Wire Survival Test of Crowbar Less, High Voltage DC, Klystron Bias Power Supply

Open Science Journal of Electrical and Electronic Engineering 2017; 4(1): 1-9 http://www.openscienceonline.com/journal/j3e Wire Survival Test of Crowbar Less, High Voltage DC, Klystron Bias Power Supply Akhilesh Tripathi *, Manmath Kumar Badapanda, Rinki Upadhyay, Mahendra Lad Department of Atomic Energy, Raja Ramanna Centre for Advanced Technology, Indore, India Email address atripathi@rrcat.gov.in (A. Tripathi) * Corresponding author To cite this article Akhilesh Tripathi, Manmath Kumar Badapanda, Rinki Upadhyay, Mahendra Lad. Wire Survival Test of Crowbar Less, High Voltage DC, Klystron Bias Power Supply. Open Science Journal of Electrical and Electronic Engineering. Vol. 4, No. 1, 2017, pp. 1-9. Received: May 31, 2017; Accepted: August 3, 2017; Published: September 25, 2017 Abstract A solid state modular -100 kv, 25 A crowbar less DC power supply is utilized for biasing the Thales make, 1 MW, 352.2 MHz, TH 2089 klystron amplifier. The klystron amplifier is susceptible to high voltage arcing because of the presence of strong electric field between its cathode and body and energy dumped during such arcing should be less than arc energy handling capacity of this klystron amplifier, otherwise irreparable damage may occur. The developed -100 kv, 25 A DC power supply has very little capacitance (10 nf) at its output due to which its stored energy is significantly reduced. Wire survival test is carried out on this power supply to ensure that stored energy contained in it is less than 20 Joule so that it can be used to bias TH 2089 klystron amplifier without requiring any crowbar. The fusing action of the power supply is measured to be 1.46 A 2 Sec which is less than fusing action tolerated by TH 2089 klystron amplifier. The power supply is capable of switching off its output within 2 µs in the event of arcing in the klystron amplifier. The paper describes the conduction of wire survival test along with the method of selection of proper test wire that can be used for this test. Various results obtained from these wire survival tests are presented in this paper. Keywords Klystron Amplifier, Wire Survival Test, Beam Power Supply, Stored Energy, Fusing Action 1. Introduction A high power RF test stand employing Thales make1 MW, 352.2 MHz, TH 2089 klystron amplifier as RF source, is developed at Raja Ramanna Centre for Advanced Technology (RRCAT), Indore for testing various RF components, cavities and accelerating structures. Klystron amplifier requires that its DC beam bias power supply must have low stored energy and low ripple in the output simultaneously. Both these requirements are contradictory in nature as low output ripple can only be met by higher filtering elements which further results in increased stored energy. Low stored energy requirement comes from the fact that higher stored energy may result in the flow of very high current inside the internal structure of klystron tube in the event of arcing inside the tube and may result in the complete failure of the klystron amplifier. Such higher current may lead to high energy dissipation in the klystron cathode region which can destroy internal elements or the vacuum seal and causes serious damage to the klystron amplifier. A conventional power supply requires larger filtering components and energy stored in them is significantly higher than what can be tolerated by klystron amplifier. Hence it uses a low impedance crowbar circuit to divert the stored energy through it in case of arcing and quenches the arc in less than few µs [1]. However, the crowbar circuits are costly and they have false-firing problems caused by electrical noise and hence have reliability issue. The life of crowbar is also short and they need quick replacement. To avoid crowbar completely, the filtering requirement of the bias power supply has to be reduced significantly and this can be achieved only by very high output ripple frequency of the bias power supply as f 1/C. Hence a novel crowbar less, solid state modular scheme is adopted to develop -100 kv, 25 A DC power supply as the beam bias power supply of TH

