- Email: [email protected]
Performance of a high-power klystron using a BI cathode in the KEK electron linac
Applied Surface Science 146 Ž1999. 84–88
Performance of a high-power klystron using a BI cathode in the KEK electron linac S. Fukuda b
a,)
, K. Hayashi b, S. Maeda b, S. Michizono a , Y. Saito
a
a KEK (High Energy Accelerator Research Organization), 1-1 Oho, Tsukuba, Ibaraki 305-0801, Japan Mitsubishi Electric Corporation, CEW, 8-1-1 Tsukaguchi, Honmachi, Amagasaki, Hyogo 661-8661, Japan
Abstract Fifty-megawatt high-power klystrons in the KEK electron linac have used a larger diameter cathode of 85 mm. Stable operation requires a careful manufacturing process of the cathode and a low-gradient gun design. Therefore, the klystron manufacturing, including a cathode processing, has been improved so as to obtain stable performance. This paper presents the klystron improvements associated with using the BI cathode and klystron operation statistics for the past 17 years. q 1999 Elsevier Science B.V. All rights reserved. PACS: 52.80; 79.40; 84.40 Keywords: Klystron; Oxide-coated cathode; BI cathode; Arcing in gun; Lifetime data
1. Introduction At the KEKB 8 GeV electron-positron linac, Sband 50-MW high-power pulsed klystrons have been used as the rf source. In this klystron, a large diameter of 85 mm is used for the cathode of a convergent-type electron gun. This large-diameter cathode requires a careful manufacturing process in order to obtain sufficient reliability, an arcing-free gun, a low barium evaporation rate, and long life. A high voltage of more then 300 kV is applied between the beam-focusing electrode and the anode under the condition of a very low pressure, such as a 10y7 to 10y8 Pa, and a high temperature, such as a 800 to )
Corresponding author. Tel.: q81-298-64-5697; Fax: q81298-64-7529; E-mail: [email protected]
10008C, depending on the cathode type. At KEK, more than 200 high-power klystrons have been used since 1983. We have experienced many problems, such as arcing in the gun region, window breakdown, and other failures. During the early stage of linac operation, the most serious problems were due to arcing; this seemed to be related to the cathode processing and the tube baking-evacuation process, since the anode of the failed klystron had many barium compounds on it. Thus, we have not only been developing the electrical design of the klystron, but also improving the manufacturing process in order to solve this problem. In this paper, we describe the development of the cathode and the electron gun of the klystrons used in the KEK linac. We also give operation life data for these last 17 years and analyze the causes of failures.
0169-4332r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 9 - 4 3 3 2 Ž 9 9 . 0 0 0 7 0 - 7
S. Fukuda et al.r Applied Surface Science 146 (1999) 84–88
2. History of the KEK S-band klystron development The 8 GeV electron linac w1x has been upgraded from the PF 2.5-GeV linac. Fifty-eight 50-MW klystrons were used as the rf source. In 1982, a 30-MW klystron was developed based on the design of the SLAC XK-5 klystron w2x by MELCO ŽMitsubishi Electric.. The XK-5, which has an output of 35-MW peak power at a 270-kV applied voltage, had been used in the SLAC 2-mile accelerator. Fig. 1 shows the klystron cathode. In the case of an S-band klystron ŽPV3030A2., 310 kV of a pulse voltage is applied between the beam-focusing electrode ŽBFE. and the anode. The nearest insulation length between them is only 18 mm, and the maximum field gradient reaches 249 kVrcm on the BFE and 235 kVrcm on the anode. The typical parameters associated with the gun conditions for the XK-5 and the 50-MW klystron w3x are listed in Table 1. The development history of the S-band klystron for these 17 years is as follows: the first 30-MW model ŽPV3030. was developed based on the XK-5, except for the shape of the BFE. An oxide-coated cathode with a diameter of 80 mm was used. During this period, we experienced on arcing troubles in the klystron gun. In order to solve these problems, we changed the shape of the BFE and examined the low-gradient gun. At the same time, we began to improve the manufacturing pro-
85
Table 1 Gun parameter of the XK-5 and 50-MW klystron Item
Unit
XK-5 Klystron
50-MW Klystron
Applied voltage Cathode diameter Cathode type
kV mm
Maximum field on BFE Maximum field on anode
kVrmm kVrmm
265 80 Oxide cathode 24.4 28.4
310 85 BI cathode 22.0 21.4
cess. In 1987, we replaced the oxide-coated cathode with a dispenser cathode Ža barium impregnated cathode.. This BI cathode model was called PV3030A1. We then adopted a low-gradient gun design for the dispenser cathode gun. This model was named PV3030A2. The PV3030A3 was introduced by replacing an insulating high-voltage ceramic seal with a larger one in order to increase the applied voltage to 310 kV. Finally, taking account of the cathode lifetime, we redesigned the gun and reformed the PV3050. This tube adopted a 85-mmdiameter dispenser cathode with a lower gradient BFE and an anode. The insulation ceramic was also replaced by a larger one; it has a capability up to a 350 kV applied pulse voltage w3x. We summarize these developments in Fig. 2.
