Development of a 200 MHz Multiple Beam Klystron Final Report U.S. Department of Energy Grant Number DE-FG02-04ER83916 July 2004 - March 2005 Calabazas Creek Research, Inc. 20937 Comer Drive Saratoga, CA 95070-3753 (408) 741-8680 RLIves@CalCreek.com 3
Introduction The Phase I program met or exceeded all the program goals. During the Phase I program, CCR personnel communicated directly with Dr. Al Moretti at Fermilab concerning the design and specifications. Results of the Phase I program were also presented at the MUON Accelerator Conference in Berkeley, California in February 2005. This review of the design by accelerator scientists provided additional verification of the approach and predicted performance. Results for specific program tasks are described below. In addition, CCR performed a preliminary investigation for a 10 MW device at the request of personnel at Fermi National Laboratory. Results of that task are also provided. Task 1. Electron Gun Design The Phase I program resulted in development of a new type electron gun that provides superior performance for immersed cathodes in flat magnetic fields. The design was achieved during a workshop on computer optimization at North Carolina State University. The students applied modern computational algorithms to optimize the geometry of the electron gun to maximize performance. The analysis produced a design that provides superior performance for low current density beams in a cathode that is simpler to manufacture than convention flat or Pierce cathodes. The electron gun design also results in significantly lower electric field gradients, making the gun easily scalable to higher power by simply increasing the operating voltage and current. A cross sectional view of one of the cathodes is shown in Figure 1. Note that the cathode is convex in shape, rather than concave as in typical Pierce type cathodes.beam focusing is also achieved with a different mechanism. In Pierce cathodes, Anode Focus Electrode Beam Tunnel Cathode Beam Centerline Figure 1. Cross section of one gun in the multiple beam klystron 4
the electrons are focused inward toward the beam axis. As they are accelerated toward the beam tunnel and are compressed by the magnetic field, conservation of canonical angular momentum imparts an azimuthal motion to create the confining v B force. In the dome cathode, the azimuthal velocity is imparted directly at the cathode by injecting the electrons into a crossed electric and magnetic field. This is similar to a magnetron injection gun. In the dome cathode, however, the injection angle is precisely chosen to provide only the azimuthal velocity required to balance the space charge in the beam. This can be seen Figure 2. Close up view of single dome cathode showing beam spiraling by the trajectory paths in Figure 2. Even better performance can be obtained by increasing the magnetic field without significantly changing the beam size. Another advantage is the increased angle between the focus electrode and the axis of the beam. In a typical Pierce gun, the angle between the focus electrode and the outer rays is approximately 67 o. The purpose is to counteract the effect of space charge in the beam near the cathode edge where there is no other confining force. In the dome cathode, the v B force is generated near the emission surface, reducing requirements on the focus electrode to confine the beam. The angle between the focus electrode and the outer electrons is closer to 80 o. Consequently, the focus electrode does not extend as far toward the anode, dramatically reducing the electric field gradient. Even though the gun for this application will operate pulsed, the field gradient is so low that the beam voltage (and gun) could be operated CW. If desired, the gun could be modified to operate grid pulsed to increase the operating efficiency, though this is less of an issue for pulse lengths greater than a few microseconds. The maximum electric field in the gun is less than 10 kv/cm. Fields exceeding 150 kv/cm are allowed for short pulsed operation, and fields below 80 kv/cm are considered acceptable for CW operation. A patent application has been submitted on this concept. The Phase I program examined several gun designs. Simulations were performed for six, eight, and sixteen beam configurations. In the end, the decision to use eight beams was based on the relative cost and performance. More beams allows reduction in the beam voltage, which impacts the cost of the power supply, circuit length (and cost), and ancillary equipment. There are also certain numbers of beams that provide convenient patterns. Figure 3 shows beam simulations for both an eight and a sixteen 5
