width was determined to be between 2.5 nsand 3.0 ns FWHM.

The best threshold level for a patch of 19 pixels with anacceptable ratio between accidental NSB and lowest triggerenergy threshold is around 24÷ 27 PhE.

A drawback of classical PMTs is so-called afterpulsenoise, which are random noise pulses of high amplitude andhigh rate. Such afterpulses could trigger the Sum-Trigger-II and need to be vetoed. The simple way of reducingthe effect of those afterpulses is to clamp the signal ata certain (optimized) amplitude. For the PMTs that areused in MAGIC the optimum is around 6÷ 8 PhE clippinglevel for the single telescope (corresponding to avoidingtriggers in a three-fold afterpulse coincidence). In stereomode (coincidence between two telescopes) this clippinglevel can be relaxed (increased) which further improves thesensitivity at lowest energies.

The Monte Carlo studies (Fig.3) show the improvementon the sensitivity and the lower threshold (around 30 GeV)respect to the 3NN digital trigger already stated [3].

Energy (GeV)20 40 60 80 1000

5

10

15

20

25

30

Comparison stereo energy threshold

30 GeV

47 GeV

Sum-Trigger-IIDigital Trigger (3NN)

Diff

eren

tial r

ate

(a.u

.)

Fig. 3: Monte Carlo simulations show the threshold com-parison between the digital trigger and the Sum-Trigger-II.

MAGIC: Sum-Trigger-II33RD INTERNATIONAL COSMIC RAY CONFERENCE, RIO DE JANEIRO 2013

2.2 Functional set-up of the overall triggersystem

The Sum-Trigger-II system is structured in several sub-systems (see Fig.4). The signals first arrive to so-calledClip-boards. After delaying, adjusting amplitude and clip-ping, the signals differentially propagate over the backplane(so-called Sum-backplane) to Sum-boards. There, they aresummed up and passed to a discriminator. Finally, the dig-italized signals propagate, again over the Sum-backplane,to a computer control board that also contains an FPGA,which counts the macrocells triggers and perform a glob-al OR that generates the final trigger signal. This over-alltrigger is connected to the MAGIC telescopes trigger coin-cidence logic (level 3) and from there to the readout.

Fig. 4: Sum-Trigger-II schematics. Each channel has ananalogue delay adjustment, an amplitude adjustment and aclipping mechanism. The sum of 19 channels is fed into adiscriminator.

2.3 The Clip-boards and the Analogue DelayModules

The Clip-board is a 9U mixed digital/analogue board wherethe necessary corrections to the signal are performed (seeFig.4, light orange shaded boxes).

• Delay: To correct individual PMTs timing differencesand fibres delays. The delays range is about 6.5 ns.

• Attenuation: To obtain perfect flat-fielding of allchannel gains. Besides the differences on gains arenormally small, we can apply attenuation from 0 to-32 dB.

• Clipping: To avoid that the PMT’s afterpulses gener-ate fake triggers. Monte Carlo shows that this is themain source of noise [3].

Because of ultra high channel density, it was necessaryto incorporate the three above functions in so-called delaymodules, of which 32 units are plugged onto the Clip-board.This is the most crucial component of the system. It containsa newly developed, adjustable analogue signal delay module.

Fig. 5: Analogue Delay Module.

The analogue delay lines have been completely tested,showing high precision during timing, amplitude and clip-ping adjustment. The temperature dependency is negligibleand the signal’s integrity is acceptable. The output pulsewith maximum delay shows just a slight widening of 0.2 nscompared to the input signal [4]. At this time, mass produc-tion and quality control of the delay lines is taking place.

There are 18 Clip-boards for each Sum-Trigger (for eachtelescope). The main components of the Clip-board arethese 32 programmable delay lines, and the CPLD (MAX-II, Altera) that allows the control of the board and the delaymodules.

Fig. 6: Picture of a Clip-board equipped with delay lines.

The gain can be set with a precision below 5% in adynamic range from -15 dB to +8 dB. The delay (pulseposition) can be adjusted with a resolution of 200 ps. Themeasured RMS noise is ≤ 0.05 PhE and the crosstalk is ≤1%, thus fulfil by far the requirements.

2.4 The Sum-backplaneThe Sum-backplane is a passive 10U printed circuit mother-board which connects the Clip-boards to the Sum-boards.From the electronic point of view, this is the most complexpart of Sum-Trigger-II. The PCB layout has been developedconsidering that 997 fast differential analogue signals haveto be routed to 55 macrocells, preserving the isochronisminside 50 ps, a bandwidth higher than 650 MHz and a lowcross-talk (≤ 1%) [3]. Both Sum-backplanes have alreadypassed the electronic test and it fulfils the requirements.

2.5 The Sum-boardsThe Sum-board (Fig. 4, dark orange. Fig.7) is the mainboard for the final event selection. It can handle up to threemacrocells, covering the whole mapping with only 19 units,still keeping the compactness of a small 3U card. It is ahigh-speed mixed analogue and digital system, which canbe controlled automatically by the computer control board[3].

Fig. 7: Sum-board.