3. Improvement of the manufacturing process Since 1985, we have improved the manufacturing process in order to solve arcing problems. A suitable
Fig. 1. Electron gun of the high-power klystron.
Fig. 2. Development history of the high-power S-band klystron.
86
S. Fukuda et al.r Applied Surface Science 146 (1999) 84–88
choice of materials was very important w4x; therefore, we introduced a vacuum-melted SUS-316L stainless steel as the BFE material and class-1 oxygen-free copper, which is also melted in a vacuum, as an inside tube material. The latter satisfies the ASTM F-68: the world’s highest standard for copper purity. For the vacuum-firing process of parts, the baking temperature of the BFE was lowered from 1000 to 6008C so as to avoid any surface roughness due to the growth of a grain boundary. It was pointed out that the impurity was concentrated on the grain boundary, which resulted in a rough surface; this was not desirable for high-voltage applications. At the same time, the BFE was polished using a paste containing small diamond particles Žparticle size of about 1–6 mm. in order to make a mirror-like surface finish. Since the anode was made of copper, the surface smoothness was achieved mechanically. For the process of pre-firing the cathode and its assembly, a new pre-firing furnace was made, and the baking temperature was carefully controlled. The cathode and its assembly were separated by a molybdenum cylinder in order to avoid any contamination of the assembly surface when the dispenser cathode was fired in order to evaporate any excess barium in the BI cathode. The temperature of a cathode post, made of stainless steel, was kept at 6008C by induction heating, similar to the case of parts baking. The cathode was carefully processed with setting a suitable filament voltage while monitoring the cathode temperature using a radiation thermometer. For the entire tube-baking process, we improved several points concerning the baking and evacuating furnace: the roughing pumping system, a manifold-baking procedure, and an investigation of residual gas using a mass analyzer. The baking temperature was raised from 480 to 5508C, and total baking hours were increased from 110 to 130 h. Due to these improvements, the final pressure in the tube reached 10y6 Pa, which was observed at the pumping head position. The forming gas Ž92% nitrogen gas and 8% hydrogen gas. was first filled in the furnace. However, it was found that hydrogen gas penetrated the tube through its copper wall; this rate increased rapidly as the baking temperature rose. Generally, the pumping speed of hydrogen is not very high for a diffusion pump or a turbomolecular pump. Although it is said that hydrogen gas was not harmful to the
cathode, any kind of gas is possibly harmful concerning arcing in the gun region. We thus changed the forming gas furnace to a vacuum furnace. Although these improvements resulted in a very low final pressure in the tube, they were finally found not to be effective for solving arcing phenomena in the tube with the oxide cathode. We thus replaced the oxide-coated cathode with a dispenser cathode. We could obtain a final pressure of less than 10y7 Pa for the BI cathode tube with the improved processing. At the same time, we decreased the arcing phenomena.
