STATUS AND FUTURE STRATEGY FOR ADVANCED HIGH POWER MICROWAVE SOURCES FOR ACCELERATORS

Transcription

1 STATUS AND FUTURE STRATEGY FOR ADVANCED HIGH POWER MICROWAVE SOURCES FOR ACCELERATORS Abstract Enrgy fficint convrsion of lctrical grid powr into radio frquncy powr is bcoming on of th ky aspcts of futur acclrators. A significant part of th initial invstmnt and running costs of ths machins will b dtrmind by th cost and fficincy of thir RF systms. For largscal acclrators, which ar proposd for instanc by th Intrnational Linar Collidr Study or by th Futur Circular Collidr Study, RF fficincy will likly b a dtrmining factor in th approval procss. Efforts ar alrady in plac to strtch th fficincy of xisting RF sourcs to highr lvls. Wll-known tchnologis such as klystrons ar bing rinvntd with modrn-day bam physics, inductiv output tubs ar bing combind for highr powr output, and modular solid stat amplifirs ar bcoming mor popular as RF sourcs for acclrators. This talk will giv an ovrviw of rcnt advancs and trnds in RF sourc dvlopmnts. It will discuss futur nds and th stratgy towards highr fficincy dvics for th bnfit of th acclrator community. THE NEED FOR HIGHER EFFICIENCY RF SOURCES Th lctrical powr consumption of futur acclrators will b drivn to a larg part by thir RF systms. This is particularly tru for lctron collidrs, circular (.g. FCC) or linar (.g. ILC or CLIC), which nd RF systms in th ordr of 100 MW or abov (s Tabl 1, [1 6])). Othr xampls with significant RF powr consumption ar high powr hadron linacs (.g. ESS) and cyclotrons (PSI) for nutron scinc or Acclrator Drivn Systms. In th cas of hadron collidrs, magnt systms dominat lctrical powr consumption whil RF systms only us a fw prcnt. At CERN, whr most acclrators ar circular, only 6% of th annual consumption ( 1.1 TWh in 2017) ar causd by th RF systms. For a 3 TV CLIC machin, around 50% of th facility powr consumption is du to th RF systm. With an stimatd yarly total of 2.74 TWh [4], and a Europan avrag pric for non-houshold consumption of 0.1 /kwh in 2017, this translats into 187 M of lctricity costs/yar for th RF systm alon. Thrfor vry prcntag point in RF fficincy has a significant impact on th running cost. For Acclrator Drivn Systms (ADS), th powr to run th acclrator complx (including cooling, vntilation, cryognics, offics, tc.) must b small (5-10%) compard to th powr, which is producd in th ADS cor. To achiv ths valus th acclrator complx must hav a total fficincy btwn 0.2 < η acc < 0.4 [7], which prsnts a challng for today s acclrators. If not achivd, th ida frank.grigk@crn.ch 12 F. Grigk, CERN, Gnva, Switzrland Tabl 1: RF Powr for Various Futur Acclrators Projct P total [MW] P RF [MW] P bam [MW] PSI LEP FCC- tbd ESS CLIC ILC CLIC total powr is for th whol facility with xprimnts, without injctor of using ADS for nrgy production is unlikly to bcom conomically viabl [8]. TODAY S TECHNOLOGIES & NEW DEVELOPMENTS In th following w will rviw today s stat of th art of high-powr RF sourcs (s Tabl 2) and point to dvlopmnts, which may chang thir fficincis or powr rach in th nar futur. As pulsd and Continuous Wav (CW) opration pos diffrnt challngs to th RF sourcs both mods of opration will b tratd. Modulator Efficincy Most high-powr RF sourcs nd a High-voltag (HV) modulator, which transforms th voltag of th lctrical grid to th HV and puls pattrn ndd by th RF sourc. For CW opration in gnral but also for pulsd opration of griddd tubs (ttrods, IOTs), th modulator is basically a HV powr supply. Pulsd powr for griddd tubs is obtaind by simply pulsing th grid using th input RF signal, whras HV pulss for th opration of non-griddd tubs (klystrons, magntrons) hav to b formd by th modulator itslf. Various topologis hav bn dvlopd in rcnt yars, which ar adaptd to diffrnt puls lngths and voltag nds [9]: i) Traditional puls transformr basd modulators with fficincis btwn 85% and 90% and ris tims in th rang of a fw hundrd microsconds. ii) High-frquncy transformr basd modulators such as th rsonant polyphas dsign mployd at SNS [10], or th stackd multi-lvl (SML) dsign dvlopd and usd by th Europan Spallation Sourc (ESS) with fficincis up to 92% and with a vry short ris tim in th rang of 100 µs. iii) Transformrlss modulators such as th Marx gnrator [11], which hav th potntial for vn highr fficincy, or dirct switch dsigns.

