CW MEASUREMENTS OF CORNELL LLRF SYSTEM AT HOBICAT

Transcription

1 MOPO67 Poceedings of SRF, Chicago, IL USA CW MEASUREMENTS OF CORNELL LLRF SYSTEM AT HOBICAT A. Neumann, W. Andes, R. Goegen, J. Knobloch, O. Kugele, Helmholtz-Zentum-Belin, 89 Belin, Gemany S. Belomestnykh, J. Dobbins, R. Kaplan, M. Liepe, C. Stohman, Conell Univesity, Ithaca, NY 853, USA Abstact In Enegy Recovey Linacs, such as the Conell ERL o BERLinPo, the main linac cavities ae opeated CW at low beam-loading. The choice of the extenal is given by two competing factos: The achievable field stability and the Hz maximum povided RF powe. To detemine the optimum extenal, LLRF measuements with the Conell system 6 Hz 5 wee pefomed at HoBiCaT to study the field stability at 5 Hz given micophonics detuning of a TESLA cavity fo diffeent gain settings and extenal values. Stable opeation at 3 extenal up to 8 was demonstated at a field phase stability of. degees. INTRODUCTION The Conell Laboatoy fo Acceleato-based ScienceS and Education (CLASSE) plans fo an X-ay light souce based on Enegy-Recovey-Linac (ERL) []. The Helmholtz-Zentum Belin is going to build the ERL demonstato facility BERLinPo []. ERLs, suitable fo light souces, need to acceleate high beam cuents up to ma. Howeve, the high beam powe is ecoveed in the cavities by the deceleated beam so, that the CW opeated supeconducting (SC) cavities of the main linac expeience zeo net beam loading. In the case of no beam loading the needed RF powe to establish a given acceleating voltage in a cavity is given by ( ) ) P f = Vacc ( ) L ( Δf + f / ( ) with the nomalized shunt impedance in linac definition, L the loaded quality facto, Δf the cavity detuning offset and f / = f / L the cavity half-bandwidth (as given in Figue ). The optimal L fo minimized powe equiements is theefoe a function of the peak cavity detuning Δf peak occuing duing opeation: f () L,opt =. () Δf peak This loaded will educe the RF powe equiements to save capital cost. Figue displays the equied fowad RF powe fo a seven cell SC cavity opeated at MV/m Wok suppoted by Bundesministeium fü Bildung und Foschung and the Land Belin. Axel.Neumann@helmholtz-belin.de now at Bookhaven National Laboatoy, Upton, NY 973, USA P f (kw) 5 Hz e7 e8 5e8 ext Figue : Requied fowad powe fo a seven cell supeconducting cavity with E acc = MV/m vesus extenal quality facto ext fo diffeent detuning (without beamloading). The black lines denote the ext analyzed in this wok. vesus the loaded quality facto fo diffeent detuning levels. The thee black lines denote the L values analyzed in this wok. Choice of L Obviously the choice of the optimal L has to fulfill two competing equiements. On the one hand fo a given detuning L has to be chosen such to minimize the equied powe level to maintain the desied acceleating voltage. On the othe hand the equied field stability limits the maximum L as the cavity bandwidth deceases with highe L. Naow bandwidth opeated cavities ae even moe susceptible to micophonics detuning and pondeomotive instabilities. ERLs typically equie a field stability of bette than. degees RMS in phase and some fo the elative field amplitude [3]. Thus, befoe fixing on the design value fo the loaded quality facto a measuement pogam has to demonstate the following key questions and tasks: What is the micophonics detuning level at vaious cavity bandwidths? To what extend does a smalle cavity bandwidth filte out highe fequency detuning components? What ae the noise souces tiggeing the micophonics detuning, especially peak events? 6 9 Opeating expeience with SRF acceleatos