4. Oxide cathode and dispenser cathode The oxide-coated cathode used in the 30-MW klystron was developed by MELCO as a mush cathode, which comprised a matrix of coarse nickel particles Ž40–100 mesh size. sintered to the underlying nickel base in order to increase the porous layer; also, the usual carbonate cathode coating was soaked into the pores. Since the conversion process was inevitable in the tube, it was important to reduce barium evaporation. In the MELCO mush cathode, the initial evaporation rate just after conversion was reported large; the authors have mentioned the reduced reaction of carbon, which came from the binder material w5x. Too much reducing agents led to a high evaporation rate, while a small amount of reducing agents was important, and was added in the nickel base. The reduced reaction due to the residual hydrogen gas, which came from the forming gas,
Fig. 3. Anode of the failed klystron with an oxide-coated cathode.
S. Fukuda et al.r Applied Surface Science 146 (1999) 84–88
87
Fig. 4. Scandate dispenser cathode of the 50-MW klystron.
seemed to play an important role for barium evaporation of the mush cathode, since the replacement of the forming gas baking furnace by the vacuum bak-
ing furnace resulted in heavier arcing and a short tube life, as shown in Section 5. Although more accurate examination may be necessary, it has not
Table 2 Tube running statistics from 1979 to 1996 Fiscal year
Cathode
79 80 81 82 83 84 85 86 87
Oxide Oxide Oxide Oxide Oxide Oxide Oxide Oxide Oxide Oxide total BI BI BI BI BI BI BI BI BI BI BI total
87 88 89 90 91 92 93 94 95 96
MTBF Žh.
Total No.
Failed No.
Failure life Žh.
Failure cause Arcing Window
Others
Living No.
Living life average Žh.
3902 9242 16,588 10,317 19,934 9950 13,409 4351 4342 11,187 34,632 29,911 23,505 19,780 14,780 69,072 – – – – 30,840
4 20 20 9 13 13ŽU 1. 12ŽU 1. 15 7 113 7 20 18 18 15 12 14 13 23 15 155
4 19 19 8 12 12 11 14 7 106 7 16 17 12 7 2 0 0 0 0 61
3902 9050 15,965 10,054 18,753 9905 13,409 3524 4342 10,783 34,632 27,948 23,505 17,637 6393 6295 – – – – 22,265
2 13 11 5 6 10 7 13 5 72 1 1 1 0 1 0 0 0 0 0 4
1 1 6 1 4 2 4 1 1 21 5 11 9 7 2 0 0 0 0 0 34
0 1 1 1 1 0 0 1 0 5 0 3 0 3 8 8 13 13 23 13 84
– 3657 11,277 2120 14,170 – – 11,568 – 8558 – 10,467 – 8571 7448 15,694 8484 6058 3989 1717 11,012
1 5 2 2 2 0 0 0 1 13 1 4 7 5 4 2 0 0 0 0 23
88
S. Fukuda et al.r Applied Surface Science 146 (1999) 84–88
yet been solved. Fig. 3 shows the anode of the failed klystron with an oxide cathode. Finally, we changed the cathode to a dispenser type. We tried to use two Ir-coated cathodes, and then adopted the scandate cathode of Spectra-mat, based on experience with the SLAC 5045 50-MW klystron. The vacuum characteristics were better than for the oxide cathode; the BI cathode was vacuumfired in a separate envelope on a separate pump station so it was not necessary to worry about contamination because any excess barium evaporation deteriorates the klystron tube inside. The BI cathode features greater strength against residual gas than an oxide-coated cathode w6x. The operating temperature was decreased to about 9508C due to recent new technology. Fig. 4 shows the scandate dispenser cathode of the 50-MW klystron.
