TRIUMF

G

ANNUAL REPORT

SCIENTIFIC ACTIVITIES

2001

ISSN 1492-417X

CANADA’S NATIONAL LABORATORYFOR PARTICLE AND NUCLEAR PHYSICS

OPERATED AS A JOINT VENTURE

MEMBERS: ASSOCIATE MEMBERS:

THE UNIVERSITY OF ALBERTA THE UNIVERSITY OF MANITOBATHE UNIVERSITY OF BRITISH COLUMBIA McMASTER UNIVERSITY

CARLETON UNIVERSITY L’UNIVERSITE DE MONTREALSIMON FRASER UNIVERSITY QUEEN’S UNIVERSITYTHE UNIVERSITY OF VICTORIA THE UNIVERSITY OF REGINA

THE UNIVERSITY OF TORONTO

UNDER A CONTRIBUTION FROM THENATIONAL RESEARCH COUNCIL OF CANADA OCTOBER 2002

The contributions on individual experiments in this report are out-

lines intended to demonstrate the extent of scientific activity atTRIUMF during the past year. The outlines are not publications

and often contain preliminary results not intended, or not yetready, for publication. Material from these reports should not be

reproduced or quoted without permission from the authors.

iv

CERN COLLABORATION

INTRODUCTION

This is the seventh year of the collaboration be-tween TRIUMF and CERN which is producing acceler-ator components for the Large Hadron Collider (LHC)project. Three technical projects and two beam dy-namics studies are presently being carried out. Twenty-four tasks have been successfully completed. The to-tal funding for this work over the two Five-Year Plans(2000–2010) is $40.5 million, of which $11 million is forTRIUMF salaries and the remainder is for equipmentand outside contract costs.

The largest contribution, in terms of dollars, is forthe production of 52 special magnets for the two beamcleaning insertions in the LHC. These twin aperturequadrupoles are warm or copper coil magnets to per-mit operation with the anticipated beam losses in thisregion where the beam is carefully collimated to avoidbeam losses in the superconducting magnets in the restof the ring. In 1999, ALSTOM Canada (Tracy, Quebec)was awarded the contract for the production of the first17 of the twin aperture quadrupole magnets. ALSTOMhad previously built a prototype magnet which did notmeet the magnetic field requirements, so changes tothe lamination design and assembly tooling were re-quired. The first series magnet was delivered to CERNin March of this year, after satisfying the required as-sembly tolerances at the factory. Magnetic field mea-surements on this magnet carried out at CERN provedthe magnet to be acceptable. The contract was then ex-tended to 52 magnets and further production started.By the end of the year, ALSTOM was producing mag-nets at the planned rate of two per month.

TRIUMF is also designing and building 6 resonantcharging power supplies (RCPS), 9 pulse forming net-works (PFN), and 20 thyratron switch tanks for theLHC injection kickers. This work is being carried out atTRIUMF in a specially built clean room with overheadcrane coverage. The RCPS were completed in 2000 andone of them was tested at 66 kV operation this year.The 9 pulse forming network tanks were assembled andtested satisfactorily at low voltage. There was excellentagreement between the pulse measurements and thePSpice predictions used for the design. Orders havenow been placed for all of the components for the 20switch tanks, with assembly scheduled for 2002.

The remaining technical work is an instrumenta-tion task which involves the design and fabrication ofelectronics for the readout of the LHC beam positionmonitors (BPM). A prototype of this VME based dataacquisition board (DAB) was successfully tested in theSPS beam during the summer of 2000. However, a de-cision at CERN to remove these electronics modules

from the high radiation environment of the LHC tun-nel required some redesign. Some additional featureswere also added. Laboratory tests were carried out atCERN in November on the revised design and satis-factory operation was demonstrated. Ten modules arenow being produced for beam tests in 2002. TRIUMFhas also arranged for the production of 2300 filter pairsfor the analogue portion of the BPM readout.

