Research progresses on Cherenkov and transit-time high-power microwave sources at NUDT Open Access

High-power microwave (HPM) has emerged in recent years as a new technology allowing new application and offering innovative approaches to existing applications. HPM sources, transforming the energy of electron beams into microwaves, are the key components in the HPM system. Recently, considerable attention has been paid to the development of HPM sources.

The relativistic backward-wave oscillators (RBWO) using the Cerenkov mechanism is one of the most promising HPM sources due to its high output power, high power efficiency and high repetition operations. In S- and X-band, RBWOs have achieved output power of 3–5 GW [1], efficiency of 20–30% and repetition rate of 100 Hz [2–5]. However, because of the pulse shortening phenomenon, pulse duration of RBWOs usually cannot exceed 20–30 ns [6–10]. Naturally, the possible reasons for this phenomenon attract great attentions in the HPM field. Typically, S.D. Polevin et al. in Russia have studied Spontaneous pulse shortening occurring in an S-band resonant RBWO at gigawatt power levels [11]. Termination of microwave output is caused mainly by emission of charged particles from the plasma forming at the slow-wave structure (SWS) surface under the action of the intense RF field. Treating the SWS surface by low-energy high-current electron beam (LEHCEB) allowed production of 3 GW, 90 ns microwave pulses with energy of ∼250 J. To further increase the beam-wave interaction efficiency and output power attract great attentions in the HPM field, too. Typically, R.Z. Xiao et al. at Northwest Institute of Nuclear Technology have presented a klystronlike RBWO with a dual-cavity extractor as shown in Fig. 1 [12]. This device combines the advantages of transition radiation with Cerenkov radiation, has the characteristics of high efficiency, high power capacity, and stable frequency, and is a potential device for power combining. The particle-in-cell (PIC) simulation results reveal that microwaves with power of 10 GW, frequency of 4.3 GHz are generated, and conversion efficiency is 48% when diode voltage is 1.2 MV and beam current 17.3 kA. Furthermore, R.Z. Xiao et al. at Nuclear Technology have researched the mechanism of phase control in a klystron-like BWO by an input cavity as shown in Fig. 2 [13]. Since the microwave field is weak during the early time of starting oscillation, it is easy to be induced, and a small input signal is sufficient to control the phase of output microwave. In simulation, an input signal with 100 kW power and 4.21 GHz frequency can control the phase of 5 GW output microwave with relative phase difference less than 6% when the diode voltage is 760 kV, and beam current is 9.8 kA, corresponding to a power ratio of output microwave to input signal of 47 dB.

The magnetically insulated line oscillator (MILO) is one of the M-type devices. The advantages of the MILO include its high power output, stable operation, self-magnetic insulation, and compact configuration. The drawbacks of the MILO are its lower efficiencies and its lack of tunability. The MILO covers the frequency from L-band, S-band, C-band, and X-band to Ku-band [14–24]. Increasing the output power, pulse duration, power conversion efficiency, and repetitive frequency are four major development directions for the MILO. The typical representative is the L-band hard tube MILO designed by Sandia National Laboratory and Maxwell Laboratory in the USA [14]. The microwave with output power of 2 GW, frequency of 1.2 GHz, pulse duration of 175 ns and efficiency of 7% is generated under the voltage of 475 kV in the experiment [15]. D. Wang et al. at China Academy of Engineering Physics have studied an L-band double ladder cathode MILO as shown in Fig. 3 [24]. The microwave with output power of 3.57 GW, frequency of 1.23 GHz, pulse duration of 46 ns and efficiency of 8% is generated under the voltage of 740 kV and the current of 61 kA in the experiment.