2 Poceedings of SRF, Chicago, IL USA MOPO67 How often and to what detuning level do micophonics peak detuning events happen? What is the ultimately achievable field stability in amplitude and phase as a function of L and loop gain? Does it fulfill the equiements of the ERL s beam dynamics? Find the optimum combination of mechanical detuning contol by e.g. piezo based fast tunes and LLRF contol settings. What ae the optimal gain and filte settings? Slow DAC/ inteface DSP Contolle xdacs FPGA Fast RAM 6xs: P fo,p ef,p tans, Beam signal Futhe, in geneato diven LLRF systems, the coupling of the Loentz-foce detuning to field amplitude fluctuations by stong micophonics may cause cavity field tips, so called pondeomotive instabilities. Also in ERLs the deceleated beam should cancel the beam loading of the acceleated beam pefectly, but due to time jitte of the beam, beam losses and small phase fluctuations of the beam in geneal, esidual beam loading may occu. This needs to be compensated fo by the LLRF system by supplying a small powe ovehead [5]. Fist tests at CEBAF and the Jeffeson Lab FEL (with beam) using the Conell system aleady demonstated an opeation up to L =. 8 with phase stability aound. degees [6] in the enegy-ecovey mode. Intense studies of micophonics detuning and its compensation by means of piezo-based tunes had been done at HoBiCaT [7] and successful compensation up to L of 8 educing the detuning by about an ode of magnitude was demonstated. In the following sections the measuements done at the HoBiCaT hoizontal cavity test facility at HZB [8] in collaboation with Conell will be pesented. SETUP AND MEASUREMENT PROGRAM Fo the measuements at HoBiCaT a nine-cell TESLA cavity equipped with a TTF-III couple, the Saclay I tune including an impoved piezo tune [7] was installed. The new vesion of the Conell LLRF system [, 9] was commisioned using a new clock geneation setup deiving all needed clock and efeence signals diectly fom a lownoise fixed fequency.3 GHz efeence souce. A pictue of the new digital boad of the LLRF system including the FPGA fo fast field contol and the Tige Shak DSP fo detuning, opeational and cavity tip contol is given in Figue. Figue 3 shows the geneal scheme of the LLRF system and the clock and efeence signal geneation. The cavity field and powe signals ae downconveted to an intemediate fequency of.5 MHz of which the field components ae detected via fou times ovesampling at 5 MHz. The cavity was diven via a W solid state amplifie o altenatively by a 7 kw CPI IOT. Unfotunately the cavity was limited to only E acc = MV/m due to stong field emission. Figue : Pictue of the Conell LLRF digital boad. 5 MHz.5 MHz.3 GHz LO=RF-.5 MHz Powe supplies Digital I/O Slow/Med boad DAC Ms/s DAC DAC ks/s I Fast Buffe FF FPGA Vitex VME connecto DSP ADTS Coldfie boad Intelock cad uench detection A D C B u f f e DAC DAC Analog Intelock signals Clock (.5, 5 MHz) LO Pf P Pt Spae Klyston HV ipple Beam cuent Linux Soft-IOC Figue 3: Scheme showing the functionality of the Conell LLRF system. Measuement Pogam The aim of the measuements was to detemine the optimal loaded and the achievable field stability in the pesence of micophonics detuning and coupled dynamic Loentz-foce detuning. Futhe the additional detuning compensation fo low fequency micophonics below the fist mechanical eigenmode of the cavity should be demonstated and finally the amping of the cavity field fom to MV/m at small cavity bandwidth of less than ten hetz in the pesence of Loentz-foce detuning of the ode of hunded hetz. Finally, the aim was to optimize the setup of detuning and LLRF field contol at highest L possible still achieving the ERL s field stability equiements in a obust way. DETUNING MEASUREMENTS Fist the mechanical chaacteistics of the CW diven cavity was tested by measuing the piezo-to-rf detuning tansfe function and the micophonics detuning spectum []. The tansfe function of the Saclay I tune-cavity 9 Opeating expeience with SRF acceleatos 63

