DEVELOPMENT OF WIDEBAND AMPLIFIER IN ITER ICRF RANGE

Transcription

1 AKHIL JHA et al. DEVELOPMENT OF WIDEBAND AMPLIFIER IN ITER ICRF RANGE Akhil Jha ITER-India, Institute for Plasma Research Bhat, Gandhinagar, India Ajesh P, JVS Harikrishna, Hriday N Patel, Manoj Patel, Rohit anand, Hrushikesh Dalicha, Paresh Vasava, Rajesh Trivedi, Aparajita Mukherjee ITER-India, Institute for Plasma Research Bhat, Gandhinagar, India Abstract High Power Amplifier 1 (HPA1) is a pre driver amplifier which will drive each RF chain of ITER ICRF source. A development work is ongoing for a tetrode tube based wide band HPA1. Aim is to achieve a -1dB bandwidth at 2MHz over any central frequency in range 36-6MHz. To achieve this specification the design of output cavity is based on wideband impedance matching circuit. Two L-C circuits connected in series are tuned to achieve a wideband response over desired frequency band. Input circuit design is based on tunable wide band impedance transformer. The amplifier is designed to operate in Grounded Grid configuration with CPI 4CW25B tube. The rated output power is 1kW/CW. A detailed calculation is performed to find operating parameters of tetrode tube at rated power. A LabVIEW based code Tetrode Tube Calculator (TTC) is developed to perform load line calculation for tetrode tube. The code requires as input, parameters like DC anode bias, input DC power, anode voltage swing etc. and calculates the parameters like output power, efficiency, output impedance, input impedance etc. The calculated parameters are used as input for cavity design. This paper discusses the tube parametric calculation using TTC code in detail. The detailed design of HPA1 cavity using CST Microwave studio software is discussed. 1. INTRODUCTION ITER-India is responsible for delivery of 8+1(prototype) RF sources to ITER project [1]. Each RF source will provide 2.5MW of RF power at VSWR 2:1 in the frequency range of 36 to 6MHz [2]. Eight such RF sources will generate total 2MW of RF power. Two RF chains containing three high power amplifiers (HPA1, HPA2 and HPA3) need to be combined to build an RF source. HPA2 and HPA3 are RF tube based amplifiers while HPA1 is a solid state power amplifier. The HPA1 (SSPA) has a maximum output power of 15kW at any frequency within frequency range of 35-65MHz. The HPA1 is fed to the input of HPA2 which can generate output power up to 12kW which finally drives HPA3 to produce output power of 1.5MW. The SSPA is a solid state device (MOSFET) based amplifier which combines ~16 pallets using two stage combiner circuits. An internal control loop is used to keep the output power of SSPA constant over the entire frequency band. SSPA is designed to deliver constant output power up to VSWR 1.5:1. and tripping level is set at VSWR 2.:1.. During operation with RF chain in certain loading scenarios SSPA faces an output VSWR up to 2.5:1. which trips SSPA and leads to loss of RF power. The high reflected power during few shots led to gate current imbalance which led to damage of MOSFETs and balancing resistor in combiners. These types of damages are quit common for multi module solid state based amplifiers and require design of fast control systems to prevent any damage. Another option for rugged tetrode tube based amplifier is being explored in the given frequency range. Tetrode tube amplifiers in comparison with solid state device based amplifiers have only single active device (The tube itself). Operation limits are decided by maximum dissipation limit of different electrodes and voltage breakdown limit of cavity elements. Using sufficient margin in dissipation limits during tube selection (depending on required output power) and proper design of cavity element can provide a rugged amplifier for high VSWR (3.:1.) scenarios. Keeping this fact in mind tube based amplifier design is initiated. 2. TUBE SELECTION AND LOAD LINE According to present requirement, to achieve an output power of 1.5MW at VSWR 2:1 required maximum output power from HPA1 is 8kW. There are two tubes that can satisfy this requirement TH 595 by Thales Electron devices and EIMAC 4CW25A/4CW25B by CPI [3]. The later one is chosen for the present development and will be implemented in Grounded Grid (Cathode driven) configuration. Absolute maximum ratings [3] for the tube in above configuration are listed in Table1. The load line calculation is done with a 1