2 Tabl 2: Paramtr Rach of Diffrnt RF Amplifirs Ttrods IOTs/MB-IOTs* Klystrons Solid stat Magntrons Gain [db] n.a. 25 P pulsd [MW] /1.3* P CW [MW] /0.15 * Input voltag [kv] < (60*) n.a. 20 T puls [ms] any any < 4 any any T ris/fall [ns] ns ns η DCtoRF [%] (>70*) * Frquncy [MHz] < Activ dvlopmnt no MB-IOT ys ys littl undr dvlopmnt, at working point In summary today s modulators ar alrady oprating with vry high fficincy (85-92%), almost indpndnt of thir output powr (kw - MW), voltag (1-100 kv), and puls lngth. For short pulss (< 500 µs), th modulator ris tim bcoms an important factor in th systm fficincy and this is whr furthr dvlopmnts such as th SML dsign can still mak a significant diffrnc. Klystrons Th principl of klystrons was publishd in 1935 by O. Hil in Grmany and thn dvlopd in 1937 by Russl and Varian at Stanford Univrsity. Elctrons ar xtractd from a hatd cathod via a high DC voltag and ar thn vlocity-modulatd with th input RF signal. Aftr a crtain distanc th vlocity modulation is transformd into a dnsity modulation. Th bunchd lctron bam thn xcits th fundamntal rsonanc of th output cavity from which th RF powr is xtractd. Typical prformanc valus ar listd in Tabl 2. Pros and Cons Advantags of klystrons ar: i) simpl, mostly solid stat input amplifirs du to larg gain, ii) high output powr powr in th MW rang, iii) long liftim up to 40 kh. Thir disadvantags ar: i) HV nds with typically > 100 kv, which translats into oil tanks for brak down protction and xpnsiv HV modulators, ii) gain curv saturats at full output powr, which mans that most klystrons ar opratd blow full output powr and blow maximum fficincy to maintain a powr margin for th LLRF systm. Dvlopmnts In ordr to rduc th fficincy loss by opration blow saturation, FNAL, DESY and othrs ar dvloping klystron linarization algorithms, aiming for opration closr to saturation. With th advanc of modrn bam dynamics tools [12], nw mthods to incras th nrgy convrsion fficincy hav bn dvlopd, such as th cor-oscillation mthod (COM), [13], th Bunching, Alignmnt and Collcting (BAC) mthod [14], and th CERN-dvlopd Cor Stabilisation Mthod (CSM) [15]. Ths mthods ar bing valuatd by th High-Efficincy Intrnational Klystron Activity (HEIKA), which was initiatd at CERN in 2014 [16]. A proof-of-principl xprimnt with th BAC mthod has alrady bn mad by rtrofitting an xisting multibam S- band tub (VDBT, Moscow, 40 bams). This boostd th output powr by almost 50% and incrasd its fficincy from 42% to 66% [17]. Griddd tubs: Ttrods, Diacrods, IOTs Th principl of griddd tubs such as triods, ttrods, diacrods, and inductiv output tubs (IOTs) gos back to vacuum diods (without grid), which wr patntd in 1904 by J.A. Flming. Thy ar built sinc around 100 yars and ar usd in many of today s acclrators with frquncis up to 400 MHz and powr valus into th MW rang. Elctrons ar rlasd from a cathod and acclratd towards th anod. In triods a control grid modulats th flow of lctrons and thrby th anod currnt. With a rsistiv load in th anod circuit, th varying currnt rsults in a varying voltag. Ttrods us an additional scrn grid to dcoupl th control grid from th anod, which incrass th gain and rducs th tndncy to slf-oscillat, spcially with inductiv loads. Inductiv Output Tubs (IOTs) wr invntd in 1938 by A.V. Haff. Thy can b undrstood as a mixtur of a klystron and a triod and ar