3 MOPO67 Poceedings of SRF, Chicago, IL USA Detuning (Hz) E acc (MV/m) K P K P Phase (deg.) Figue : RF detuning tansfe function in dependance of piezo tune modulation fequency (uppe plot). The lowe plot shows the phase lag between piezo excitation signal and the cavity detuning esponse t (s) Figue 6: Cavity field amplitude (E acc ) vesus time fo diffeent popotional gains at L = Mechanical eigenmodes by helium pessue fluctuations and to compensate detuning duing the field amping. A f (Hz) Tubo pump Figue 5: Detuning spectum of a TESLA cavity opeated CW at L =5 7 and E acc = MV/m. The data wee taken by eadout of the khz onboad ing buffe. combination is shown in Figue displays the detuning amplitude esponse and the phase lag between detuning and the piezo modulation fequency signal. The typical goups of mechanical eigenmodes between 5 and 35 Hz can be obseved and also the fist mechanical eigenmodes with a athe low esponse amplitude and a high mechanical quality facto at Hz and 35 Hz. The phase esponse indicates a goup delay of -3μs. Due to the naow bandwidth in geneal well below 5 Hz, most high fequency components ae filteed by the cavity itself and only the fist two eigenmodes ae pesent in the detuning spectum given in Figue 5. Data taken at L of 5 7 showed an RMS detuning of Hz with 5 Hz peak detuning. A fist commissioning of the LLRF s piezo contol loop showed the limitation of the loop gain by the fist mechanical eigenmode. To suppess any excitation only low gains with lowpass filteing below Hz wee possible. In the following tests the detuning contol was theefoe mainly used to contol slow difts FIELD STABILITY MEASUREMENTS Afte successfully closing the LLRF loop at L =5 7 an excitation of the 8/9-π passband mode about 8kHz below the π-mode damped by 7dB was obseved. As pedicted by contol theoy calculations this lead to instabilities of the loop oscillating with 5.8 khz at a popotional gain of. Changing the main loops filte settings to suppess the next passband mode and futhe optimization of the loop gain esulted in a stability of σ Φ =. degee and elative amplitude eo of 6 5. At that time the efeence souce had a malfunction so that all measuements pesented in the following had to be done by a standad fequency synthesize. This had a facto of fou wose phase noise chaacteistics limiting the achievable stability. Figues 6 and 7 show the fist gain scans at field levels of MV/m and L =5 7 fo diffeent popotional gains and zeo integal gain. The esidual eo follows as expected the /(+ ) dependance and at low gains of - stong coupling between phase and amplitude eos hint at micophonics amplified by Loentz-foce detuning. Gain Scans vesus L Figues 8, 9 and summaize the gain scans pefomed at L of 5 7, 7 and 8. Shown is in log scale as a colo code the achieved RMS phase stability fo a given setting of integal and popotional gains. The best values achieved ae maked by ed o dak blue spots. White aeas denote esults with phase eos highe than. degee o cavity field tips. Cavity field tips wee caused by intinsically instable gain settings o pondeomotive instabilities due to too low feedback gain leading to highe esidual amplitude deviations. In Table the esults of 6 9 Opeating expeience with SRF acceleatos