5. Run data statistics and BI cathode performance Table 2 shows the run data of the last 17 years for KEK high-power klystrons. This table gives the run data contributed to the KEK-TRISTAN operation up to 1996. This was divided into two parts; the upper is the data of klystron with the oxide cathode; the lower is the data of that with the BI cathode. The main cause of failures for that above was arcing; it reached about 70% of total tubes. Also, tubes manufactured in fiscal years 1986 and 1987 were found to have short lifetimes of around 4000 h due to the baking furnace change from the forming-gas one to the vacuum furnace, as described. Living tubes with an oxide cathode were all disposed while upgrading the linac. The lower table shows the lifetime data of a klystron with a dispenser cathode. The main causes of failures changed from arcing to window failures. In spite of the fact that the average applied voltage to the cathode increased and the average output power increased, arcing decreased predominantly. MTBF of the BI cathode klystron is about three times higher than that of the oxide cathode klystron. However, the lifetime of the failed BI cathode klystron was only
two times higher than for others, since dead tubes were fewer and the contribution to lifetime came mainly from the initial failure of BI cathode tubes. Other causes of failure included the operation stop of the old tubes to replace them with 50-MW tubes. They could be used, and, in fact, were used in other facilities. The PV3030A2 was all replaced with new klystrons, such as the PV3030A3 and 50-MW. Taking into consideration these cases, the realistic lifetime is expected to be more than 30,000 h. Window problems will be the main failures in the future, since our windows were fed the biggest power in a single pill box window when we use 50-MW klystrons.
6. Summary The developments of the high-power klystrons, which used a large-diameter cathode, are described. We succeeded in developing a stable 50-MW klystron by improving the manufacturing process, replacing the oxide-coated cathode with a BI cathode and adopting a low-gradient gun design. A very low pressure in the klystron, even with heater power on, was realized by performing careful cathode processing. Serious arcing problems were solved by using a low-gradient gun with a BI cathode. From the data of lifetime during these 17 years, we concluded that the cause of failures due to arcing decreased from 70% to 3% after replacement using BI cathode.
References w1x A. Enomoto, Proc. of 1996 Int. Linac Conf., CERN, Geneva, Switzerland, 1996, p. 633. w 2 x G.T. Konrad, Proc. of 1984 Int. Linac Conf., SeeheimrDarmstadt, West Germany, 1984, p. 293. w3x S. Fukuda, S. Michizono, K. Nakao, Y. Saito, S. Anami, Nucl. Instrum. Meth. Phys. Res. A 368 Ž1996. 561. w4x W.H. Kohl, Materials and Techniques for Electron Tubes, Reinhold, New York, 1960. w5x T. Hata, J.Kai, M. Koitabashi, K. Sano, Mitsubishi Denki Gihou, Vol. 40-6, 1966, p. 1034 w6x R.O. Jenkins, Vacuum 19 Ž8. Ž1969. 353.
Performance of a high-power klystron using a BI cathode in the KEK electron linac S. Fukuda b
a,)
, K. Hayashi b, S. Maeda b, S. Michizono a , Y. Saito
a
a KEK (High Energy Accelerator Research Organization), 1-1 Oho, Tsukuba, Ibaraki 305-0801, Japan Mitsubishi Electric Corporation, CEW, 8-1-1 Tsukaguchi, Honmachi, Amagasaki, Hyogo 661-8661, Japan
Abstract Fifty-megawatt high-power klystrons in the KEK electron linac have used a larger diameter cathode of 85 mm. Stable operation requires a careful manufacturing process of the cathode and a low-gradient gun design. Therefore, the klystron manufacturing, including a cathode processing, has been improved so as to obtain stable performance. This paper presents the klystron improvements associated with using the BI cathode and klystron operation statistics for the past 17 years. q 1999 Elsevier Science B.V. All rights reserved. PACS: 52.80; 79.40; 84.40 Keywords: Klystron; Oxide-coated cathode; BI cathode; Arcing in gun; Lifetime data
1. Introduction At the KEKB 8 GeV electron-positron linac, Sband 50-MW high-power pulsed klystrons have been used as the rf source. In this klystron, a large diameter of 85 mm is used for the cathode of a convergent-type electron gun. This large-diameter cathode requires a careful manufacturing process in order to obtain sufficient reliability, an arcing-free gun, a low barium evaporation rate, and long life. A high voltage of more then 300 kV is applied between the beam-focusing electrode and the anode under the condition of a very low pressure, such as a 10y7 to 10y8 Pa, and a high temperature, such as a 800 to )
Corresponding author. Tel.: q81-298-64-5697; Fax: q81298-64-7529; E-mail: [email protected]
10008C, depending on the cathode type. At KEK, more than 200 high-power klystrons have been used since 1983. We have experienced many problems, such as arcing in the gun region, window breakdown, and other failures. During the early stage of linac operation, the most serious problems were due to arcing; this seemed to be related to the cathode processing and the tube baking-evacuation process, since the anode of the failed klystron had many barium compounds on it. Thus, we have not only been developing the electrical design of the klystron, but also improving the manufacturing process in order to solve this problem. In this paper, we describe the development of the cathode and the electron gun of the klystrons used in the KEK linac. We also give operation life data for these last 17 years and analyze the causes of failures.
