SLAC-PUB+543 December 1994 A HISTORY OF THYRATRON LIFETIMES (A) AT THE STANFORD LINEAR ACCELERATOR CENTER David B. Ficklin Jr. Stanford Linear Accelerator Center, Stanford Universit~ Stanford, CA 94309 Abstract The Stanford Linear Accelerator Center (SLAC) No experience was available to establish or reliably project has been in almost continuous operation since the mid- thyratron lifetimes. Even limited operation at 120 pulses dle 1960s, providing a remarkable opportunity to amass per second for a month did not establish any benchmarks. The prediction of the rate of consumption of thyratrons thyratron data. This paper reviews the history of this in the linac modulators was therefore based on conjec- thyratron usage, focusing primarily on data collected dur- ture. Modulator reliability in the Stanford Linear Collider ing the last ten years of accelerator operation. ~~LC) configuration is discussed by A. R. Donaldson et There have been two distinct operating conditions “SLAC Modulator Operation and Reliabiltiy in the during the history of operation at SLAC. Prior to 1985, S~C Era” in the1992 20th Power Modulator Symposium, the fundamental th ratron operating points were 46 kV IEEE Confrence Record CH318@7/92/000@ O152. anode voltage (Epy f , 4.2 kA peak current, 3.8 ps equiva- The linac commenced operation at 120 pulses per sec- lent square pulse (esp), with a maximum repetition rate ond in July 1990. In July, 11 thyratrons were changed of 360 pulses per second (pps). The accelerator was UP in the linac modulators, followed by 18 in August, 4 in graded during 1985, and the thyratron operating points September, 19 in October and 10 in November, when the are now 46 kV Epy, 6.3 kA, 5.4 ps esp, with a maximum run was completed. During this run cycle, the initial repetition rate of 120 pps. symptoms of premature thyratron failure due to internal The SLAC high-energy physics research program r~ problems with Grid 1 appeared. During the three years quires that each of the available modulator klystron units of SLC operation at 10, 30 or 60 pps, the average replace provide a stable microwave energy source. Within these ment rate was 4.5 tubes per month. From 1988 through constraints, this paper explores historical thyratron life 1991, running at 60 pps, the replacement rate increased “ times at SLAC, reviewing the available data to determine to 7.5 tubes per month. Running at 120 pulses per sec- how long thwe thyrat rons can be expected to operate be ond, the replacement rate is 14 tubes per month. These fore failure currently or recently used in the 243 acceler- rates include thyratron changes for any reasons. ator modulators. An extensive testing program was established in March 1991 to verify the status of all thyratrons removed INTRODUCTION from operation in the linac. Of all thyratrons undergoing The Stanford Linear Accelerator Center functioned failure verification testing, 10% are recertified for opera- for many years w originally designed. In early 1985, tion in the linac modulators. This program provides hard there was a major upgrade to accommodate the newly data that is shared with the vendor to improve the relia- developed 50 Megawatt pulsed S-Band Klystron. This r~ bility of their product. work of all linear accelerator (linac) modulators altered The nature of the operational cycle at SLAC includes -- the thyratron operating condition to an Epy of 46kV at periods of time when the power is turned off for installa- 5,600 amps into a 1:14 pulse transformer to drive the new tion work? repairs, and upgrades. During the initial start klystrons. up followlng an extended maintenance period, thyratron Prior to and during the conversion of the knac mod- replacements run above average. The actual number of ula~ors, concern was expressed about the existing ITT changes varies, without correlation to time turned off, or F143 and Wagner CH1 191 thyratrons being capable of prior operating conditions. During a recent holiday down operating at the higher required power. Therefore, 48 of time, the thyrat ron filament, reservoir, and dc keepahve the 243 modulators used in six sections of the accelerator circuits were left on in an attempt to reduce replacements. were modified to use two thyratrons each. After addi- This decreased the number of replacements, but required tional testing, the existing Wagner CHl191s and the ITT operating 243 power supplies at 121 kilowatts. At three F143/CHl 191 rebuilds were found to be fully capable of weeks, cost of power is cheaper; at five weeks, power costs operating the new modulator configuration, so all of the are greater than savings. dual thyratron modulators were eventually reconfigured Of the 440 thyratron replacements in the linac that back to single thyratron operation. occured