somtims calld klystrod. A control grid vry clos to th cathod rgulats th lctron currnt, which mans that IOTs us currnt modulation (as in triods) instad of vlocity modulation (as in klystrons) to crat a bunchd lctron bam. Th bunchd bam is acclratd with a DC voltag and thn passs an output cavity just as in a klystron. From thr th RF powr can b xtractd via inductivly coupld coaxial transmission lins. Pros and Cons Ttrods and diacrods provid output powr into th MW rang but ar limitd to frquncis of 400 MHz. Thir limitd gain of 15 db mans that highpowr ttrod amplifirs nd 2-3 stags of amplification, which drivs up th cost and rsults in complicatd amplifir systms. Evn though ttrods ar usd in many acclrators, 13

3 th ovrall markt is rlativly small, which mans: i) that many tubs ar no longr producd, and ii) thr ar vry fw companis, which ar capabl and willing to build ttrodbasd RF amplifirs. Nvrthlss, onc commissiond, ttrod and diacrod RF amplifirs ar xtrmly rliabl and can work for many dcads if maintaind adquatly. IOTs hav lss gain than klystrons but highr gain than ttrods (s Tabl 2), which puts thir input powr nds within rach of commrcially availabl solid stat amplifirs. As thy do not nd a long drift spac lik klystrons, IOTs ar quit compact and thrfor cost fficint. IOTs and indd all griddd tubs ar limitd in thir frquncy rach by th distanc of th control grid from th cathod. Th RF priod has to b smallr than th tim of flight from cathod to this grid. Thrfor th frquncy of IOTs is limitd to around 1.3 GHz. Th maximum powr of singl bam IOTs is limitd to around 100 kw. As thy ar usd for broadcasting of digital signals, thr is a commrcially viabl markt, which mans that long-trm availability sms rlativly scur. Dvlopmnts Svral yars ago Thals dvlopd th diacrod, an improvd vrsion of th ttrod, which can ithr doubl th powr at a givn frquncy, or doubl th frquncy at a givn output powr [18,19]. It is succssfully usd sinc 2015 at th LANSCE acclrator in Los Alamos but has not gaind widsprad us. In ordr to mak us of th advantags of IOTs for MW-class RF amplifirs, Thals, CPI, and L3 dvlopd two prototyp 704 MHz multi-bam IOT s (MB-IOT) for ESS with an output powr > 1 MW. Both tubs wr tstd at CERN and rsults ar rportd at this confrnc [20]. Magntrons An arly form of th magntron was invntd in 1910 by H. Grdin, followd in 1920 by A. Hull s invntion of th split-anod magntron, and thn prfctd in 1936 by Alksrff and Malaroff in Russia, who built a 3 GHz dvic dlivring 300 W of powr. Magntrons ar so-calld fr running oscillators. Elctrons ar rlasd by hating th cathod (s Fig. 1) and ar thn acclratd by a voltag diffrnc towards th anod. A magntic fild prpndicular to th anod/cathod plan maks th lctrons rotat around th cathod thrby xciting a pi-mod in th anod cavitis. Th pi-mod modulats th lctron currnt into bunchs, which thn furthr xcit th cavity filds. A currnt starts to flow as soon as th lctrons start hitting th anod and RF powr can b xtractd from on of th cavitis. Pros and Cons Advantags of magntrons ar: i) high DC/RF fficincy of up to 85%, and ii) low pric. Thir disadvantags ar: i) difficult phas/amplitud, which nds furthr R&D to mak thm usabl for multi-cavity acclrators, and ii) opration blow th working point may dcras th fficincy considrably. 