0169-4332r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 9 - 4 3 3 2 Ž 9 9 . 0 0 0 7 0 - 7
S. Fukuda et al.r Applied Surface Science 146 (1999) 84–88
2. History of the KEK S-band klystron development The 8 GeV electron linac w1x has been upgraded from the PF 2.5-GeV linac. Fifty-eight 50-MW klystrons were used as the rf source. In 1982, a 30-MW klystron was developed based on the design of the SLAC XK-5 klystron w2x by MELCO ŽMitsubishi Electric.. The XK-5, which has an output of 35-MW peak power at a 270-kV applied voltage, had been used in the SLAC 2-mile accelerator. Fig. 1 shows the klystron cathode. In the case of an S-band klystron ŽPV3030A2., 310 kV of a pulse voltage is applied between the beam-focusing electrode ŽBFE. and the anode. The nearest insulation length between them is only 18 mm, and the maximum field gradient reaches 249 kVrcm on the BFE and 235 kVrcm on the anode. The typical parameters associated with the gun conditions for the XK-5 and the 50-MW klystron w3x are listed in Table 1. The development history of the S-band klystron for these 17 years is as follows: the first 30-MW model ŽPV3030. was developed based on the XK-5, except for the shape of the BFE. An oxide-coated cathode with a diameter of 80 mm was used. During this period, we experienced on arcing troubles in the klystron gun. In order to solve these problems, we changed the shape of the BFE and examined the low-gradient gun. At the same time, we began to improve the manufacturing pro-
85
Table 1 Gun parameter of the XK-5 and 50-MW klystron Item
Unit
XK-5 Klystron
50-MW Klystron
Applied voltage Cathode diameter Cathode type
kV mm
Maximum field on BFE Maximum field on anode
kVrmm kVrmm
265 80 Oxide cathode 24.4 28.4
310 85 BI cathode 22.0 21.4
cess. In 1987, we replaced the oxide-coated cathode with a dispenser cathode Ža barium impregnated cathode.. This BI cathode model was called PV3030A1. We then adopted a low-gradient gun design for the dispenser cathode gun. This model was named PV3030A2. The PV3030A3 was introduced by replacing an insulating high-voltage ceramic seal with a larger one in order to increase the applied voltage to 310 kV. Finally, taking account of the cathode lifetime, we redesigned the gun and reformed the PV3050. This tube adopted a 85-mmdiameter dispenser cathode with a lower gradient BFE and an anode. The insulation ceramic was also replaced by a larger one; it has a capability up to a 350 kV applied pulse voltage w3x. We summarize these developments in Fig. 2.
3. Improvement of the manufacturing process Since 1985, we have improved the manufacturing process in order to solve arcing problems. A suitable
Fig. 1. Electron gun of the high-power klystron.
Fig. 2. Development history of the high-power S-band klystron.