2 FIP/P8-16 targeted output power in range of 1kW to 15kW (max.) and according to that the maximum anode bias voltage/ current is chosen to be 6kV/4.5A. In class B operation, for an electrical efficiency of 6% above biasing condition provide an output power roughly equal to 16.5kW with anode dissipation of 1.8kW. The above data is within tube maximum rating of anode bias 6.5kV/5A and anode dissipation 15kW. The two operating points of load line are chosen to be 6kV/A (point 1) and 1kV/13.5A (point 2) for DC screen bias of.75kv. Further calculations were performed by graphical method known as Eimac Tube performance computer by CPI [4]. This method provides instantaneous values of the different electrode currents flowing at every 15 of the electrical cycle. These instantaneous current values are then used to calculate the resultant RF and DC current values. These current values together with the voltage operating point provide the value of RF output power, required drive power, various electrode dissipation, gain, efficiency, input and output tube impedance. Finally the calculated input and output tube impedance is used to design the input and output impedance matching circuit. The HPA1 needs to be operated in different operating conditions depending on the loading condition provided by the HPA2 input. To provide required constant output power during these different loading conditions HPA1 needs to be operated on different load lines. Each load line calculation required repetition of above calculation procedure which was quite tedious and time taking. For convenience above calculation was coded in MS excel and finally the logic was implemented to develop Tetrode Tube Calculator (TTC) code which is LabVIEW based. TABLE 1. ABSLOUTE MAXIMUM RATING DC Plate voltage 6.5kV DC Screen voltage 1.5kV DC Plate current 5.A Plate dissipation 15kW Screen dissipation 45W Grid dissipation 2W 3. TETRODE TUBE CALCULATOR (TTC) The TTC code can be divided in two major parts. First part is digitization and extraction of data points from the tube constant current characteristic curve. In the second part, the available data points and input parameters are used to calculate the instantaneous current values over the load line. Finally the instantaneous current values are used to calculate the various tube parameters. In the first part of TTC, the tube constant current characteristic curve (which is supplied with the tube datasheet) is imported in image format. This characteristic curve is calibrated by defining the range of x-axis and y-axis as shown in Fig. 1(a). Once calibrated, the data of each constant current curve is taken separately as shown in Fig. 1(b). This process generates a matrix of tube plate voltage vs. grid voltage for each constant current curve. This matrix can be exported in.csv or.text format. Once the data matrix corresponding to perticulr screen grid (SG) voltage is available, it can be scaled for any other SG volatge as per three half power law [5] in TTC itself. Calibration Importing Image Calibration Calibration of x - axis Defining axis range Defining constant current value Sampling of data points Calibration of y - axis 1(a) 1(b) FIG. 1. Screenshot of TTC showning (a) calibration of graph axis (b)data sampling and naming for each curve

3 AKHIL JHA et al. In the second part of TTC, input parameters like anode bias voltage, anode peak voltage, maximum anode current, maximum anode dissipation and class factor (defines the class of operation) are given to code as shown in Fig. 2(a). For example, for HPA1 above parameters are 6kV, 5kV, 4.5A, 15kW (Table 1.) and 3. (for Class B) respectively. In another tab, the SG voltage corresponding to data matrix is defined and sampled data corresponding to each electrode current (anode current, control grid (CG) current and SG current) is identified in two steps as shown in Fig. 2(b). In first step, for each electrode current row range is defined