The main beam dynamics involvement has been thedesign and optimization of the beam optics for the col-limation insertions. The work this year was aimed atimproving the robustness of the collimation system andmaking modifications to the design for cost savings.TRIUMF is also assisting with the study of beam-beaminteractions at the LHC collision points using a simu-lation code developed at TRIUMF.

The LHC cost review carried out at CERN inthe latter part of the year indicated that the cost-to-completion of the accelerator and experimental ar-eas had increased by about 18% over the original esti-mate. The impact of this cost overrun on the comple-tion schedule for the LHC is still not determined. Inany case the components being provided by TRIUMFshould be delivered as scheduled.

BEAM DYNAMICS

The simulation study of beam instability inducedby interactions between the two beams at the LHCcollision points has been extended to include the effectof “parasitic collisions” between bunches approachingand leaving the crossover; a promising start has alsobeen made to include longitudinal motion within thebunches in the simulation. For the LHC collimationsystem, the priority was updating the cleaning inser-tion optics to take account of economy-driven changesin the magnet lattice. Studies of the robustness of theLHC collimation system against alignment and mag-netic field errors were thereby delayed, but have nowrestarted with fresh impetus from the new CollimationTask Force at CERN. The study of resonance-free op-tics in the LHC was completed by investigations con-firming its insensitivity to octupole and skew sextupoleerrors.

Coherent Beam-Beam Effects in the LHC

This year we continued to study coherent modes ofbeam-beam interactions using a simulation code basedon a hybrid fast-multipole (HFM) field solver, origi-nally developed at TRIUMF for space-charge calcula-tions. Of particular concern are the so-called “parasiticcollisions”, where counter-rotating bunches of protonspass each other in close proximity as they approachthe actual collision point. Although these long-range

206

encounters involve weaker electric fields than the head-on collisions, there are many more of them and hencethere is a greater potential to excite unstable modes.

We found that the HFM method, which easily ac-commodates varying distance scales, provides efficientand accurate field solutions for parasitic collisions, al-lowing a fully self-consistent multi-particle simulation.With traditional mesh-based field solvers, such simu-lations become prohibitively expensive due to the ex-ponential scaling of computation time with the size ofthe solution region.

A number of different scenarios involving head-oncollisions with different betatron tunes, beam intensi-ties, and beam sizes, as well as early tests and resultsfor long-range collisions, have now been studied [W.Herr et al., Phys. Rev. ST Accel. Beams 4, 054402(2001)], with further long-range results to be reportedin 2002. For both types of collision, the simulations re-vealed the expected coherent modes with frequenciesin good agreement with analytical predictions.

For head-on collisions with equivalent beams, thepotentially unstable π mode is isolated from the inco-herent continuum of frequencies. On the other hand,sufficient differences of intensity, betatron tune, orbeam size between the two beams will shift the π modeinto the continuum, where it will be Landau-damped.Our simulation tool allows us to quantify these ef-fects and hence predict what operational parametersare needed to avoid coherent beam-beam instabilities.

Later in the year, we began a new study of anotherissue relatively unexplored by simulation: the effect oflongitudinal motion in coherent beam-beam interac-tions. Extending the model fully to three dimensionsescalates the cpu requirements to the supercomputerrealm. However, we have developed a longitudinal slic-ing algorithm that makes the problem tractable on asmall computing cluster. The simulation code has beenextended to support this model and has been tested ona single processor. In 2002 we plan to parallelize thecode with MPI for use on a small Linux cluster beinginstalled as a “seed” facility at TRIUMF.LHC Collimation

Lattice work included rematching the beam opticsfor the collimation insertions in LHC Version 6.3. Asa result of an engineering proposal for standardiza-tion of the connection cryostat (to save 0.5 MSF), thequadrupole Q7 (left and right, IR3 and IR7) had to bemoved by 3 m. This was done without a serious loss ofperformance.