86
S. Fukuda et al.r Applied Surface Science 146 (1999) 84–88
choice of materials was very important w4x; therefore, we introduced a vacuum-melted SUS-316L stainless steel as the BFE material and class-1 oxygen-free copper, which is also melted in a vacuum, as an inside tube material. The latter satisfies the ASTM F-68: the world’s highest standard for copper purity. For the vacuum-firing process of parts, the baking temperature of the BFE was lowered from 1000 to 6008C so as to avoid any surface roughness due to the growth of a grain boundary. It was pointed out that the impurity was concentrated on the grain boundary, which resulted in a rough surface; this was not desirable for high-voltage applications. At the same time, the BFE was polished using a paste containing small diamond particles Žparticle size of about 1–6 mm. in order to make a mirror-like surface finish. Since the anode was made of copper, the surface smoothness was achieved mechanically. For the process of pre-firing the cathode and its assembly, a new pre-firing furnace was made, and the baking temperature was carefully controlled. The cathode and its assembly were separated by a molybdenum cylinder in order to avoid any contamination of the assembly surface when the dispenser cathode was fired in order to evaporate any excess barium in the BI cathode. The temperature of a cathode post, made of stainless steel, was kept at 6008C by induction heating, similar to the case of parts baking. The cathode was carefully processed with setting a suitable filament voltage while monitoring the cathode temperature using a radiation thermometer. For the entire tube-baking process, we improved several points concerning the baking and evacuating furnace: the roughing pumping system, a manifold-baking procedure, and an investigation of residual gas using a mass analyzer. The baking temperature was raised from 480 to 5508C, and total baking hours were increased from 110 to 130 h. Due to these improvements, the final pressure in the tube reached 10y6 Pa, which was observed at the pumping head position. The forming gas Ž92% nitrogen gas and 8% hydrogen gas. was first filled in the furnace. However, it was found that hydrogen gas penetrated the tube through its copper wall; this rate increased rapidly as the baking temperature rose. Generally, the pumping speed of hydrogen is not very high for a diffusion pump or a turbomolecular pump. Although it is said that hydrogen gas was not harmful to the
cathode, any kind of gas is possibly harmful concerning arcing in the gun region. We thus changed the forming gas furnace to a vacuum furnace. Although these improvements resulted in a very low final pressure in the tube, they were finally found not to be effective for solving arcing phenomena in the tube with the oxide cathode. We thus replaced the oxide-coated cathode with a dispenser cathode. We could obtain a final pressure of less than 10y7 Pa for the BI cathode tube with the improved processing. At the same time, we decreased the arcing phenomena.
4. Oxide cathode and dispenser cathode The oxide-coated cathode used in the 30-MW klystron was developed by MELCO as a mush cathode, which comprised a matrix of coarse nickel particles Ž40–100 mesh size. sintered to the underlying nickel base in order to increase the porous layer; also, the usual carbonate cathode coating was soaked into the pores. Since the conversion process was inevitable in the tube, it was important to reduce barium evaporation. In the MELCO mush cathode, the initial evaporation rate just after conversion was reported large; the authors have mentioned the reduced reaction of carbon, which came from the binder material w5x. Too much reducing agents led to a high evaporation rate, while a small amount of reducing agents was important, and was added in the nickel base. The reduced reaction due to the residual hydrogen gas, which came from the forming gas,
Fig. 3. Anode of the failed klystron with an oxide-coated cathode.
S. Fukuda et al.r Applied Surface Science 146 (1999) 84–88
87
Fig. 4. Scandate dispenser cathode of the 50-MW klystron.
seemed to play an important role for barium evaporation of the mush cathode, since the replacement of the forming gas baking furnace by the vacuum bak-
ing furnace resulted in heavier arcing and a short tube life, as shown in Section 5. Although more accurate examination may be necessary, it has not
Table 2 Tube running statistics from 1979 to 1996 Fiscal year
Cathode
79 80 81 82 83 84 85 86 87
Oxide Oxide Oxide Oxide Oxide Oxide Oxide Oxide Oxide Oxide total BI BI BI BI BI BI BI BI BI BI BI total
87 88 89 90 91 92 93 94 95 96
MTBF Žh.
Total No.
Failed No.
Failure life Žh.
Failure cause Arcing Window
Others
Living No.
Living life average Žh.