and in second step, each row is assigned with corresponding electrode current value. Finally this data with input parameters is processed to get instantaneous current points (not shown in figures) and output parameters like DC anode current, input anode DC power, peak fundamental RF current, output RF power, anode efficiency, input/output tube impedance, SG/CG current and anode/sg/cg dissipation (for both grounded grid and grounded cathode configuration) as shown in Fig. 2(a). I/p parameters Max anode dissipation DC Anode bias Peak anode volt Max anode current Class factor Tube Parameters O/p parameters (grounded cathode) Grounded grid DC anode current DC anode input power Peak Fund RF current O/p RF power Anode efficiency (%) Anode dissipation O/p tube impedance SG voltage Define row range Anode current range Input Parameters SG current range CG current range DC SG current SG dissipation DC CG current DC CG dissipation Anode CG SG Assign SG current value Assign CG current value Drive power I/p tube impedance Assign Anode current value Gain 2(a) 2(b) FIG. 2. Screenshot of TTC showning (a) input (I/p) and output ( O/p) parameters in tube parameter tab (b) defining of row range and assigning of electrode current values in input parameter tab 4. HPA1 TUBE PARAMETERS Tube parameters for HPA1 was calculated using TTC for grounded grid configuration at SG bias voltage 75V with three different operating points corresponding to anode bias voltage of 4kV, 5kV and 6kV as shown in Table2. Table 2. CALCULATED HPA1 TUBE PARAMETERS FOR GROUNDED GRID CONFIGURATION Operating point 1 Operating point 2 DC anode bias (V adc ), kv Idle anode current (A)... V apeak (kv) V apeakmin (kv) I p (A) DC anode current (A) Anode power input (DC), kw Peak fundamental RF current (A) Output power(p out ), kw Anode dissipation, kw Efficiency (%) Output tube impedance (ohm) DC screen current (A) Screen dissipation (W) DC CG current (A) CG dissipation (W) Drive power (P drivegg ), W Input tube impedance (ohm) Gain (P out /P drivegg ), db

4 FIP/P8-16 Calculated parameters show that, with anode bias voltage in range of 4 to 6kV the maximum RF output power achievable is 1 to 15kW with a drive power of ~ 1.2 kw and efficiency in range of 73 to 78%. The maximum anode bias voltage 6kV and DC anode current ~3.3A are well within the tube absolute maximum rating (Table1). All three electrode dissipations i.e. anode, SG and CG are approximately 4kW, 15W and 1W respectively which are well within maximum dissipation rating (Table1). Though the gain by the amplifier is on lower side it can be increased by shifting the operating point towards class AB with idle anode current up to 2A. This will increase gain up to 13.8dB, but on the cost of efficiency and increased anode dissipation which will be ~64% and ~8kW respectively for operating point 6kV. 5. HPA1 IMPEDANCE MATCHING CIRCUIT The tube parameter calculation provides input and output tube impedance in range of 67 ohm to 72 ohm and 52 ohm to 88 ohm respectively as shown in Table 2. These tube impedances are needed to be matched with nominal transmission line impedance of 5 ohm using an impedance matching circuit. This impedance matching circuit can be divided in two parts i.e. input and output impedance matching for input and output impedance of tube respectively. At the time of writing this paper design of input matching circuit is in progress so only output matching circuit is discussed in detail. The design of output impedance matching circuit of HPA1 can be divided in two parts. First part is design and analysis of lumped RF circuit and second part modeling and simulation of coaxial RF circuit Design and analysis of lumped RF circuit The bandwidth requirement for complete RF chain about any central frequency is 1dB at 1MHz which provides a minimum bandwidth requirement for HPA1 i.e. 1dB at 2MHz. A wideband circuit containing two L networks coupled together in series is used to achieve the above requirement. Lumped element representation of HPA1 output matching circuit with AC source (tube), anode load impedance (R s ), inter electrode capacitance (C out ) and output load impedance (R L ) is as shown