CERN Chopper Status F Caspers M Paoluzzi CERN

CERN Chopper Status F. Caspers, M. Paoluzzi CERN, HIPPI 08 meeting October 29 th, 2008

Contents u Introduction n u CERN chopping scheme Layout Technical requirements Evolution of the SPL chopper n n Modifications in 2005 Status by September 2006 l n n n u u What happend since end of 2006 Status by April 2008 Some photos What is still to be done Drivers Status Acknowledgements Most of these slides were already shown in May 2008 at the WP 4 meeting CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 2

SPL Layout u u Superconducting Proton Linac (SPL) Fast beam chopping at Ekin = 3 Me. V ; thus is about 8% n u u u Fast chopper required to establish desired beam pattern RFQ: Radio frequency quadrupole DTL: Drift tube linac CCDTL: Coupled cavity DTL PIMS (PI-Mode Structure) = 0. 65, = 0. 92 : superconducting cavities CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 3

CERN Chopping Scheme u Most demanding scheme for SPL operation: n n n cutting out three bunches out of eight repetition rate 44 MHz bunch spacing 2. 84 ns ; 10 to 90% rise and fall time required < 2 ns 8*2. 84 ns ( 44 MHz) CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 4

Chopper Line (1) u u Chopper line lattice designed such as to magnify the kick from the chopper; this reduces required kick field The chopper plates have to be installed in the quads; this saves length and reduces space-charge related emittance growth Chopper off Kick-magnifying quad Chopper on CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 5

Chopper Line (2) bunching cavities chopper beam dump CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 6

Modifications in 2005 u u The chopper plates are no longer DC-wise floating (no more triaxial mode of operation) Now we have a coaxial instead of a triaxial chopper structure The triaxial version was meant for simultaneous dual mode of operation, i. e. 0 to 10 MHz: electrostatic deflector, above 10 MHz travelling wave mode Removal of isolating units both in water cooling circuits and coaxial driving lines CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 7

Meander lines u u u Initially samples of the meander lines were produced at CERN While all parameters were basically ok (electrical, vacuum), reproducibility of certain electrical parameters (electrical length, match) was not always perfectly satisfactory After the accomplished proof of principle with CERN technology, a supplier that can well control all the process parameters was needed; a possible change in technology is not a problem as long as the key properties are preserved Kyocera was eager to enter in a cooperation with CERN and willing to adapt their technology to our needs The plate that was recently furnished compared well in all aspects with the best CERN samples and we hope that the promised good reproducibility will show up in reality CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 8

Meander line: Technology u Technological differences between the CERN and Kyocera meander structures: CERN Kyocera CERN orders from Wesgo (D) ceramic plates with fixation holes and 10 to 15 m thick homogeneous Mo. Mn layer (fired at 1400 C in hydrogen atmosphere Kyocera produces the alumina plates in-house. Then a meander pattern is created in thick film silver paste of about 10 to 15 m thickness. This thick film silver paste has after firing a much higher resistivity than bulk silver (factor 5 to 10) The meander structure is etched into the Mo. Mn layer. Afterwards a 1 to 2 m thin-film layer of Ag is attached by sputtering. Onto the Ag thick-film layer 30 to 40 m copper are deposited electrochemically In a final step this silver-coated Mo. Mn layer gets Finally 1 to 2 m of Au are applied for good another 30 m silver by electrochemical contacts and protection against oxidation deposition In the meantime (october 2008) I learned that it is possible to produce nonmagnetic Nickel layers by adding 18 % of phosphor [F. C. ] CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 9

CERAMIC PLATE Process Flow Ceramic incoming inspection Metallize printing (Ag thick-film) Firing Cu Plating Machining Au Plating Electrode for electrochemical deposition removed by grinding Machining Resistance Testing Grinding Area Final inspection RF Property Testing Packing CERN HIPPI meeting October 2008 Courtesy: Kyocera Caspers, Paoluzzi CERN chopper status October 2008 10

Status by September 2006 u u First meander line plate from Kyocera were received in June 2006 but it turned out that the attenuation was too high In the second iteration the technological parameters were properly adjusted and the last sample was very satisfactory After extensive electrical tests this single plate was installed in the chopper tank Vacuum, leak and heat tests performed successfully CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 11

Electric Measurements: Transmission Attenuation u u Measurements performed on single chopper plate with an image plane 10 mm above the line’s surface to simulate the presence of a second plate Frequency domain transmission A DC resistance of 1. 1 was measured, which agrees very well with the low-frequency limit of the measured attenuation 3 d. B bandwidth 940 MHz. If there was no phase distortion the rise time would be Attenuation over one chopper plate Generator bandwidth u All rise times quoted are 10 to 90% values CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 12