2 Akhilesh Tripathi et al.: Wire Survival Test of Crowbar Less, High Voltage DC, Klystron Bias Power Supply 2089 klystron amplifier which simultaneously meets the requirement of low ripple as well as low stored energy in its output. The switch power modules used in this power supply are suitably staggered and connected in series to provide -100 kv DC output with very high effective ripple frequency which reduces output filter capacitor requirement substantially. The power supply contains 176 numbers of AC- DC converter based switch power modules which are connected in series to provide -100 kv output and hence the ripple frequency at the output is 176 times the module operating frequency [2]. Suitable R-L-C based filter has been incorporated at the output of this power supply to keep the output voltage ripple below 0.5%. This filter contains both inductance (L, stored energy ½ LI 2 ) and capacitance (C, stored energy ½ CV 2 ). During arcing of klystron amplifier, its arc resistance reduces considerably (~ 0.1 Ω) and hence the discharge of the inductor stored energy becomes slower (with time constant L/R), but capacitive stored energy discharges very quickly (with time constant RC), thereby the latter becomes crucial for the protection of klystron amplifiers. Some klystron manufacturers keep restriction on allowable fusing action of the arc current where fusing action is defined as the time integral of the square of the current (I 2 t rating), while others keep restriction on the maximum allowable energy that can be allowed to pass through these arcing klystrons. The TH 2089 klystron amplifier employed in this test stand, can tolerate about 20 Joules of energies called critical energy under arcing and 40 A 2 sec of fusing action called critical fusing action under arcing beyond which irreparable damage may occur [1, 3, 4]. Wire survival test is conducted on -100 kv, 25 A DC power supply power supply to ensure that its stored energy is less than 20 Joule. The fusing action of this power supply is also measured in this paper which is hardly reported in literature. The fusing action of a test wire is a physical quantity which depends upon the area of the test wire but is independent of the length of the test wire. In conventional power supply, limiting elements are needed to limit the short circuit current and hence switch off time is less significant in these power supplies. But in crowbar less power supply, there are no limiting elements and hence it needs to be switched OFF immediately in case of arcing in the load. The output sensing circuit employed in this power supply to keep the HV switch OFF time around 2 µs is also presented in this paper. Section 2 briefly describes the phenomenon that leads to arcing in side klystron amplifier. Section 3 gives details of wire survival test conducted on -100 kv, 25 A DC power supply to qualify it for biasing this klystron amplifier. It also describes the selection of test wire for conduction of wire survival test. Section 4 focuses on the output monitoring scheme employed in this power supply to keep least possible switch OFF time. Section 5 presents the analysis of the results drawn from these tests. It calculates the fusing action of the power supply. 2. Arcing Inside Klystron Amplifier The gun region of klystron amplifier includes the cathode made of porous tungsten impregnated with an emissive material, a heater and a beam forming electrode [3]. The assembly is mounted in the tube itself by ceramic-metal seals, providing high voltage insulation. The beam forming electrode (at cathode potential) is intentionally operating at a much lower temperature than the cathode to avoid it becoming an emitter. The shape of the klystron gun region is given in Figure 1. Klystron is a vacuum device and operates at very high vacuum condition in the order of 10-9 Torr. The gun material contains trace amounts of gases that may come out of the material when high voltage is applied between cathode and collector. When the electrons which are continuously emitted from cathode, collide with these gas molecules there is electron multiplication due to impact ionization which may lead to arcing. This degassing process is mostly responsible for arcing in the gun region. Other cause of arcing in the gun region is creation of electron avalanche due to multipactoring. This multipactoring aspect is taken care by the klystron manufacturer but it may still happen in the various operating condition whenever the travel time of electrons between any two electrodes inside the gun structure is integer multiple of half of RF period (T s /2) and electron emission factor is greater than or equal to 1. In the klystron window region, arcing is caused by increased reflected power (poor VSWR). When reflected power is excessive, it causes back scattering of electron and creates arcing like situation. Figure 1. Cut-away view of electron gun in TH 2089 klystron amplifier. 3. Wire Survival Test Wire survival test is conducted to ensure the suitability of high voltage bias power supplies for feeding to sensitive klystron amplifiers [5, 6, 7, 8, 9]. In this test, a wire of suitable material and dimension is selected and the klystron arcing condition is intentionally created to get an idea about the amount of fault energy dumped into the klystron amplifier under arcing condition from its bias power supply. The survivability of wire during wire survival test ensures that the stored energy in the bias power supply is not sufficient to cause any damage to the arcing klystron amplifier under actual operating condition.