Improving the robustness of the collimation systemwas the main subject during the second half of 2001.This was done as part of the work of the CERN Col-limation Task Force (see http://www.cern.ch/LHC-collimation), whose final statement is to be issued in

November, 2002. Ongoing studies include: modificationof the collimator set-up (probably longer collimators)to ensure protection against accidental impact on themby a large fraction of the LHC beam; modelling steadymulti-turn losses with DIMAD; and a search for themost dangerous optical error that would affect momen-tum collimation.

Resonance-Free LHC Optics

This study was completed in 2001. Resonance-freeLHC optics make it possible to cancel the effect of non-linear resonances. However, the resonance-free condi-tion itself can be destroyed by large gradient errors.For nominal and resonance-free lattices we computed:

• Q1′′ and Q2′′, the second derivatives of tuneswith respect to momentum, associated with theskew sextupole field error a3;

• resonance driving terms and the reduction in longterm dynamic aperture due to octupolar errors:b4 and a4.

It was demonstrated that resonance-free opticsmake it possible to tolerate the nominal a3 error and3–10 times the nominal b4, a4 errors, without takingadvantage of their correction systems. Gradient errorsthree times larger than the nominal ones do not destroythese benefits (see Fig. 193).

CONTROLS AND INSTRUMENTATION

LHC Orbit System Components

The electronics intended to read out the LHCbeam position monitors (BPM) comprise a calibra-tor, 70 MHz low-pass filter, wide-band time normal-izer (WBTN), and 40 MHz digitization. As a resultof radiation tests performed at CERN in 2000, it wasdecided to move much of the LHC BPM electronics out

Fig. 193. Second order chromaticity (400 seeds) causedby skew-sextupole field error in main LHC bends. Theresonance-free optics (right scale) have smaller parasiticchromaticity than the nominal, even for three times thenominal gradient errors.

207

of the radiation fields in the tunnel. This required re-design of the CERN normalizer and the TRIUMF dig-ital acquisition board (DAB) which uses two complexprogrammable logic devices. The WBTN was repack-aged as two modules: an analogue part remaining inthe LHC tunnel, and a digital part moved to a mezza-nine card on the DAB module. The WBTN analoguesignal is transmitted through a 3 km fibre optic cableto the DAB II module which accepts two mezzaninecards, one for each plane of the BPM. After reviewof these changes, the specification document was re-vised in May and the following operation modes weredefined:

• Orbit: continuous, real time closed orbit mea-surement.

• Capture: turn-by-turn measurement for se-lected/all bunches/batches and data averaging.

• Histogramming: record the distribution of datawith one global orbit acquisition, and histogramfor data validation.

• Post-Mortem: provide a history of the beam or-bit and turn-by-turn position in order to diag-nose and understand beam loss or sudden beamdumps.

• Calibration: asynchronous acquisition (not re-liant on the 40 MHz bunch clock and the 11 kHzrevolution frequency).

All three acquisition modes (orbit, capture and his-togramming) can proceed in parallel. The post-mortemand histogramming additions resulted in considerableincrease in design complexity, as did the decision to ac-quire horizontal and vertical signals without the beamsynchronous timing. Unlike the first DAB prototype,the DAB II is not responsible for providing timing andcontrol for WBTN calibration. Instead, so as not torely on beam synchronous timing, calibration is donein the LHC tunnel using an oscillator on the WBTNfront-end.

Three DAB II cards were assembled, including a“dummy” WBTN mezzanine card to simulate BPMdata. After successful local tests of the redesignedhardware and application software, the cards were de-livered in November to CERN where laboratory testson the DAB/WBTN combination demonstrated theircorrect functioning. During this time the prototypeLHC BPM system became operational in the SPS, andthe bunch-by-bunch measurement capability was usedextensively to understand the electron cloud instabilitywhich may limit LHC-beam intensity. Figure 194 showsthe horizontal oscillation amplitude of each bunch inthe LHC batch for 1000 turns. Marginal stability couldbe reasserted by changing the machine chromaticity.