4 Poceedings of SRF, Chicago, IL USA MOPO67 Phase (deg.) K P K P K P deg t (s) Figue 7: Cavity field phase vesus time fo diffeent popotional gains at L =5 7. Figue 9: RMS phase stability (log scale) of the cavity field fo diffeent integal ( ) and popotional gain settings at L = 8. The ed dot maks the achieved absolute minimum of this gain scan. 8.8 deg deg Figue 8: RMS phase stability (log scale) of the cavity field fo diffeent integal ( ) and popotional gain settings at L =5 7. The ed dot maks the achieved absolute minimum of this gain scan. Figue : RMS phase stability (log scale) of the cavity field fo diffeent integal ( ) and popotional gain settings at L = 8. The dak blue dot maks the achieved absolute minimum of this gain scan. the gain scans ae summaized. Fo MV/m in all thee cases the cavity could be opeated at about kw powe o below. Depending on the cavity bandwidth the RMS micophonics level vaied fom -9 Hz. Best field stability was achieved fo the lowest L as expected, but also at 8, a half-bandwidth of only 3.5 Hz(!), the cavity was opeated with a vey high stability of about. degee phase deviation. Duing the measuements the piezo Table : Cavity field stability esults at E acc = MV/m σ f σ Φ σ A /A P f L (Hz) (deg) (kw) contol loop mainly kept the cavity on esonance contol- ling the cavity in the sub-hetz egime. Also amping of the cavity field within second to MV/m was demonstated fo all thee quality factos. OUTLOOK In summay the measuements showed that a loaded quality facto of is feasible achieving highest field stabilities. Even at peak detuning of 5 Hz about 5 kw of installed RF powe would suffice to stably opeate the cavity at MV/m. Nevetheless, futue measuements with a bette pefoming cavity have to demonstate opeation at field gadients as high as MV/m and a eliable long tem opeation with an automated field ecovey afte a cavity tip. Futhe, it is planned to impove the pefomance of the piezo tune algoithm to effectively cancel micophonics detuning, maybe allowing opeation at even highe L. 9 Opeating expeience with SRF acceleatos 65

5 MOPO67 Poceedings of SRF, Chicago, IL USA ACKNOWLEDGEMENTS We would like to thank the pesonal at both laboatoies fo suppoting this wok, especially the IT depatments in suppoting the tansfe of the LLRF system in the HZB/BESSY EPICS envionment. Futhe thanks to Sascha Klauke, Michael Schuste, Ande Fahm,Dik Pflückhahn and Stefan Rottedam fo suppoting the opeation of HoBiCaT. We also would like to acknowledge Klaus Ludwig s design of the new clock and efeence system. REFERENCES [] J.A: Cittenden et al., Developments fo Conell s X-ay ERL, Poc. of the 3 d PAC (9), Vancouve, Canada, [] M. Abo-Bak et al., BERLinPo- An Acceleato Demonstation Facility fo ERL-based Light Souces, Poc. of the 5 th LINAC (), Tsukuba, Japan, [3] M. Liepe, S. Belomestnykh, RF Paamete and Field Stability Requiements fo the Conell ERL Pototype, Poc. of the th PAC (3), Potland, USA, [] M. Liepe et al., Expeience with the New Digital RF Contol System at the CESR Stoage Ring, Poc. of the st PAC (5), Knoxville, USA, [5] M. Liepe, J. Knobloch, Supeconducting RF fo enegyecovey-linacs, Nuclea Instumentes and Methods in Physics Reseach A, 557 (6), pp [6] M. Liepe et al., Pushing the limits: RF field contol at high loaded, Poc. of the st PAC (5), Knoxville, USA, [7] A. Neumann, W. Andes, O. Kugele, J. Knobloch, Analysis and active compensation of micophonics in continuous wave naow-bandwidth supeconducting cavities, Phys. Rev. ST Accel. Beams 3, 8,. [8] O. Kugele, A. Neumann, W. Andes, and J. Knobloch, Adapting TESLA technology fo futue cw light souces using HoBiCaT, Review of scientific instuments, vol. 8, Jul., p. 77. [9] M. Liepe, S. Belomestnykh, J. Dobbins, R. Kaplan, C. Stohman, A new Digital Contol System fo CESR-C and the Conell ERL, Poc. of the th PAC (3), Potland, USA, [] A. Neumann, Compensating micophonics in SRF Cavities to ensue beam stability fo futue Fee-Electon-Lases,Phd thesis, Humboldt Univesität Belin, Gemany, Opeating expeience with SRF acceleatos