14 Anod Cathod Figur 1: Elctron movmnt in a magntron. Dvlopmnts Up to now magntrons ar only usd in acclrators that us a singl RF sourc, such as lctron machins for X-ray production. Th combination of svral magntrons with a prcis phas and amplitud control has bn dmonstratd with proof-of-principl xprimnts [21, 22] but has so far not bn mployd in multicavity acclrators. In th US a consortium of CCR, CPI, and FNAL is activly dvloping a magntron basd RF systm for acclrators using 1.3 GHz magntrons capabl of 100 kw pak powr and a 10% duty cycl [23]. Solid Stat Solid stat, transistor-basd RF amplifirs promis costfficint RF powr gnration and th advantags of modular systms, which may allow hot-swapping of faulty units during opration. Most of today s systms oprat blow 100 kw at frquncis blow 1.3 GHz with DC to RF fficincis blow 55%. Pros and Cons Singl amplification units ar limitd to around 1 kw. As thr is no significant commrcial markt for highr powr transistors it is unlikly that this valu will b incrasd significantly in th nar futur. This mans on nds to combin a larg numbr of singl units for highpowr RF amplifirs. Traditionally this powr combination is don in pairs of 2 and sinc ach combination carris a crtain loss it is rlativly infficint to go byond som 10s of kilowatts. In most acclrator scnarios, th RF amplifirs hav to withstand significant amounts of rflctd powr and if th amplifir itslf cannot withstand ths rflctions, circulators ar usd to dviat powr into watr-coold RF loads. For solid stat amplifirs, th addition of circulators for all singl units and/or for th combind output oftn maks th whol systm too xpnsiv. Dvlopmnts Th ky to making cost-fficint solid stat amplifirs for powr valus > 100 kw lis in th ffctiv combination of th singl units. A promising solution is th us of combinr cavitis, which can combin th output powr of all singl units in on stag. Howvr, matching hundrds of input antnnas and minimising th rflctd

4 powr du to manufacturing tolrancs of th lctronics or du to faild units is far from trivial. CONCLUSIONS ON PARAMETER REACH AND OVERALL RF SYSTEM EFFICIENCY Th powr and frquncy rach of th various RF sourcs is summarizd in Fig. 2. Th chart is limitd to frquncis < 1.3 GHz and th MW rang bcaus highr valus can basically only b covrd by klystrons and with vry short pulss. In th rang btwn 10s of kw to 100 kw solid stat has to compt with ttrods in th low-frquncy rang and with IOTs in th highr frquncy rang and in both cass th fficincy of solid stat amplifirs is lowr than th comptition. In som cass, howvr, th modularity of solid stat systms may outwigh this disadvantag and thr th combining cavitis ar most likly th way towards highr powr units. Th dvlopmnt of multi-bam IOTs, which may rival klystron-basd systms is ncouraging but it may still tak som nginring ffort to mak thm as rliabl and cost-fficint as thir countrparts. Th dvlopmnt of high-fficincy klystrons and klystron linarisation algorithms is highly promising and will mak klystrons not only mor fficint but smallr with incrasd avrag output powr, and it will lowr th HV nds thrby rducing th nds for HV protction and making th modulators chapr. P [kw] P [kw] Ttrods pak (<1 ms) MB IOT [<3ms] Klystrons (<2ms) ttrods IOT pak (<10 ms) x Magntrons (<10 ms) 2 MBIOT prototyps klystrons magntron IOTs solid stat ttrods Ttrods CW Solid stat CW Klystrons CW IOT CW Magntrons CW x frquncy [MHz] klystrons magntron IOTs