in Fig. 3(a). The inter electrode capacitance between anode and SG remains in the range of 23pF to 28pF as per tube datasheet [3]. As first step of circuit design this inter electrode capacitance is resonated at the design frequency with an inductor (L ). Now the further matching circuit is designed by taking R s as source impedance and R L as load impedance. A virtual resistance R is considered in such a way that its value is geometric mean of R s and R L [6]. Then the loaded Q of this circuit is calculated as given in equation 1 [6]. Q = R 1 = R s 1. (1) R L R AC Anode load (Rs) Cout Lo L1 C1 R L2 C2 RL AC Anode load (Rs) Cout Leq C1 L2 C2 RL 3(a) 3(b) FIG. 3. (a)hpa1 output matching circuit with resonating inductance L and virtual resistance R (b)final configuration of output matching circuit The calculated loaded Q along with R s, R L and R are used to calculate the final circuit elements L 1,L 2, C 1 and C 2 as shown in Fig. 3(a). The final configuration of output matching circuit along with R s, R L and C out is as shown in Fig.3(b). Two inductance L and L 1 in parallel are replaced with an equivalent inductance L eq. To cover entire frequency band of 36 to 6MHz, initial calculations were performed at 35 and 62MHz with an intermediate value of R s and C out as 786 ohm and 25.5pF respectively as shown in Table 3. Table 3. CALCULATED VALUE OF HPA1 OUTPUT MATCHING CIRCUIT ELEMENTS Frequency(MHz) L eq (nh) C 1 (pf) L 2 (nh) C 2 (pf)

5 Insertion Loss (db) Return Loss (db) AKHIL JHA et al. The output matching circuit was simulated using CST design studio [7]. Starting with the available data at 35MHz (from Table 3), further tuning and optimization was performed to achieve final S-parameter response as shown in Fig. 4(a) and 4(b). Parametric sweep was performed for three different values of C out covering its range as given in tube datasheet [3] while keeping the tube resistance constant at 786 ohm. Insertion loss for the frequency range of 33 to 37MHz is well within -.3dB for all three cases (shown in Fig. 4(a)) while the return loss is better than -14 db for C out less than 25.5pF but it goes up to db in the lower side of band (shown in Fig. 4(b)) at 28pF , , , , Frequency (MHz) 786 Ohm/25.5(pF) 786 Ohm/23pF 786 Ohm/28pF 4(a) 4(b) FIG. 4. S-parameter response of HPA1 output matching circuit at center frequency 35MHz (a)insertion loss (b)return loss The value of various output circuit elements after the tuning and optimization is as shown in Table 4. Similar simulation was also performed at 62MHz to achieve approximate circuit element values. Thus the RF circuit simulations provided range of the values for circuit elements for frequency band of 36 to 6MHz , , , , Frequency (MHz) 786 Ohm/25.5pF 786 Ohm/23pF 786 Ohm/28pF Table 4. VALUE OF OUTPUT MATCHING CIRCUIT ELEMENTS AFTER OPTIMIZATION R s (ohm) C out (pf) C 1 (pf) C 2 (pf) L 2 (nh) L eq (nh) Modeling and simulation of coaxial RF circuit The final HPA1 output matching circuit has coaxial elements. Thus the optimized circuit element values (as given in Table 4) were used to find out the dimensions of equivalent coaxial elements. According to transmission line theory, a shorted stub having electrical length less than quarter wavelength at any frequency works as an inductor and the corresponding inductance can be calculated using equation (2) [8]. X L = Z tan (βl). (2) Where, reactance of stub (X L ) = L, characteristic impedance of stub = Z, propagation constant (β) = 2. To achieve an optimum height (~ 15mm) and floor space (~ 5 mm X ~ 5mm) for output matching circuit the characteristic impedance for stub was fixed to 6 ohm. Thus the calculated maximum electrical length (air as dielectric medium) of L eq and L 2 at centre frequency 35MHz were 1547 mm and 145mm respectively. The output matching circuit was modelled and simulated using CST Microwave Studio (MWS) [7]. The inductors ( L eq and L 2 ) were modelled as shorted stubs with rectangular outer conductor (OC) and the capacitors were modelled as coaxial cylindrical capacitors. A cut view of detailed model for output impedance matching circuit (or output cavity) with tube, decoupling capacitors (or DC