Open Science Journal of Electrical and Electronic Engineering 2017; 4(1): 1-9 3 Figure 2. Scheme of -100 kv, 25 A DC power supply. Klystron manufacturer also suggest that fusing action of the arc current (I 2 t in A 2 sec) should be less than 40 A 2 sec along with stored energy criterion. The stored energy of-100 kv, 25 A DC power supply along with its I 2 t rating is measured in this paper to ensure that this power supply is safe in all respect to be used as bias power supply for TH 2089 klystron amplifier. The scheme of the power supply along with its output filter is shown in Figure 2. 3.1. Selection of Test Wire A copper wire of appropriate dimension is employed for conducting wire survival test, due to its easy availability in various dimensions [10]. The dimensions of wire are chosen keeping two factors in view. First, to keep the energy restriction, a wire of appropriate volume need to be specified, as the energy to fuse a wire depends only on it s volume. Second, to keep the I 2 t fusing action restriction, a wire of appropriate cross section need to be specified, as the fusing action to fuse a wire depends only on it s area and is independent of it s length. Again the minimum length of the copper wire should be chosen appropriately to avoid the corona discharge in the air. Considering the electric field required for corona discharge in air to be 24 kv /cm, the length of the copper wire should be greater than 4.16 cm to be used with -100 kv power supply. The heat required to raise the temperature of any wire is equal to the change in internal energy if the loss of energy to the surroundings is zero. Therefore, dq = du = mc( T) dt (1) where dq is the incremental change in heat energy, du is the incremental change in internal energy, m is the mass of the wire, C(T) is the specific heat capacity and dt is the incremental change in temperature. Here it is assumed that the heating of the wire occurs in a short enough amount of time and the system is adiabatic. Integrating and substituting ρ la for m, where ρ is the density, A is the cross-sectional area, and l is the length, equation 1 becomes: U = ρ la Tf C( T ) dt (2) where T o is the ambient temperature and T f is the melting temperature of the wire. The specific heat capacity can be approximated with a power series and U is the change in internal energy of the wire. Applying the first law of thermodynamics to the system, To de = dq dw (3) Here de, dq, and dw are the incremental change in energy, heat transfer from the wire and work done on the wire respectively. Under the assumption that the system is adiabatic, equation 3 reduces to: de = dw (4) where -dw is equal to Pdt (where P is the power dissipated in the wire, i.e. i 2 (t)r(t) or v(t)i(t)). As there is no change in kinetic or potential energy of the wire, so de= du. Hence from equations 1, 2 and 4, it is found that: Pdt. = ρlac( T) dt (5)

4 Akhilesh Tripathi et al.: Wire Survival Test of Crowbar Less, High Voltage DC, Klystron Bias Power Supply On integrating both the sides Tf Tf Pdt = ρ la C( T) dt (6) To Therefore to calibrate a 20 Joule copper wire, Tf To ρ la C( T) dt = 20Joule (7) To For a copper wire, ρ= 8900 kg/m 3, C= 385 J / (Kg K), T f = T Melting = 1083 C and T o = T ambient = 25 C, the volume of copper wire is calculated as la = 5.5168 mm 3. A graph showing the length vs wire gauge (function of the wire area) is given in Figure 3. From this curves it is possible to choose a wire gauge and length needed for wire survival test. A copper wire of 0.18 mm diameter (AWG 33) is selected for this wire survival test, its area comes out to be 0.0254 mm 2 and from above equation, length is calculated as 217 mm. 3.2. Wire Survival Test Conduction The wire survival test is carried out by instantly connecting the wire under test to the terminals of high voltage power supply set to deliver the maximum output voltage. As soon as the power supply is switched ON, its output current increases as both the terminals of power supply are shorted through test wire and when this current crosses the set over current limit, the power supply trips. Survivability of copper wire ensures that stored energy in the power supply is less than energy rating of the copper wire and fusing action of the power supply is calculated from the observed short circuit current graphs during wire survival test. In the laboratory, this test is conducted by shorting two cables which are going from HVPS -100 kv terminal and 0 kv terminal to klystron amplifier as shown in Figure 4. The high voltage cables used are from Dielectric Sciences, USA and total length of cable including the high voltage terminal of HVPS to klystron and klystron to HVPS ground terminal is 25 meter. The shorted cables are put in series with the test wire as shown in Figure 4. Various lengths of AWG 33 copper wires are used for this test. Power supply output over current limit is set to 25 A. This test is conducted by operating a series relay switch which allows the power supply short circuit current to flow through the specified copper wire. This fault current is sensed by the protection system of the power supply and the power supply tripped immediately due to its over current setting on operation of series relay. Simultaneously input power is also isolated from the power supply. Initially the wire length is selected to be 21.7 cm. In this case the energy of the copper wire is calculated to be 20 Joule. Figure 3. Dimension of test wire for wire survival test.