Fig. 194. Bunch-by-bunch oscillation amplitude (mm).

Although further hardware modifications will be re-quired to correct minor problems with VME interrupthandling, and possibly to add further functionality, it isanticipated that ten DAB II modules will be deliveredto CERN in June, 2002 for beam tests in the SPS and,depending on the outcome, 60 pre-production modulesmight be ordered in 2003.Filter pairs

The matching criterion for the filter pairs now callsfor the 3 dB corner frequencies to be within 1 MHz.The filter manufacturer, Lorch Microwave, promisedthat the time responses will match within 100 ps. Thepre-production run of 150 pairs of 70 MHz low pass fil-ters was received from Lorch in April. Ten pairs wereretained at TRIUMF and 140 pairs were sent to CERNwhere testing of 26 pairs confirmed electrical measure-ments made by Lorch during the manufacture.

Testing of the radiation hardness of the filters wasbegun concurrently at CERN and TRIUMF in Au-gust. Three filters tested at TRIUMF, using a beam of67.3 MeV protons with a dose rate of 7.3 Gy/min, wereexposed to 200 Gy, which is the anticipated lifetimeexposure at the LHC. Sensitive tests with a networkanalyzer found no change in electrical properties, andno corrosion was found when the irradiated filters wereopened for inspection. The lengthy test at CERN, withan estimated dose rate of 20 Gy/week, accumulated anexposure of 900 Gy including the possibly more dam-aging effects of neutrons. CERN measurements con-firmed no changes in the electrical properties and nocorrosion. A mis-steered proton beam in the LHC couldcause up to 2 W dissipation in the 50 Ω resistor inside afilter. At TRIUMF, a filter heated by 2 W rose 47C intemperature and the time delay through it increasedby 15 ps, which is considered acceptable. The othercharacteristics remained constant.

208

The production run of 2150 pairs arrived at TRI-UMF in October, much ahead of the original deliveryschedule. Lorch Microwave measured 13 electrical pa-rameters of each filter and made this database availableas an Excel spreadsheet. A LabVIEW program, inter-faced to a network analyzer, will be used to test a 5%sample of the filter pairs, and shipment to CERN isanticipated in February, 2002.

MAGNET DEVELOPMENT

TRIUMF has agreed to provide CERN with fifty-two twin aperture resistive quadrupole magnets for theLHC cleaning insertions. In 1999, TRIUMF awardedALSTOM Canada Inc. (Tracy, Quebec), a contract tomanufacture the first 17 quadrupoles. During 2000,ALSTOM prepared tooling and started working onthe first production magnet. Some technical difficultiesin welding the stacks of laminations to the requiredstraightness tolerance, and a period of labour prob-lems, delayed completion of this first magnet. Mechan-ical measurements of the assembly accuracy are carriedout with a special pole distance measurement device(PDMD) at the factory. This device measures the dis-tance between flat-parts of the poles at four gaps foreach aperture. The acceptance criterion requires a tol-erance of ±100 µm for 90% of the length of the magnet.Figure 195 shows the results for one of the aperturesof the first magnet.

The first magnet satisfied these requirements andwas air-freighted to France, arriving at CERN onMarch 5. CERN made preliminary mechanical andmagnetic field measurements of the magnet, and onApril 3, accepted the magnet. Measurements on themagnet were reported in a paper presented at MT17[Clark, Proc. 17th Int. Conf. on Magnet Technology(MT17), Geneva, September 24–28, 2001 (IEEE Trans.Appl. Superconductivity, in press)].

After the first magnet was delivered, ALSTOM andTRIUMF negotiated the extension of the contract from

Fig. 195. Measured pole distance errors on MQW001 as afunction of the longitudinal position in one of the apertures.

Fig. 196. Layout of the so-called TRIUMF shop at AL-STOM.