blocking capacitors) and tuning rods is as shown in Fig. 5. Two section of the output cavity comprising stubs L eq and L 2 are shown. Both sections are separated by metal plate and are electrically coupled together by coaxial capacitor C 1 and finally stub L 2 is coupled to the output transmission line through coaxial capacitor C 2. Distance of shorting plungers from the base plate and output plate will be varied (practically) by the help of tuning rods. The tuning rods will be motor driven which will drive the shorting plungers by screw-nut arrangement. The copper shorting plunger is electrically connected with the Aluminium outer and copper inner through Beryllium Copper (BeCu) finger contacts. The tube is 5

6 FIP/P8-16 inserted inside the inner conductor (IC) of shorting stub L eq, details of tube terminal connections and decoupling capacitors are defined as detail A (shown in Fig. 6(a)). The details of capacitor C 1 and C 2 is defined as detail B (shown in Fig. 6(b)). FIG. 5. Detailed view of HPA1 output cavity All the decoupling capacitors are modeled by sandwiching layers of KAPTON film between two metal cylinders. For the frequency band of operation, a sufficiently low capacitive reactance (1 to 2 ohm) can be achieved with a capacitance of ~ 1nF or above. Five layers of KAPTON HN [9] film with each of thickness 125 micron provides capacitance for the anode and SG decoupling capacitor in range of 2 to 3nF. The isolation provided by the KAPTON layer (better than 5kV) is sufficient for the peak RF voltage of 11kV over the anode. Inner cylinder of anode decoupling capacitor connects with the anode flange through finger contacts while outer cylinder is bolted with the IC of L eq stub as shown in Fig. 6(a). The KAPTON layer extends out from the overlapping region to prevent breakdown from fringing fields. The inner cylinder of the SG decoupling capacitor connects with SG flange through finger contact while outer cylinder is bolted with cavity OC thus setting SG at RF ground potential. The anode decoupling capacitor with IC sits over the base plate through Delrin spacers while the tube sits over the cathode decoupling capacitor and filament connection. The inner cylinder of cathode decoupling capacitor connects with the cathode flange through finger while the outer cylinder forms IC of input cavity. Inner cylinder of CG decoupling capacitor forms OC for input cavity while outer cylinder connects with CG flange through finger contacts. The ICs of cylindrical capacitors C 1 and C 2 are modelled in two parts with a Teflon and Delrin spacer in between respectively. The upper part of IC of C 1 (shown as Upper IC) is connected with the IC of stub L eq (shown in Fig. 5) while the lower part (shown as Lower IC) is connected with the IC of stub L 2 through finger contacts as shown in Fig. 6(b). The Lower IC is moveable in and out of the overlapping region through Delrin shaft 1. In capacitor C 2 left hand side part is IC of stub L 2 (shown as Left IC) while the right hand side part is a copper cylinder (shown as Right IC ). The Right IC is moveable through the Delrin shaft 2 as shown in Fig. 6(b). The Right IC slides over the inner conductor of output port through finger contacts. 6(b) 6(a) FIG. 6. (a)tube connections and decoupling capacitors detail (b)cylinderical capacitors C 1 and C 2 details

7 Insertion Loss (db) Return Loss (db) Insertion Loss (db) Return Loss (db) AKHIL JHA et al. The model of output cavity was simplified for initial electro-magnetic (EM) simulation in CST MWS [7]. The tube, decoupling capacitors, tuning rods and shafts were removed as shown in Fig. 7(a). Port 1 was defined at the anode decoupling capacitor connection and port 2 was defined at the output transmission line. Various sections of output cavity and corresponding line diagram are shown in Fig. 7(a) and 7(b) respectively. Txin, Txint and Txout define the transmission line sections before, in between and after the coaxial circuit elements. The value of R s and C out were defined externally in CST design studio module. Txin C1 Txint C2 Txout Rs PORT 1 Cout PORT 2 Leq L2 7(a) 7(b) FIG. 7. (a)simplified output cavity model with different