beam gun. For the RF power level required and a perveance for good efficiency, the voltage difference between an eight beam gun and a sixteen beam gun is approximately 11 kv (68 kv versus 57 kv). Figure 3. Beam simulations for eight and sixteen beam guns. It is not necessa to simulate all beams. One need only simulate one primary beam and all the nearest neighbor beams to determine the predicted performance. The relative power supply costs were discussed with Jeff Casey at Diversified Technologies, Inc. (DTI), one of the countries leading innovators in solid state power supplies. Mr. Casey indicated that there were two principal components for the power supplies. The initial stage converts the raw line power to the appropriate level for energizing the solid state switches. This represents more than half of the cost and would be the same for operation at both 68 kv and 57 kv. The actual voltage is achieved by stacking solid state switches. The incremental difference between the two voltages would represent a modest increase in the total power supply cost. CCR discussed the cost of the cathodes for the two configurations with Mr. Kim Gunther of Heatwave Laboratories, Inc. Heatwave provides both standard and exotic cathodes for a wide variety of applications. Mr. Gunther was formerly the chief engineer at Spectra-Mat, Inc., one of the two largest suppliers of cathodes in the United States. According to Mr. Gunther, each of the cathodes for the MBK would cost approximately $2-3k in large quantities. Therefore, the difference in the cost of cathodes between the eight and sixteen beam configurations would be $16k-$24k. According to Mr. Casey at DTI, this would significantly exceed the difference in cost of the power supplies. Sixteen beams would also complicate the circuit design. As can be seen in the sixteen beam pattern, not all beams would be at the same radius. Consequently, the beam coupling to the cavity fields would be different between the inner and outer beams. Interestingly enough, the circuit length would not be significantly reduced for the sixteen beam design because of the reduced coupling (R/Q) in the cavities. The eight beam design achieved in the Phase I program provides the required beam power with less than 5% beam scallop. The magnetic field is 2.5 times the Brillouin 6
value, which provides excellent beam confinement in the presence of space charge bunching in the circuit. The beams are all sufficiently close to the klystron axis to allow fundamental mode cavities. In summary, the Phase I program developed an eight beam electron gun design that meets all the requirements for the klystron with minimum cost. Task 2. Design of a Fundamental Mode Cavity Without Parasitic Oscillations The goal of the program was to develop RF cavities with good coupling to the electron beam. Both fundamental and ring resonator cavities were initially investigated. The fundamental mode cavity provides the most compact design, with a radius of approximately 90 cm compared to approximately 150 cm for the ring resonator approach. The advantage of the ring resonator is that it provides more room for the electron beams and it is easier to maintain a uniform electric field across the beam tunnels. A disadvantage, in addition to its larger size, is that there are more parasitic resonances near the operating frequency. Parasitic resonances exist within 16 MHz of the desired operating mode for the ring resonator cavity. Accepted practice for fundamental multiple beam cavities is to locate all electron beams within lambda/4 of the cavity axis 5. This keeps the electric field variation across the beam tunnel to manageable values. Several configurations are possible for the beam tunnel section of the cavity. Figure 4 shows three of the configurations simulated. Figure 4. Three fundamental mode cavities analyzed with different variations on the cavity noses Of these three, the reentrant configuration a provided the best combination of field variation across the beam and beam coupling to the cavity fields. Figure 5 shows the electric fields for this configuration as simulated in MAGIC 3D. 7
Figure 5. Ez fields for tapered reentrant, fundamental mode cavity Figure 6 shows the final cavity configuration used in the eight beam klystron design. Task 3. Analysis and Optimization of Electron Beam-RF Wave Interaction Circuit The initial circuit design was performed using KLSC. KLSC is a 2D large signal code for designing single beam circuits. The approach was to determine the circuit parameters for one of the eight beams in the MBK and optimize the performance. Once the single beam parameters were determined, this design was simulated in JPNDISK and 2D MAGIC to verify the performance. In addition, the Figure 6. Cavity configuration used in the eight beam multiple beam klystron design 8