Transmission Step Response u u Response for 0 to 700 MHz low pass step function (Kaiser Bessel weighting function with = 6) Comparison between a measurement with and without the image plane. Due to the high electric field energy in the alumina the kick field does not change much when the symmetry is broken Measured rise time trm = 1. 771 ns, to be compared with tri = 1. 407 ns of input pulse; structure rise time This is a conservative estimate of tr since the tti is rather short and we get into the highly dispersive region of the response CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 13

Phase Distortion u u From the measured phase without image plate the electrical delay of 16. 73 ns (linear term) was removed With image plane (realistic configuration) the electrical length was 16. 83 ns, within 0. 1 ns of the required value The remaining phase is not flat as for a dispersion-free line; thus we have phase distortion The phase distortion is due to coupling between adjacent lines in the meander structure. This coupling increases quickly with frequency like in a microstrip directional coupler In an ordinary first-order low-pass the 45 degree points coincide with the 3 d. B points. Here they are at 375 MHz, i. e. much lower than the 3 d. B points CERN HIPPI meeting October 2008 Phase without image plane Generator bandwidth u Caspers, Paoluzzi CERN chopper status October 2008 14

u u Very good impedance match of meander line to 50 : reflection in frequency domain of the order of -30 d. B below 500 MHz These data were measured on a test jig consisting of a single plate with SMA connectors fixed on either side CERN HIPPI meeting October 2008 Generator BW Reflection Caspers, Paoluzzi CERN chopper status October 2008 15

Reflection Step Response u u Response for 0 to 700 MHz low pass step function S 11 very small, of the order of 0. 02 which is another indication of good match The line impedance is not perfectly constant over the meander length as can be seen from the bump at t = 10 ns. Towards the end of the line an apparent increase in line impedance can be seen. This is an artifact caused by the lossy line; could be corrected numerically CERN HIPPI meeting October 2008 Twice the line length of ¼ 17 ns Caspers, Paoluzzi CERN chopper status October 2008 16

u u u It was tried to adjust the electrical delay of a chopper plate by modifying the metal ground plane Cutting a longitudinal groove into the ground plane reduces the effective e and thus increases the group velocity on the line. Since the variation in line impedance is over distances much shorter than the wavelength, the other electrical properties should not be affected much For two 5 mm wide and 3 mm deep grooves a 5% decrease in the electrical length was found on a CERN plate CERN HIPPI meeting October 2008 Generator bandwidth Tuning of Electrical Length Caspers, Paoluzzi CERN chopper status October 2008 17

What happend since end of 2006 ? (1) u u We have seen that the originally used aluminum support plates for the alumina ceramic are not usable since they liberate internal stress when heated and show considerable deformation ( up to several 100 micron); this has even led to the destruction of ceramic substrates when mounted (fortunately only old ones, which we were allowed to destroy) This deformation is a very critical issue since thermal contact betwen the ceramic plate and the metal support depends strongly on the small vacuum gap between the ceramic and the metal. In fact this point has been subject of many discussions and the heat transfer between the ceramic and metal plate is essentially an electromagnetic tunneling effect ! The wavelength corresponding to room temperature is about 10 micron we the average space between metal and ceramic plate has to be below a few microns in order to meet the tunneling condition (similar to a microwaveguide below cutoff signal transmission) CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 18

What happend since end of 2006 ? (2) u u u Thus it was decided to replace those aluminum plates by stainless steel plates which are heat treated (to release all internal stresses) and copper coated (without using nickel as intermediate layer [magnetic field!) and finally having a flash of gold to prevent oxydation. The copper coating towards the ceramic layer is required for keeping losses of the image currents from the microstrip structure within reasonable limits. One may raise the question about degradation of thermal conductivity due to the use of stainless steel n n The anser is that this completely irrelevant since thermal impedance of the vacuum gap is by far the dominating contribution. This has been shown experimentally in vacuum in 2006. If one would solder/braze the ceramic plate directly onto the metal support thermal resistance would reduce by more than factor of 30. This is relevant and also under discussion in the frame of applying this concept at GANIL CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 19

What happend since end of 2006 ? (3) u u u We also noticed that due to changes in production technology at Kyocera (now all parameters are well under control and production is reproducible) certain samles did not meet out specs and we had to return them. As a consequence we found a considerable scatter (up to 1. 5 ns) in electrical delay as well as a systematic offset between the different ceramic plates which were delivered over the last 3 years. Thus we had to apply means in reducing the scatter and getting the mean value of the delay right (16. 7 ns nominal) For this problem we decided to apply the „groove“technique for fine tuning , which of course only works on one direction i. e. reducing the el. delay. In this context the question comes up: n n n How is the electrical delay (which is frequency dependent, see privious slides) exactly to be defind? Do we take the el. delay at low frequencies or at 100 MHz or at 200 MHz ? Or do we refer to the 50% point of the rising slope of the step excitation ? CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 20