3902 9242 16,588 10,317 19,934 9950 13,409 4351 4342 11,187 34,632 29,911 23,505 19,780 14,780 69,072 – – – – 30,840
4 20 20 9 13 13ŽU 1. 12ŽU 1. 15 7 113 7 20 18 18 15 12 14 13 23 15 155
4 19 19 8 12 12 11 14 7 106 7 16 17 12 7 2 0 0 0 0 61
3902 9050 15,965 10,054 18,753 9905 13,409 3524 4342 10,783 34,632 27,948 23,505 17,637 6393 6295 – – – – 22,265
2 13 11 5 6 10 7 13 5 72 1 1 1 0 1 0 0 0 0 0 4
1 1 6 1 4 2 4 1 1 21 5 11 9 7 2 0 0 0 0 0 34
0 1 1 1 1 0 0 1 0 5 0 3 0 3 8 8 13 13 23 13 84
– 3657 11,277 2120 14,170 – – 11,568 – 8558 – 10,467 – 8571 7448 15,694 8484 6058 3989 1717 11,012
1 5 2 2 2 0 0 0 1 13 1 4 7 5 4 2 0 0 0 0 23
88
S. Fukuda et al.r Applied Surface Science 146 (1999) 84–88
yet been solved. Fig. 3 shows the anode of the failed klystron with an oxide cathode. Finally, we changed the cathode to a dispenser type. We tried to use two Ir-coated cathodes, and then adopted the scandate cathode of Spectra-mat, based on experience with the SLAC 5045 50-MW klystron. The vacuum characteristics were better than for the oxide cathode; the BI cathode was vacuumfired in a separate envelope on a separate pump station so it was not necessary to worry about contamination because any excess barium evaporation deteriorates the klystron tube inside. The BI cathode features greater strength against residual gas than an oxide-coated cathode w6x. The operating temperature was decreased to about 9508C due to recent new technology. Fig. 4 shows the scandate dispenser cathode of the 50-MW klystron.
5. Run data statistics and BI cathode performance Table 2 shows the run data of the last 17 years for KEK high-power klystrons. This table gives the run data contributed to the KEK-TRISTAN operation up to 1996. This was divided into two parts; the upper is the data of klystron with the oxide cathode; the lower is the data of that with the BI cathode. The main cause of failures for that above was arcing; it reached about 70% of total tubes. Also, tubes manufactured in fiscal years 1986 and 1987 were found to have short lifetimes of around 4000 h due to the baking furnace change from the forming-gas one to the vacuum furnace, as described. Living tubes with an oxide cathode were all disposed while upgrading the linac. The lower table shows the lifetime data of a klystron with a dispenser cathode. The main causes of failures changed from arcing to window failures. In spite of the fact that the average applied voltage to the cathode increased and the average output power increased, arcing decreased predominantly. MTBF of the BI cathode klystron is about three times higher than that of the oxide cathode klystron. However, the lifetime of the failed BI cathode klystron was only
two times higher than for others, since dead tubes were fewer and the contribution to lifetime came mainly from the initial failure of BI cathode tubes. Other causes of failure included the operation stop of the old tubes to replace them with 50-MW tubes. They could be used, and, in fact, were used in other facilities. The PV3030A2 was all replaced with new klystrons, such as the PV3030A3 and 50-MW. Taking into consideration these cases, the realistic lifetime is expected to be more than 30,000 h. Window problems will be the main failures in the future, since our windows were fed the biggest power in a single pill box window when we use 50-MW klystrons.
6. Summary The developments of the high-power klystrons, which used a large-diameter cathode, are described. We succeeded in developing a stable 50-MW klystron by improving the manufacturing process, replacing the oxide-coated cathode with a BI cathode and adopting a low-gradient gun design. A very low pressure in the klystron, even with heater power on, was realized by performing careful cathode processing. Serious arcing problems were solved by using a low-gradient gun with a BI cathode. From the data of lifetime during these 17 years, we concluded that the cause of failures due to arcing decreased from 70% to 3% after replacement using BI cathode.
References w1x A. Enomoto, Proc. of 1996 Int. Linac Conf., CERN, Geneva, Switzerland, 1996, p. 633. w 2 x G.T. Konrad, Proc. of 1984 Int. Linac Conf., SeeheimrDarmstadt, West Germany, 1984, p. 293. w3x S. Fukuda, S. Michizono, K. Nakao, Y. Saito, S. Anami, Nucl. Instrum. Meth. Phys. Res. A 368 Ž1996. 561. w4x W.H. Kohl, Materials and Techniques for Electron Tubes, Reinhold, New York, 1960. w5x T. Hata, J.Kai, M. Koitabashi, K. Sano, Mitsubishi Denki Gihou, Vol. 40-6, 1966, p. 1034 w6x R.O. Jenkins, Vacuum 19 Ž8. Ž1969. 353.