17 to 52 magnets. ALSTOM reorganized the project,bringing in a full-time project manager/engineer. Theproduction shop was also reorganized and a second coilmold and stacking table fabricated to enable the mag-nets to be produced at a rate of two per month. DEL-STAR Inc. (Montreal, Quebec), replaced SIGMA asthe coil winder, with the coil insulation and impreg-nation continuing to be carried out at ALSTOM. Fig-ure 196 shows the layout of the so-called TRIUMF shopat ALSTOM.

The year-end saw 3 production magnets at CERNand 3 magnets in transit to CERN. The productionrate peaked at 3 magnets per month just before Christ-mas. ALSTOM expects to deliver at least 24 magnetsin 2002.

The method of reviewing the quality of the twotypes of laminations punched for these magnets wasalso presented at MT17 [Clark, op. cit.].

KICKER MAGNETS

In collaboration with CERN, TRIUMF has de-signed and is in the process of assembling and testing 4CERN LHC injection kicker systems plus spares. EachLHC injection kicker system will consist of 1 resonantcharging power supply (RCPS), two 5 Ω pulse form-ing networks (PFN), 4 thyratron switch tanks, and 2kicker magnets. The LHC injection kicker magnet sys-tems must produce a kick of 1.3 T m each with a flat-top duration which is variable up to 7.86 µs, a risetime of 900 ns, and a fall time of less than 3 ms. Thecombination of ripple and stability in the field from allkicker system components must be less than ±0.5%. Aprototype kicker magnet has been assembled at CERNand is ready for testing. Six RCPS, including the proto-type, have been assembled at TRIUMF. The prototypehas been extensively tested at CERN and one produc-tion model has undergone preliminary testing at 66 kV

209

Fig. 197. Four completed RCPS and 8 PFN in storage onthe proton hall roof, prior to tests.

operation. Fully documented RCPS tests will be per-formed early in 2002. Nine PFN have been designedand assembled at TRIUMF and tested at low voltage.Figure 197 shows the completed PFN and RCPS instorage on the proton hall roof beams. Testing of pro-totype thyratron switch tanks has been completed atCERN, and design of the production models has beencompleted at TRIUMF. Orders have been placed forall of the components for 20 switch tanks that will op-erate at 66 kV.

RCPS

An RCPS has two parallel outputs, to charge two5 Ω PFN to 66 kV. The RCPS has a 2.6 mF storagecapacitor bank charged to 3 kV. A thyristor is usedto switch the energy on the capacitor bank onto theprimary of a 1:23 step-up transformer of low leakageinductance. The output of the secondary is transferredto two 5 Ω PFN through two coaxial cables, two diodestacks and two 70 Ω resistors. The RCPS is designedso that the PFN can be charged up to 66 kV in lessthan 1 ms at a repetition rate of 0.2 Hz.

The TRIUMF Kicker group has completed the se-ries production of 5 RCPS, which incorporate sev-eral improvements over the prototype RCPS. The twodummy loads that were used for prototype PFN test-ing at CERN were returned to TRIUMF. New 1.1 µFdummy load capacitors rated at 66 kV, which werereceived from ZEZ Silko (Czech Republic), have beeninstalled and tested. Each 1.1 µF dummy load repre-sents a 28 cell PFN. A standard test procedure hasbeen developed for RCPS to be tested at TRIUMF.The tests will proceed early in 2002.

PFN

Each 5 Ω PFN is composed of two parallel 10 Ωlumped element delay lines (Fig. 198) consisting of4.3 m long precision wound coils, high voltage and high

Fig. 198. One of 9 PFN under construction at TRIUMF.

current capacitors, and damping resistors. There is athyratron switch tank at each end of the 5 Ω PFN.