design was simulated by Dr. Ed Wright at Communications and Power Industries, Inc. using their 2DRF large signal code. MAGIC3D was used to simulate the entire multiple beam circuit. This is a significant computational challenge, so only a limited number of simulations could be performed in the Phase I program. Previous experience at CCR and other sites indicated that most of the performance information can be obtained from the 2D simulations. The Phase I program also investigated the linearity requirement for the Tevatron application. In the Tevatron, five injector cavities will be driven by five klystrons. The first injector requires a quiescent RF power of 3 MW, and the fifth cavity requires a quiescent power of 4.35 MW. For this discussion, assume that cavities 2, 3, and 4 require, in order, 3.3 MW, 3.6 MW, and 3.9 MW. As the proton beam loads the five cavities in succession, the RF Input Power Feedback Loop will increase the RF power into each cavity by 15%. These powers are indicated in Table 1. The RF Input Power Feedback Loop requires a well defined gain interval in order to affect these incremental changes in RF power. Figure 7 shows the RF output power versus RF input power. Note that for the five cavities listed in Table 1, all RF power values are well defined. For example, for Injector Cavity No. 5, the quiescent RF power is 4.35 MW for an RF input power of 67 W, and the final RF power is 5 MW for an RF input power of 99 W. These values give Table 1: Tevatron Injector Power Requirement Injector Cavity Quiescent Final Power (MW) Power (MW) 1 3.0 3.45 2 3.3 3.8 3 3.6 4.1 4 3.9 4.5 5 4.35 5.0 Output Power (MW) Input Power (Watts) Figure 7. Simulated output power versus input power a difference in gain of 0.6 db with a difference in RF input power of 32 W, or equivalently, a difference of 1.5 db. 9
Dr. Al Moretti at Fermi National Laboratory reviewed the results of this analysis and the RF performance predictions from the simulation codes. Dr. Moretti confirmed that these results meet all the requirements for the proton drift tube linac. Task 4. Design of the RF Windows, Input and Output Coupler, and Spent Beam Collector Information on the output window requirement was provided by Dr. Moretti at Fermi National Laboratory. It is desired that the final output from the klystron be in coaxial waveguide; consequently, a coaxial window was designed for the output cavity. Since the R/Q of the output cavity is low, it will be necessary to extract output power in two waveguides located 180 o apart. This will avoid field distortions that could affect performance. The output window was designed using CCR s scattering matrix code CASCADE. This code is used at universities and industries around the world, and its accuracy is well validated. CCR s optimization feature was used to design a window with superior performance around the operating frequency. Figure 8 shows the simulated voltage standing wave ratio (VSWR) of the window. Figure 8. Predicted VSWR of the output window 10
Figure 9 shows a solid model of the output window assembly. This model does not show the cooling circuit. The average power level in the window is fairly modest, so no significant thermal issues are anticipated. The input cavity will be driven through a coax cable loop coupled to the cavity fields. It is anticipated that all beams will be collected in a common collector. The low current density and the size of the collector will allow grounded structures along the axis. This will ensure that no electrons are reflected from a potential well in the collector. Such a To Output Cavity Alumina Ceramic Matching Elements Flange Figure 9. Solid model of the preliminary output window design potential depression can occur when the individual beam tunnels are terminated and the beams are affected by space charge forces from adjacent beams. Results from beam simulations are shown in Figure 10. These simulations are for operation without generation of the RF power. This represents the worst case in terms of power deposition, since all the original beam power remains. Note the Figure 10. Collector simulation with three of the eight electron beams. The left view shows a side view, and the right image is an end view. The mechanical structure is sliced along one axial plane for these views. mechanical structure on the axis. This structure will prevent creation of a potential well and will be heavily cooled to handle the thermal load from beam impact. 11
Task 5. Preliminary Mechanical Design Figure 11 shows a solid model of the complete klystron and magnet assembly. The magnet consists of series of coils producing a sufficiently flat, 200 G, magnetic field at the electron beam positions. The four support rods for the magnet coils are iron and serve as the return path for the magnetic field, so no additional iron structures are required around the klystron. The square plates at each end are also iron. The magnetic field is relatively simple to produce because of the immersed electron gun. The open configuration will greatly simplify installation and setup, since all parts of the klystron, such as water and electrical connections, will be directly accessible through the coils. Also, the klystron and magnet will be shipped as a single unit on a movable sled.. Figure 11. Solid model of 5M, 201 MHz, multiple beam klystron with magnet Note also that the high voltage ceramic is designed to operate in air, which eliminates requirements for an oil tank. The cathode and heaters will be energized by leads brought in through the gun polepiece (not show). There will be two output windows. The necessary waveguide hardware to combine the power into a single guide will be developed and built in the Phase II program.figure 12 shows a sectioned view of the klystron. A preliminary outline drawing of the klystron and magnet are shown in Figure13. The total length of the tube and magnet is approximately 16 feet. Preliminary calculations indicate that the klystron will weight approximately 5,800 lbs and the magnet 12