The relativistic klystron amplifier (RKA) transfers kinetic energy of relativistic electron beam to HPM based on the principle of velocity modulation. RKA is a promising candidate for spatial coherent combining of HPM and has been investigated intensively during the past decades all over the world. Theoretical studies about RKA were carried out by Friedman et al. at the Naval Research Laboratory since 1970s, and typical experimental results were reported for the first time in 1985. A microwave with peak power of 6 GW and pulse duration of 80 ns is produced by the L-band RKA at 1.3 GHz when the diode voltage and current are 1 MV and 35 kA, respectively [25]. To enhance power conversion efficiency and pulse duration, Friedman proposed a new type RKA with inductively loaded wide gaps [26], which was demonstrated to be capable of generating 2.85 GW L-band HPM with efficiency of about 60% in the experiment. Furthermore, the triaxial klystron amplifier (TKA) is put forward to realize RKA operating in high frequency bands. Researches on RKA in China started in 1970s by Huang et al. at China Academy of Engineering Physics. Typically, HPM with peak power of 1 GW and pulse duration of 140 ns was produced by the S-band RKA in 2011 [27], and moreover coherent power combining of two S-and RKAs was realized experimentally for the first time. In addition, X-band TKAs are investigated actively by CAEP as well as the National University of Defense Technology (NUDT) to pursue long pulse HPM generation. Fig. 4 displays the multi-beam TKA put forward by China Academy of Engineering Physics, in which the electron beam and propagation tubes are divided into 18 small regions to suppress self-oscillation and leakage of TEM and TE modes. Typical experimental results are given in Fig. 4(c), which demonstrates a 0.98 GW HPM generated with pulse duration of 95 ns, when the diode voltage and beam current are 650 kV and 4.5 kA, respectively [28]. The injection microwave power is 50 kW, which yields a gain of about 43 dB. However, the multi-beam structure is fairly complicated, and self-oscillation is still observed in PIC simulation, which requires further investigation.

The relativistic transit-time oscillator (TTO) is one of the most promising high power microwave sources due to its virtues such as high power, high stability, monochromatic output RF signal and compact structure [29–31]. The first TTO named monotron was proposed in 1930s [32,33]. The initial experiment proved that transit-time effect could generate microwave, but the power conversion efficiency was only 0.2%. After decades of effort, scientists in Sandia National Laboratories developed a new TTO called split-cavity oscillator (SCO), whose power conversion efficiency was over 20% [34]. Since then, much progress in pursuit of higher output power, higher operation frequency and higher conversion frequency has been made in both theory and experiment [35–37]. The traditional relativistic TTO usually employs the conducting foils to guide the intense relativistic electron beam instead of the additional guiding magnetic field. Although this can certainly reduce the system's volume, weight, and energy consumption, the intense relativistic electron beam easily produces plasma at the conducting foils. The plasma is deleterious to the repetitive and long pulsed operation. To solve the above problems, the researchers presented the foilless TTOs, e.g. the researchers from China Academy of Engineering Physics proposed an X-band five-unit foilless TTO, which can deliver 780 MW HPM at X-band [38]. Its structure and the microwave waveform are presented in Fig. 5. However, its impedance is above 100 Ω, which limits the input power, and its guiding magnetic field is above 1 T. To lower the magnetic field and diode impedance, a coaxial foilless TTO, generating a microwave above 2 GW in L-band, is put forward in NUDT [39]. Due to the potential application of the Ku-band HPM in high power radar, communication, and other fields, it is meaningful to expand the operation band of the coaxial foilless TTO to Ku-band. However, the smaller size of the device at higher band certainly will cause the problems of electron beam collecting, electron beam guiding, RF-field breakdown and et al. Therefore, it becomes a focus of our future research to solve the above problems.

In this section, we report recent progress in achieving long pulse HPMs with GW-class power level in three bands [40,41]. In our design, special attentions are paid on two aspects. One is to keep a large gap between anode and cathode and a far distance between the downstream of the SWSs and the location of e-beam collector, through which the displacement of the cathode and collector plasmas can be negligible for pulse length of ∼100 ns; another is to optimize the electrodynamic structure to decrease the electric field on the surface of SWSs while keeping a relatively high power efficiency. The maximum RF electric field strength on SWS surfaces in the three designed HPM sources is under 1 MV/cm at the GW output power level.

The S-band RBWO is designed based on the basic principle of resonant RBWOs with its schematic diagram shown in Fig. 6. The resonant cavity is employed to enhance the power efficiency and to decrease the RF field. The main features are presented as follows: (1) the resonant cavity replacing the cutoff neck is used to enhance the efficiency. The electron beams can achieve pre-modulation through the resonant cavity, which is beneficial to beam-wave interaction; (2) the resonant cavity can completely reflect the microwave (the TM₀₁ mode) transmitted from the SWSs to the diode; (3) the radial distance between the cathode and the resonant cavity is about 5 mm, greater than 2 mm (the conventional RBWO with cutoff neck), thus avoiding electrons' scraping the cavity surfaces and reducing the RF field.