Nine PFN have been built and tested at low volt-age at TRIUMF to ensure that performance is withinspecifications. High voltage tests will be carried outfollowing the manufacture of high voltage switch tanksat TRIUMF. The PFN were charged to 10.0 V and dis-charged into a precision 5 Ω load using a fast switch-ing MOSFET with very low (11 mΩ) on-state resis-tance. The low voltage charging and switch circuitswere modelled with PSpice in order to understand andtherefore compensate for the measurement system cal-ibration errors. Measurements in which the PFN wasreplaced with a 5 mF capacitor bank with a low par-asitic inductance and series resistance established thecalibration. All of the PFN, with numbering from 2through 10, were measured without damping resistorsinstalled; PFN #2 was measured with and withoutdamping resistors. The ripple without damping resis-tors was less than ±0.5% on 8 of the 9 PFN, and ±1%on PFN #4. Voltage ripple for PFN #2, with dampingresistors, was better than ±0.3%. The predicted kickrise time for PFN #2, from 0.2–99.8%, was 900 ns fora pulse with a flat-top of 8.54 µs duration and a rippleof less than ±0.2%, which meets the required specifi-cations. There was no attempt to set up a low voltagedump switch to measure the fall time at low voltage.The absolute average measured flat-tops (Fig. 199) ofthe 9 PFN were within ±0.1% of each other, indica-tive of quality control during manufacture of the PFN.A procedure was developed for accurately measuringpulse performance of the PFN. The results of pulsemeasurements are compared with PSpice predictionsin Fig. 200. There is a 0.25% discrepancy between theabsolute value of the flat-top as measured and calcu-lated by PSpice. However, the detailed shapes of the

210

97

97.5

98

98.5

99

99.5

100

100.5

101

2 3 4 5 6 7 8 9 10

PFN NUMBER

AV

ER

AG

EV

OLT

AG

E%

ΩΩΩΩ shields installed

ΩΩΩΩ shields not installed

Fig. 199. Measured average flat-top voltage normalized to200 V charging voltage (from 2 µs to 9 µs) with no dampingresistors installed (error bars represent the maximum peakto peak ripple).

99.5

99.6

99.7

99.8

99.9

100.0

100.1

100.2

100.3

100.4

100.5

0 1 2 3 4 5 6 7 8 9 10Time ( µµµµs)

Nor

mal

ized

puls

ehe

ight

%

Measurements with andwithout damping resistors

Pspice calculations with andwithout damping resistors

Fig. 200. PFN #2 measured and calculated pulse wave-forms with and without damping resistors installed.

measured and calculated pulses are in excellent agree-ment.

Thyratron Switch Tanks

The thyratron switch tanks will mount on the endsof the PFN tanks. The main switch (MS) thyratronwill be connected to a 5 Ω transmission line kickermagnet via 10 parallel 50 Ω coaxial cables, and thekicker magnet output is connected to a 5 Ω resistiveterminator. The dump switch (DS) thyratron will alsobe connected to a 5 Ω resistive terminator.

The TRIUMF Kicker group, in collabora-tion with CERN, has designed a prototype HVgrid/heater/reservoir bias circuit for the thyratrons.Two prototype bias circuits were built and benchtested at TRIUMF, and sent to CERN in Septem-ber, 2000. CERN installed a prototype bias circuitin a prototype thyratron switch tank and completedpower tests. CERN requested several modificationsand a new feature. Two new bias boxes were built andbench tested at TRIUMF, and shipped to CERN inNovember. However, CERN has requested, followingsuccessful completion of power tests early in 2002, thatTRIUMF supply 34 bias boxes; 20 of these will be usedin the LHC injection kickers and 14 are to be used inthe MKE (SPS extraction) CERN kicker system.

Redesign of all 20 thyratron switch tanks was com-pleted. 184 drawings have been reviewed, all compo-nents have been ordered, and contracts awarded to thefollowing machine shops: Sicom Industries (BC), Tal-van (BC), Sunrise Engineering (BC), PDE, Acrodyneand Carleton University. Components started to arriveat the end of 2001, and most should be received in Jan-uary/February, 2002. We expect to complete the designand place orders for the dump switch terminating re-sistor assemblies, thyratron transport jigs and liftingframes for the switch tanks early in 2002.

Specifications

Two specifications were issued.

211