Open Science Journal of Electrical and Electronic Engineering 2017; 4(1): 1-9 5 The 21.7 cm copper wire has survived wire survival test which ensures that the fault energy of this power supply is less than 20 Joule. The waveform showing the output voltage and current during this test is presented in Figure 5. Again, the wire survival test is conducted with 10 cm long copper wire corresponding to energy of 9.2 Joule. This 10 cm long wire has also survived wire survival test and that ensures that fault energy of this power supply is less than 9.2 Joule. The wave from associated with this test is presented in Figure 6. Figure 4. Wire survival test setup with HV cable from HVPS end to klystron amplifier. Figure 5. Wire survival test with 21.7 cm wire.

6 Akhilesh Tripathi et al.: Wire Survival Test of Crowbar Less, High Voltage DC, Klystron Bias Power Supply Figure 6. Wire survival test with 10 cm wire. 4. Switch off Time The switch OFF time of the power supply should be as low as possible so as to cut off the high voltage in case of arcing in the load of the power supply (TH 2089 klystron amplifier) to minimize the impact of arcing [5]. The reduction of switch OFF time up to 2 µs is achieved in this power supply by quick sensing of the output signal (voltage and current) through analog monitoring system of the power supply which is presented in Figure 7. The analog monitoring system is used to transmit measurement data from a high voltage potential to earth potential via fiber optical links. The reference value for the trip level is communicated to the HV side through an optical fibre channel from the ground based control and monitoring unit. The digital receiver re-establishes the analogue trip level with the help of an DAC at the HV side. In case the value sensed by analog monitoring system exceeds the threshold value set by control system, the analog comparator output goes high and it is transmitted to the ground referred control system of the power supply via optical fiber. The control system immediately removes the ON command of module IGBT and trips 11 kv input circuit breaker. HV switch OFF time is the time delay occurred in removal of ON command of module IGBTs from the moment of arcing and it is shown in Figure 8. During wire survival test, high voltage switch off time of the power supply is measured to be 2 µs that implies that within 2 µs of occurrence of arc fault signal, IGBT ON commands are removed. Figure 7. Analog monitoring system of -100 kv, 25 A DC power supply.