solid stat frquncy [MHz] Figur 2: Powr/frquncy chart of RF sourcs for pulsd (uppr chart) and CW (lowr chart) opration up to 1.3 GHz CERN NEEDS & DEVELOPMENTS At CERN around 80 high-powr RF systms (> 100 kw) ar in opration, half of which basd on ttrods, followd by klystrons and som first solid stat units, which ar prsntly bing tstd for th 200 MHz SPS RF systm upgrad. An ovrviw of all systms is givn in Tabl 3. Ttrods & klystrons Whn Linac2 is rplacd by Linac4 in 2021 most high powr systms will b basd on klystrons, making ths th most important R&D targt for CERN. Within th HEIKA ntwork fforts ar mad to dvlop a 400 MHz high-fficincy klystron basd on th CSM mthod for th LHC [24] and th FCC study. Th goal is to kp th xisting LHC modulators and powr supplis, and to incras th availabl output powr by having an improvd fficincy. Dvlopmnts for 800 MHz hav also startd in th FCC fram. A similar dvlopmnt may also bcom intrsting for th 352 MHz klystrons of Linac4 in ordr to limit th cost of rplacing xisting units. In a qust for high pak powr CERN stablishd thr 12 GHz klystron-basd tst stands to condition and charactris high-gradint acclrating structurs. In th X-band tst facilitis 1 and 2 (Xbox-1 & 2) [25] singl 50 MW klystron/modulator units provid 1.5 µs pulss of 50 MW with a 50 Hz rptition rat. A puls comprssor [26] transforms ths into 250 ns long pulss with 140 MW pak powr. In Xbox-3 [27] a novl approach was dvlopd, combining two tims two 6 MW 5 µs klystrons, capabl of running at 400 Hz rptition rat. Aftr puls comprssion and powr combination ach pair provids 2 tst placs with 45 MW in 250 ns pulss with a 200 Hz rptition rat [28]. To maintain th availability of ttrods, CERN has mad agrmnts with industry to rciv limitd manufacturing licncs in cas crtain tub typs go out of production. IOTs CERN has built a tst stand stand for MW-class IOTs, which has bn usd to succssfully tst two MB- IOTs, which hav bn procurd by ESS and built by CPI/Thals and L3. Whthr MB-IOTs can compt with th nw high-fficincy klystrons will dpnd on th final cost of th sris units. Solid stat CERN is prsntly working vry closly with Thals on th ralisation of 200 MHz solid stat amplifir towrs basd on combinr cavitis. Each of th 32 towrs shall b capabl of > 100 kw, which will thn b combind to a total powr of MW for th SPS travlling wav acclrating cavitis [29, 30]. In Linac3 thr ar plans to rplac two pulsd 100 MHz ttrod amplifirs, dlivring kw ach, with solid stat units. ACKNOWLEDGEMENT I want to thank th following collagus from th CERN RF group for thir advic and input: Olivir Brunnr, Eric Montsinos, Carlo Rossi, Igor Syratchv, Hlga Timko, and also Andrs Sunsson and Mats Lindroos from ESS. 15

5 Tabl 3: Ovrviw of High-Powr (> 100 kw) RF Systms at CERN Typ P [MW] f [MHz] T puls [µs] Rp. rat [Hz] N systms N tubs/units Linac2 Ttrod Linac3 Ttrod / Linac4 Klystron REX Ttrod / RFQD Ttrod PS Ttrod < 200 ms PS Ttrod / SPS Ttrod s 43 khz/ SPS (LIU) SSA s 43 khz/ SPS IOT s 43 khz/ LHC Klystron CW CW XBOX1,2 Klystron XBOX3 Klystron REFERENCES [1] P. Lbrun, Enrgy Efficincy of high-nrgy particl acclrators and collidrs, talk prsntd at 4th Workshop Enrgy for Sustainabl Scinc, Magurl, Romania, [2] CERN lctricity consumption flyrs, 2017, wb.crn.ch/consumption-flyrs. [3] Acclrator Baslin Dsign, Volum 3.II, in Th Intrnational Linar Collidr, Tchnical