Transmission Attenuation Status April 2008 Loss values of 1 db at 200 Mhz are ok. and consistent with previous data. CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 21

Transmission Step Response Status April 2008 We will place plate 1 and plate 2 (red and green trace) into one chopper unit as they match rather well (c. f. 50% point) and plate 3 and plate 4 into the other unit CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 22

El. Delay, Status April 2008 CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 23

Reflection (frequency domain) Status April 2008 CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 24

Reflection (time domain, step response) Status April 2008 Note that for the blue and the green trace the char. impedance is slightly too low (refl factor -2 %) and for the red trace is slightly too high (+2 %) CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 25

Some photos (1) Rear side with water cooling circuit and RF connectors for rigid coax lines to the vacuum feedthrough grooves Note, that the grooves are good for venting (vacuum) the confined space between the ceramic plate and the metal support CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 26

Some photos (2) CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 27

What is left to be done u u u u Final assembly of the chopper plates and the tanks …. . done by now (October 2008) the chopper tanks are already in the beamline under construction. Final vacuum test…done by now (October 2008) Final electrical test (DC high voltage and DC high current) under vacuum; such tests were done already 2 years ago , but not on the same ceramic plates DC tests with beam (DC bias voltage on the plates with open termination = Static deflection) RF power tests under vacuum without and with beam when the pulser is available. Production of spare quadrupoles ! Whatever else I might have forgotten…. CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 28

Drivers status FID Technology Pulsers u u u Positive and negative units have been delivered and characterized. Units succesfully tested at 1 MHz, 1 ms burst and 50 Hz repetions rate for two weeks continuos operation. Reliability problems due to incorrect triggering could be highligted. CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 29

Drivers status Achieved parameters PARAMETER REF. SPEC. ACHIEVED Output voltage Vout 700 V 670 V Load impedance Zout 50 OK Pulse length TWout 8 ns to 1 s OK Minimum rep. freq. fmin Single pulse OK Maximum burst length TB 1 ms f. Bmax 50 Hz Maximum burst rep. freq. CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 30

Drivers status Parameters to improve (1) PARAMETER Minimum off pulse time REF. Toff SPEC. 14 n No No ACHIEVED 40 ns OK OK Maximum repetition frequency fmax 45 MHz 15 MHz Rise time (10 % - 90) Rise time (3 % - 90) Fall time (90 % - 10) Fall time (90 % - 3) TR TRR TF TFF <2 ns <2. 5 ns <3. 7 ns Max. voltage between two pulses |Vn| <2 % of Vout <10 % Toff dep. on TWout Toff dep. on rep. freq. CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 31

Drivers status Parameters to improve (2) PARAMETER Propagation delay time REF. TD No No TD dependent on TWout TD dep. on rep. freq. Pulse dist. | TWout - TWin| TPd dep. on TWout TPd freq. CERN HIPPI meeting October 2008 SPEC. <500 ns ACHIEVED <100 ns No Yes: ≤ 0. 2 ns @ 1 MHz ≤ 1. 5 ns @ 15 MHz Pd <5 ns No No <3 ns Yes: ≤ 0. 2 ns @ 1 MHz ≤ 1. 2 ns @ 15 MHz Caspers, Paoluzzi CERN chopper status October 2008 32

Drivers status Synchronization electronics Fast synchronization electronics prototype has been developed. It is composed of : • a synchronism detector. • a fast adjustable delay. • digital pulse to pulse loop that locks the rising front of the amplifier output pulse to a reference pulse. • This compensates slow delay variations (max 30 ps/pulse) and stabilizes the delay within ~100 ps. A similar loop can also be built to compensate the delay variations. CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 33

… finally u u u The achievement of the full specifications seems to be a very complicated task and the experience gained until now is very important. The rise and fall time are only part of the difficulties as pulse delay and pulse length distortion are also very difficult to be maintained within a fraction of ns. Even if only partially fulfilling the specifications, 3 units of each polarity will be delivered at the end of November 2008. They will already implement some improvements. The availability of the drivers will allow testing in 2009 the overall chopper system without beam. The development will then continue to achieve full specs. CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 34

Acknowledgements u u u We would like to thank the AB-RF workshops for assembling the tank, F. Wurster, M. Nagata and Mr U. Behrens from Kyocera for fruitful cooperation in development and implementation of the technologies for printing the meander structure J. Borburgh for assistance with the heat transfer measurements and in particular Vladimir Bretin and his crew from the RF mechanical workshop for his infinite patience and diligence in mounting and demounting this structure. Thanks also to R. Garoby and T. Linnecar for support CERN HIPPI meeting October 2008 Caspers, Paoluzzi CERN chopper status October 2008 35