will weigh approximately 4,200 lbs. These values are very rough and should not be used for detailed planning. Figure 12. Sectioned view of the 201 MHz multiple beam klystron Figure 13. Preliminary outline drawing of the MBK and magnet. The dimensions are in feet. The mechanical design indicates that the klystron and magnet can easily meet the size specifications for the program. 13
Task 6. Preliminary Investigation of a 10 MW MBK Personnel at Fermi National Laboratory requested that CCR investigate a 10 MW MBK at the same operating frequency. Essentially no geometrical changes were made to the 5 MW design, other than a slight increase in the beam tunnel diameter. The only significant change was an increase in the voltage from 68 kv to 86 kv. The total current becomes 200 A divided in eight beams. The cathode loading is only 1.2 A/cm 2, well below the value that would result in an issue with lifetime. Simulations with KLSC indicate that 10 MW can be achieved with high efficiency and little change in klystron size or configuration. Plots of efficiency and output power as a function of input power are provided in Efficiency (%) Output Power (MW) Input Power (W) Input Power (W) Figure 14. KLSC simulation results for 10 MW MBK Figure 14. Some additional design work will be required to improve the shape of the output power versus input power response, but this is not anticipated to be a problem. In summary, the Phase I program achieved a preliminary multiple beam klystron design that achieved all the goals of the program. An email from Dr. Al Moretti, manager of the RF system for the Tevatron confirmed that all specification are met. Simulations predict that the efficiency will exceed 55% with sufficient gain for the application. The number of electron beams was selected based on the total cost of the klystron and power supply. CCR believes this represents the most cost effective configuration when the total system is considered. There appear to be no significant technical challenges that would prevent the klystron from being successfully built, tested, and delivered in a Phase II program. 14
References [1] Confined Flow Multiple Beam Guns for High Power RF Applications, L. Ives, M. Miram, A. Krasnykh, Valentin Ivanov, 13th IEEE Conf. Plasma Sci., Las Vegas, NM, June 2001. [2] Lawrence Ives, George Miram, David Marsden, Max Mizuhara, Tom Robinson, Jorge Guevara, Anatoly Krasnykh, Valentin Ivanov, Construction and Test of a Confined Flow Multiple Beam Gun for a 50 MW Klystron, Fifth IEEE Intern. Vacuum Electronics Conference, Monterey, CA pp 284-285, April 2004. [3] Development of a 50 MW Multiple Beam Klystron, U.S. Department of Energy Small Business Innovation Research Program, DE-FG03-02ER83379, July 2002 - June 2005. [4] A. Balkcum, E. Wright, H. Bohlen, M. Cattelino, L. Cox, M. Cusik, K. Eppley, S. Forest, F. Friedlander, A. Staprans, L. Zitelli, Operation of a 1.3 GHz, 10 MW Multiple Beam Klystron, Fifth IEEE Intern. Vacuum Electronics Conf., Monterey, CA, April 2004, pp 280-281. [5] Dr. Edward Wright, Communications and Power Industries, Inc., private communications, October 2004. [6] Development of a 10-MW, L-Band, Multiple-Beam Klystron for TESLA, E. Wright, A. Balkum, H. Bohlen, M. Cattelino, L. Cox, L. Falce, F. Friedlander, B. Stockwell, and L. Zitelli, Communications and Power Industries, Inc. [7] Gridded Sheet Beam Gun for High Power RF Applications, U.S. Department of Energy Small Business Innovation Research Grant Number DE-FG03-01ER83209, September 2001 through May 2004. [8] Sheet Beam Klystron for Accelerators, U.S. Department of Energy Small Business Innovation Research Grant Number DE-FG03-03ER83617, July 2003 through April 2004. [9] 20 kw CW, L-Band Klystron for Driving Superconducting Accelerator Cavities, U.S. Department of Energy Small Business Innovation Research Grant Number DE-FG03-02ER83378, September 2002 through April 2005. [10] Design of a 10 MW, 91 GHz Gyroklystron for Linear Accelerators, U.S. Department of Energy Small Business Innovation Research Grant Number DE- FG03-99-ER82754, September 1999 through June 2003. [11] Design of a 10-MW, 91 GHz Frequency Doubling Gyroklystron for Advanced Accelerator Applications, W. Lawson, R.L. Ives, M. Mizuhara, J. Neilson, and M.E. Read, IEEE Trans. Plasma Sci., Vol. 29, No. 03, June 2001. [12] S Bowater; D Findlay; D Wilcox, RF Tube comparisons and other requirements, RF Basic technology bid Rutherford Appleton Laboratory, 23rd Feb 05, http:// www.hep.ph.ic.ac.uk/~longkr/uknf/rf/2005-02-23/. 15