Open Science Journal of Electrical and Electronic Engineering 2017; 4(1): 1-9 7 Figure 8. Switch OFF time of -100 kv, 25 A DC power supply. 5. Analysis and Result From the above tests, it is confirmed that the stored energy of this power supply is less than 9.2 Joule which meets20 Joule criterion specified by klystron manufacturer. Now the fusing action of this power supply is measured. Fusing action of the power supply is a measure of the thermal energy delivered to each ohm of the klystron circuit by the shortcircuit current during the time t. Figure 9 is redrawn from Figure 2 to explain the short circuit phenomenon of the tested wire. In this circuit L is the equivalent short circuit inductance which mainly contains inductances of the connecting cables and C is the output capacitance of output filter. Figure 9. Equivalent circuit of short circuit across output capacitor during wire survival test. The cable inductance is measured from the formula of the DC inductance of straight single wire which is: L = 2. l µ r.[ln(2 l/ a) 0.75] nh (8) Where l is the length of wire in cm, a is the radius of wire in cm and µ r is the relative permeability of copper conductor which is 0.999994. Since the cable length used is 25 meter and cable diameter is 1.5 inches hence the self inductance of the cable is calculated as 35.6 µh. The cable capacitance is too low (in pf) and hence it is ignored. Now from Figure 9, if C is taken as 10 nf, R is taken as 20 Ω and L is taken as 19.8 µh, in that case it is clear that L > ¼ R 2 C (as ¼ R 2 C = 0.55 x 10-6 ) which leads to damped sinusoidal oscillations of short circuited current. The period of this damped wave is found as T= 1/f= 2π (LC) = 2.78 µs. Waveform shown in Figure 5 and Figure 6 is a clipped sinusoidal waveform with period of 2.8 µs which matches closely with calculated period. The current in these waveforms is clipped and is shown only up to 50 A. It is clipped due to the limited output range of the DAC (digital to analog converter) used in the output monitoring of the main control system. The actual peak of short circuit current can be calculated from this clipped waveform. For I 2 t measurement, the short circuit current and voltage waveforms are taken for much longer duration as shown in Figure 10. In this figure, scale for voltage is 1 V 12 kv and scale for current is 20 V 20 A. So the peak value of current is 40 x 3 = 120 A and time duration taken for this current to become zero is 203.18 µs. The calculation of I 2 t value is done by integrating the area with the approximation of a triangle as shown in Figure 10: I 2 t = (120 A) 2 x 203.18 µs/ 2 = 1.46 A 2 sec.

8 Akhilesh Tripathi et al.: Wire Survival Test of Crowbar Less, High Voltage DC, Klystron Bias Power Supply Table 1. Summary of wire survival tests. Wire Length(Size:AWG 33) Wire Energy Wire Survival Fusion Energy HV switchoff time 21.7 cm 20 J Yes < 1.46 A 2 sec 2 µs 10 cm 9.2 J Yes < 1.46 A 2 sec 2 µs Figure 10. Current and voltage waveforms for I 2 t calculation during wire survival test. Since the area covered by the current graph is much smaller than the area of the triangle shown in above graph, the actual I 2 t rating of the power supply will also be smaller than the one measured in this paper. The summary of all these wire survival tests is presented in Table 1. 6. Conclusion Wire survival tests were carried out to ensure the suitability of -100 kv, 25 A DC high voltage bias power supply for feeding to sensitive klystron amplifier TH 2089. The wire survival test is a stringent test and survivability of wire ensures complete protection of klystron amplifier from arcing under actual operating condition. This test method is based on equating the klystron arc energy to the wire fusing energy. Wire survival test of this power supply is done with 10 cm long AWG 33 copper wire which has survived during this test insuring that the fault stored energy in this power supply is less than 9.2 Joule. The waveform of the short circuit current during wire survival test is damped sinusoidal and its time period is measured as 2.8 µs. This time period of the sinusoid wave is calculated and found to be 2.78 µs which is in close match with experimental result. The critical fusing action under arcing of the power supply is found to be less than 1.46 A 2 sec. The switch off time of high voltage is measured as 2 µs hence this -100 kv, 25 A DC power supply is suitable to power the cathode of 1 MW, 352.2 MHz, TH 2089 klystron amplifier. References [1] Bora D. et al. High power continuous wave microwave system at 3.7 GHz. Review of Scientific Instruments volume, 2001, 72(3):1869-1875 [2] Manmath Kumar Badapanda, Rinki Upadhyay, Akhilesh Tripathi, Rajeev Tyagi and Mahendra Lad. AC-DC converter power modules of a solid state modular high voltage DC power supply. IEEE International Conference on Electrical Power and Energy Systems (ICEPES), 2016, pp. 100-104. [3] TH 2089 data sheet UTH2089. [4] Cassel R. and Nguyen M. N. A unique power supply for the PEPII klystron at SLAC. IEEE Proceedings of the Particle Accelerator Conference, 1997, 3: 3482-3484. [5] Cortazar O. D., Ganuza D., Fuente J. M. De La, Zulaika M., Perez A., and Anderson D. E. A 100 kv, 60 A solid state 4 khz switching modulator for high power klystron driving. Review of Scientific Instruments, 2013, 84: 054706(1-6). [6] Srinivas Y. S. S. et al. Results of 10-Joule wire-burn test performed on 70 kv rail-gap crowbar protection system for high power klystrons and gyrotron. IEEE Symposium on Fusion Engineering, 2002, pp. 91-94. [7] Srinivas Y. S. S., Babu Rajan, Makwana Azad, Parmar Kirit, Kulkarn S. V. Development of 70kV, 22 A DC Power Supply for High Power RF and Microwave Tubes. Journal of Physics: Conference Series, 2010, 208(1): 012031.