Dsign Rport, [4] A Multi-TV Linar Collidr Basd on CLIC Tchnology, CLIC Concptual Dsign Rport, CERN , [5] E. Jnsn, Dvlopmnts Towards Highr Efficincy of RF Systms, talk prsntd at FCC wk 2016, Rom. [6] ESS, ACCSYS Enrgy Invntory [7] J-P. Rvol, Enrgy fficincy and proton drivrs for ADS, prsntd at Proton Drivr Efficincy Workshop, PSI, Switzrland, [8] M. Haj Tahar, F. Méot, and S. Pggs, Enrgy fficincy of high powr acclrators for ADS applications, in Proc. IPAC2016, Busan, Kora, 2016, papr TUPOY044. [9] C.A. Martins, Modulators for high powr RF sourcs, prsntd at Proton Drivr Efficincy Workshop, PSI, Switzrland, [10] W.A. Rass t al., Oprational rsults of th Spallation Nutron Sourc polyphas convrtr-modulator for th 140 kv klystron RF systm in Proc. PAC 2001, Chicago, USA, papr MPPH110. [11] M.A. Kmp t al., Final dsign of th SLAC P2 Marx klystron modulator, in Proc. IEEE PPC 2011, Chicago, USA, pp [12] J.C. Cai and I. Syratchv, KlyC: 1.5D larg signal simulation cod for klystrons, papr accptd for publication, in IEEE Trans. on Elctron Dvics, to b publishd. [13] A. Yu. Baikov, C. Marrlli, I. Syratchv, Toward high-powr klystrons with RF powr convrsion fficincy of th ordr of 90%, in IEEE Trans. on Elctron Dvics, vol.62, no.10, pp , [14] I.A. Guzilov, BAC mthod of incrasing th fficincy in Klystrons, in Proc. od IEEE Vacuum Elctron Sourcs Confrnc 2014, St. Ptrsburg, Russia. [15] G. Burt t al., Particl-in-cll simulation of a cor stabilization mthod klystron, in Proc. of IEEE Intrnational Vacuum Elctronics Confrnc, IVEC 2017, London, UK. [16] I. Syratchv, Introduction to HEIKA. Tntativ structur and objctivs, prsntd at CLIC Workshop 2015, CERN, Gnva, Switzrland. [17] C. Marlli Rcnt dvlopmnts towards vry high fficincy klystrons, sminar at ESS, Fb [18] Thals Elctron Dvics, TH 628, vry high-powr RF Diacrod, 200 MHz, 1 MW, CW or 4.5 MW pulsd. [19] Thals Elctron Dvics, TH 680, High-powr Diacrod, 50 kw, CW or 80 kw in long pulss. [20] M. Jnsn t al., Tsting of th ESS MB IOT prototyps in Proc. of IPAC 2018, Vancouvr, Canada, papr WEXGBF1, this confrnc. [21] A.C. Dxtr, Phas lockd magntrons for acclrators, in Proc. of LINAC 2014, Gnva, Switzrland. [22] B. Chas t al., Prcision vctor control of a suprconducting RF cavity drivn by an injction lockd magntron, J. of Instr., vol. 10, [23] T. Bui t al., Advancd, phas-lockd, 100 kw, 1.3 GHz magntron, in Proc. of IEEE IVEC 2016, Montry, USA. [24] O. Brunnr t. al., RF powr gnration in LHC, in Proc. of PAC 2003, Portland, Orgon, USA. [25] N. Catalan Lashras t al., Exprinc oprating an X-band high-powr tst stand at CERN, in Proc. of IPAC 2014, Drsdn, Grmany, papr WEPME016. [26] A.A. Bogdashov, A 12 GHz puls comprssor and componnts for CLIC tst stand, in Proc. of RuPAC 2010, Protvino, Russia, papr TUPSA005. [27] N. Catalan Lashras t al., Commissioning of XBOX3: a vry high capacity X-band RF tst stand, in Proc. of LINAC 2016, East Lansing, USA, papr TUPLR047.

6 [28] M. Volpi t al., High powr and high rptition rat X-band powr sourc using multipl klystrons, prsntd at IPAC 2018, Vancouvr, Canada, papr THPMK104, this confrnc. [29] E. Shaposhnikova, E. Ciapala, E. Montsinos, Upgrad of th 200 MHz RF systm in th CERN SPS, in Proc. of IPAC 2011, San Sbastian, Spain, papr MOPC058. [30] E. Montsinos, CERN LIU-SPS 200 MHz RF upgrad SSPA amplifirs, prsntd at CWRF 2016, Grnobl, Franc. 17