The performances of the Sum-board meet the require-ments. The bandwidth is around 550 MHz and its flatness isinside 0.5 dB up to 100 MHz. The peak to peak electronicsnoise is quite low, similar to 0.6 PhE. The cross-talk is lessthan 1% up to 300 MHz. The maximum skew in the ana-logue part (pile-up synchronization) is 80 ps and its linear-ity is perfect up to 3.7 V. The comparator threshold has aresolution of 0.1 PhE, a perfect linearity and stability.

The digital trigger output signal goes to the Astro-boardwhere the final decision will take place.

MAGIC: Sum-Trigger-II33RD INTERNATIONAL COSMIC RAY CONFERENCE, RIO DE JANEIRO 2013

2.6 The computer control BoardThe Astro-board (see Fig.4, blue boxes) is a fully digital 9Uboard. It is the final stage of the Sum-Trigger-II electronicschain. It contains a Cyclone IV FPGA and a FOX-G20Linux computer. The combination of a FPGA with theLinux computer is very powerful and also simple. Thecomplete Sum-Trigger control program communicatingwith the MAGIC central control is running on this computer.

Fig. 8: Astro-Board. The board contains an FPGA (AlteraCyclone IV) and an embedded Linux PC (FOX-G20),bottom right corner of the board.

It take cares of:

• The FPGA manages 55 macrocell triggers and adjuststhe thresholds. At the same time, it merges these localtriggers through a global OR, finally triggering theMAGIC-DAQ.

• The trigger rate is kept quite stable and below the lim-its, adjusting continuously the discriminator thresh-olds of the Sum-boards. The correct parameters arecomputed by the embedded PC.

• Interface with Central Control Software: The Sum-Trigger gives reports and can be control by the Cen-tral Control of the telescopes thanks to this embeddedcomputer.

• Automatic calibration of delay, amplitude, and clip-ping can be perform since the PC can control the ap-plied voltages on the Clip-board pins, so it can ful-ly control the delay lines. An automatic calibrationprocedure is under development.

2.7 Calibration processDue to occasionally required tuning of the high-voltagesapplied to the PMTs, the amplitude and signal transit timecan change for individual pixels and have to be re-adjustedon a regular basis for optimal trigger performance. Hence,Sum-Trigger-II includes a computer controlled automaticadjustment of amplitude and delay of the signals. In order tokeep the complexity of the new circuits low, an innovativemeasuring technique based on the evaluation of a seriesof rate measurements is introduced, requiring only veryfew additional electronics. In particular, the discrete triggeroutput of the discriminator is used to measure amplitudes bycounting the number of events that exceed the discriminatorthreshold within a certain time span. Similarly, the ratesto determine the delay of each channel are derived fromthe output of a D-type flip-flop that is used to compare thearrival times of two pulses.When performing the gain adjustment, the discriminator

threshold is fixed to the target amplitude level, and theattenuation value is varied while counting the number oftrigger signals from the discriminator. Likewise, the optimaldelay is derived by tuning the delay line module. Here, thecounter is incremented by the D-type flip-flop comparingthe signal arrival time with the timing reference channel.The result is a series of rate measurements of the transitionregion from maximum to minimum number of triggercounts. Due to the time and amplitude jitter inherent inthe signals, the rate scans show a cumulative distributionfunction, which is used to derive the optimal settings, beingfound at 50% of the maximum rate (Fig.9).

Fig. 9: Principle of the measurement process. Here, “value”means gain or delay, depending on the property calibrated.

3 SummaryThe analogue Sum-Trigger is a new trigger concept that hasbeen invented, developed and tested on the single MAGICtelescope. It was designed to lower the trigger thresholddown to 25 GeV. The proof of principle was the detectionof the Crab pulsar in 2008.The Sum-Trigger-II has been designed for stereo observa-tions. Thanks to the computer control it is able to compen-sate the signal delays, to equalize the gain and to stabilizethe trigger rates such that observations at lowest thresholdsinside the night sky regime are possible.

The Sum-Trigger-II has been tested and mass productionof the individual elements and boards has started. Theinstallation, on the MAGIC telescopes, is foreseen theend of this Summer (2013). This system will allow us tolower the trigger threshold of MAGIC and observe manyinteresting objects such as Pulsars, high redshift AGN andGRBs.

Acknowledgement: Technical support: M. Bettini, D. Corti,A. Dettlaff, D. Fink, M. Fras, P. Grundner, C. Knust, R. Maier, S.Metz, M. Nicoletto, O. Reimann, Ma. Reitmeier, Mi. Reitmeier,K. Schlammer, T. S. Tran, H. Wenninger, P. Zatti.

References[1] T. M. Kneiske, K. Mannheim, and D. H. Hartmann.

Astronomy Astrophysics, 386(1): 1-11, April 2002. doi:10.1051/0004-6361:20020211

[2] MAGIC collaboration. Science, 322(5905): 1221-1224,November 2008. doi: 10.1126/science.1164718.

[3] Francesco Dazzi. PhD Thesis, University of Udine, March2012.

[4] Dennis Hafner. Diploma Thesis, Munchen, November 2010.