sections (b)line diagram of output cavity Simulations were performed at frequencies 35MHz and 62MHz (as center frequency) with value of R s as 52 ohm, 698 ohm and 88 ohm corresponding to three different anode biasing points while, the value of C out was kept constant at 25.5pF. Initial S-parameter response of output cavity at 35MHz is shown in Fig. 8 (a) and 8 (b). S-parameter response shows that with present configuration and dimensions -1dB bandwidth can be achieved over 4MHz band about centre frequency 35MHz. Increasing the value of R s reduces the bandwidth of circuit i.e. at 52ohm insertion loss (IL) is better than -.48dB while it increases up to -.89dB at 88 ohm as shown in Fig. 8 (a). Similar behaviour is observed in the return loss (RL) response of the output cavity as shown in Fig 8 (b)..5 33, , , , , , Frequency(MHz) 52 Ohm/25.5pF 698 Ohm/25.5pF 88 Ohm/25.5pF , , , , , Frequency(MHz) 52 Ohm/25.5pF 698 Ohm/25.5pF 88 Ohm/25.5pF 8(a) 8(b) FIG. 8. S-parameter response of output cavity at center frequency 35MHz(a)Insertion loss (b)return loss The S-parameter response at center frequency 62MHz is rather wideband compared to response at 35MHz. The IL is better than -.15dB at R s value 52ohm and goes up to -.5dB at 88ohm over the band of 4MHz at center frequency 62MHz. Similarly the RL response is better than -18dB at 52 ohm and goes up to -12dB at 88 ohm , , -.1 6, , -.5 6, , Frequency(MHz) 52 Ohm/25.5pF 698 Ohm/25.5pF 88 Ohm/25.5pF 6, , , , , , (a) 9(b) FIG. 9. S-parameter response of output cavity at center frequency 62MHz(a)Insertion loss (b)return loss -5-1 Frequency(MHz) 52 Ohm/25.5pF 698 Ohm/25.5pF 88 Ohm/25.5pF 7

8 FIP/P FUTURE SCOPE OF WORK Output cavity model used for initial EM simulation was a simplified one. In the next stage of EM simulations, the response of detailed output cavity model including decoupling capacitors, tuning rods and Delrin shafts will be studied. Further adjustment in the dimension of L 2 may be required to improve the RL response at lower edge of frequency band. Also a study of output cavity response will be carried out at few intermediate frequencies i.e. 45MHz, 55MHz etc. A detailed design of input cavity for the range of loading conditions provided by the tube cathode will be carried out. Since the range of tube input impedance is 67 to 72ohm which close to transmission line impedance of 5 ohm a coaxial wideband impedance transformer can provide a good match over the frequency band of operation. Further design process is ongoing in this direction. 7. SUMMARY A development program is underway for a wideband tube based pre driver amplifier (HPA1) in the ITER ICRF range. Selected tube for this development is EIMAC 4CW25B. To perform detailed tube parametric calculation a LabVIEW based code Tetrode Tube Calculator is developed. Tube parameters for different anode biasing conditions are calculated for output power range of 1 to 15kW. The output impedance matching circuit is designed as two L network connected in series which provides desired bandwidth response. A detailed 3D model of output cavity with corresponding S-parameter response at band edge frequencies 35MHz and 62MHz is provided. Output cavity in present configuration is able to achieve insertion loss better than -1dB for 4MHz band about the central frequency. REFERENCES [1] APARAJITA MUKHERJEE et al., Status of R&D activity for ITER ICRF power source system, Fusion Engineering and Design (215) [2] RAJESH TRIVEDI et al., Outcome of R&D program for ITER ICRF Power Source System, this conference [3] CPI, EIMAC DIVISION, 4CW25A and 4CW25B Tube Data Sheet, USA. [4] PREPARED BY THE STAFF OF CPI, EIMAC DIVISION, Tube performance computer for r-f amplifiers, Care And Feeding Of Power Grid Tubes, 5 th edition, pp 37-51, CA 947, USA [5] PREPARED BY THE STAFF OF CPI, EIMAC DIVISION, The three halves power law, Care And Feeding Of Power Grid Tubes, 5th edition, pp , CA 947, USA [6] CHRIS BOWICK, Impedance matching, RF Circuit Design, 2ed Edition, pp 69-72, Elsevier, USA (27) [7] [8] Pozar D M, Transmission line theory, Microwave Engineering, 4th edition, pp 59, John Wily & Sons Inc., USA (212) [9] DUPONT, KAPTON HN, datasheet