during the three years since September of 1991, During the period of the linac upgrade, it was found the earlier internal problem with the Grid 1 failure was a that the new 5405 klystron could operate at 350kV, with significant cause. Originally appearing in the F241s, this 60 megawatts of power out. This required modification of type of failure mode also appeared in other models, and all 1:14 pulse transformers to 1:15, increasing the peak pri- will be discussed as part of the data presented for each mary current through the thyratron from 5,600 to 6,100 thyratron model. The large number of changes has signif- amps. With this change, we could continue to use the di- icantly altered lifetime profiles. minishing thyratron inventory while actively engaged in We look at current lifetime profiles and age profiles, a search for new thyratrons that would operate the new grouped by specific thyratron model. This data is dy- namic. Here, it is presented in a static format. All men- 5045 klystrons. tion of hours refers to the thyratron high-voltage running Initial checkout of the modulator systems was accom- time recorded by electromechanical meters on the front plished at 10 pps. Most of the accelerator commission- of each modulator. The age and lifetime profiles are ex- ing work done through 1990 was accomplished at 60 pps tracted from the thyratron data base and may be viewed or less. During part of the conversion process, the ac- on computer. These graphs are based on run-time meter celerator continued to operate particle beam experiments readings taken approximately at the end of each month. with the pulse rate limited to 10, 30 or even 60 pps. Data current as of May 31, 1994, was used to generate the hours history plots shown for the thyratrons models * Work supported by Department of Energy contract presented. Filament run-time data is also recorded, but DGAC03-76SFO05 15. is not included in this discussion. Presented at the 1994 21st International Power Symposium Advanced Program West in South Coast Plaza Hotel in Costa Mesa, California, June 27–30, 1994 Litton Model 4888 Thyratron Ltietime Profile 1002 as of 5/31 /94 The thyratron high voltage run time hours for Lit- ton’s model 4888 thyratrons are in the range of one to six 25 ~ Number ofHT;yratrons: 145 thousand hours, from a sample statistically small com- 20 Mean : 7606 pared to other vendors. Design work, and prototyping produced the first tube in June 1993, with production i i 15} deliveries started August of 1993. Litton is still in the i post development stage of production, correcting of minor problems, improving and opt imization testing procedures. 10 The first thyratron was installed into the Hnac in June 1993. Today there are 16 in use. During a maximum 8750 5 hours of high-voltage running time, the earlier problem with Grid 1 failure has been seen in only one thyratron 0 to date. Due to the very short period of time available to 0 10 20 30 40 50 60 70 collect meaningful lifetime data, it is too soon to project Thousand of Hours ~A15 an accurate lifetime. EM OmniWave Model 1002 Figure 2. The quantity of tubes verus the high-voltage run- ning time hours at the time of removal from the finac mod- Production or rebuilding of the Wagner carcasses udtor. There were 19 failures with 1= than 500 hours of started in late 1984 with the first deliveries beginning in high-voltage running time. January 1985, and ceased nearly 6 1/2 years later after approximately 250 tubes had been rebuilt. The Thyra- Fifteen of the CX1836AS currently active in the linac. tron Age Profile graph (Figure 1) plots quantity versus A problem manifested as external arcing from Grid 1 to age for the 10% of the total model 1002 thyratrons which ground reduced the number of CX1836A thyratrons ac- are still in active service. Most of the 15 thyratrons cen- tive in the linac. A fix was found by working closely with tered in the bell of the distribution curve at 20,000 hours English Electric Valve, and all of the removed CX1836AS were produced late in the production. The two oldest have been modified and recertified for use. The CX1836A thyratrons have close to 50,000 hours. is the only thyratron model that has not suffered from the Grid 1 problems prevalent with all other models used in Thyratron Age Profile 1002asof5/31194 the linac modulators. 25 ~ English Electric Valve Ltd. Model CX241O Number of Thyratrons: 26 The CX241O is the result of a SLAC request for a six- 20 Minimum HV: 12102 nch diameter thyratron with an oxide cathode. Deliveries Mean HV: 21634 started July 1992 and ran through July 1993. Outgrowth Maximum HV: 49728 15 of this particular design is the CX2412, which supports — - t 1 and brings Grid 1 through the ceramic in the same fwh- E Active ion as Grid 2.