Open Science Journal of Electrical and Electronic Engineering 2017; 4(1): 1-9 9 [8] Patel P. J., Singh N. P., Tripathi V., Gupta L. N. and Baruah U. K. A regulated power supply for accelerator driven system, Proc. IAEA, 2009, pp.1-8. [9] Srinivas Yellamraju Sham Sunder, Sanjay V. Kulkarni. Performance and modeling of 70kVdc power supply with solid-state crowbar. Fusion Engineering and Design,2013, 88: 868 871. [10] Parmar K. M., Srinivas Y. S. S., Kadia B. R., Kulkarni S. V. and ICRH-RF Group. Wire Burn Test on 35 kv, 20 A, High Voltage Power Supply for 91.2 MHz, 200 kw Stage RF Generator. 24 th National Symposium on Plasma Science and Technology, PLASMA, 2009. Bibliography Akhilesh Tripathi received his Bachelor of Technology degree in Electronics and Communication Engineering from Uttar Pradesh Technical University, Lucknow in 2007. He was awarded gold medal by his college for standing first in Electronics and Communication Engineering in 2007. He joined RRCAT in 2008 as Scientific Officer after graduating from 8 th batch of BARC Training School, Indore. He also received Master of Technology degree from Homi Bhabha National Institute (HBNI), Mumbai in 2010. He has expertise in the field of high voltage power supplies, RF power amplifiers and related control-interlock system. He has published several papers in various conferences and journals. Rinki Upadhyay received her Bachelor of Engineering degree in Electrical Engineering from University of Rajasthan (renamed as Rajasthan Technical University), Jaipur in 2006. She was awarded gold medal by her college for standing first in Electrical Engineering in 2006. She joined RRCAT in 2007 as Scientific Officer after graduating from 7 th batch of BARC Training School, Indore. She also received Master of Technology degree from HBNI, Mumbai in 2010. She has expertise in the field of high voltage power supplies, RF power amplifiers and related control-interlock system. She has published several papers in various conferences and journals. Manmath Kumar Badapanda received his Bachelor of Engineering degree in Electrical Engineering from University College of Engineering (renamed as Veer Surendra Sai University of Technology), Burla, Odisha in 1988 and Master of Technology degree from Indian Institute of Technology, BHU, Varanasi in 1990. He joined RRCAT, Indore in 1993 and presently he is Head, RF Power Supplies Laboratory, RF Systems Division at RRCAT. He is mainly looking after design and development of precession regulated HVDC power supplies and related control, interlock and protection aspects for various sensitive RF amplifiers employed with high energy particle accelerators developed in this centre. He has also received Post Graduate Diploma in Business Administration from IIMS, Kolkata in 1999. He has published more than 45 numbers of technical papers in various conferences and journals. Mahendra Lad received his Bachelor of Engineering degree in Electronics and Telecommunication Engineering from Devi Ahilya University, Indore in 1986. He joined RRCAT, Indore in 1987 as Scientific Officer after graduating from 30 th batch of BARC Training School, Mumbai. Since then he is involved in RF System development for various Synchrotron Radiation Sources in this centre. His research interests include development of low level RF system for particle accelerators. Presently he is Head, RF Systems Division at RRCAT. He has published several papers in various conferences and journals.