The schematic of the C-band RBWO is given in Fig. 7. A smooth waveguide is inserted between the cut-off neck and the SWSs. By adjusting the length of the inserter, the phase difference between the fundamental forward harmonic and the–1st backward harmonic can be varied, and the power efficiency can be increased compared with that of the structure without an inserter.

In order to test performance of the designed HPM sources, a long-pulsed high current electron beam accelerator is built, which can provide a 10–15 GW pulsed electric power on a 40 Ω matched load with the pulse duration of 150 ns at the single-shot mode or 1–20 Hz repetition mode.

Firstly, experiments are carried out in S- and C-band at the single-shot mode. Under the condition of the diode voltage 700 kV, beam current 7 kA and the guiding magnetic field 2 T, the microwave power is about 1 GW and 1.2 GW in S- and C-band, respectively. Fig. 8 shows typical experimental waveforms of C-band RBWO in single-shot mode. At such power level, the pulse duration usually kept 100 ns stable in both bands, and even reached over 110 ns occasionally. The experimental results agree very well with simulation results, which suggest almost no pulse-shortening phenomenon happening. Increasing diode voltage and beam current to about 1 MV and 10 kA, output microwave power for both sources increased to about 2 GW with a little decreased pulse duration of 90–100 ns, indicating that the pulse-shortening phenomenon does happen in such power level. After experiments, many traces of arcing are found nearby the downstream end of the cut-off necks in both sources, where simulation results show the RF fields are most intensive. In addition, a few spots are also found nearby the iris of SWSs.

In order to lengthen the pulse duration, measurements in two aspects can be taken. One is to improve the smoothness and cleanness of SWS surfaces and the vacuum level to increase the breakdown threshold. Experimental results have indicated that treating SWS surface with the LEHCEB can effectively increase the breakdown threshold from 1 MV/cm to about 2.5 MV/cm. The electrochemistry polishing or mechanical polishing are also the effective technological approaches. The other is to optimize the electrodynamic structures to decrease the electric field on the surface of SWSs while maintaining relative high power efficiency.

In 20 Hz repetition mode with burst duration of 5 s, the S-band RBWO was tested with diode voltage of 850 kV and beam current of 9 kA approximately, and the experimental results are given in Fig. 9. The output power is around 1.8 GW with pulse duration of 110 ns approximately, and the power efficiency is about 24%. The good overlapped waveforms suggest that the S-band RBWO has a stable performance in such power level.

In X-band, the single-mode RBWO is excluded in our design because the maximum electric field on SWS surface in such structure exceeds 1 MV/cm with output power of over 1 GW. It is evitable to employ overmoded SWSs with the D/λ > 1. Our investigations show that overmoded SWSs can effectively reduce the RF field on SWS surface, but the reduction is not as notable as the increase of D/λ due to the change of field distribution. Moreover, too large D/λ will result in much more difficulties in mode-selection. Therefore, it is suggest that the increase of D/λ should be moderate.

An X-band HPM source using overmoded SWSs with D/λ ≈ 3 is proposed and schematically shown in Fig. 10. The electrodynamic structure of two SWS sections separated by a drift tube is employed to improve beam-wave power efficiency through adjusting the length of the drift tube. A resonant cavity is used at the upstream of the first SWS section to reflect generated microwave going to the diode region. The design of the device is described in detail in Ref. 44. Both the power capacity and output mode purity of an X-band overmoded RBWO are substantially improved through optimizing the profile of the SWSs and geometric parameters of the whole electrodynamic structures.

Fig. 11 shows the experimental results of the improved device in 30 Hz repetition mode with diode voltage of 730 kV, beam current of 9.8 kA, and guiding magnetic field of 0.7 T. Through radiation power integrated at the far field, the measured output power is 2 GW with power efficiency of 28%, and pulse duration is 116 ns on average. The microwave spectrum measured in the experiments agrees well with that observed in the PIC simulation. Currently, we believe that the explosive emissions plasma formed on the surface of SWSs due to RF breakdown is the predominant cause of pulse shortening in RBWOs, and in our design, the measures taken is effective to suppress the pulse shortening phenomenon.

Since the first RBWO was developed in 1970, there have been many reports on RBWOs operating in the high frequency region (S-band, X-band, and millimeter wave), but discussions on low operation band (L- and P-bands) are scanty. The main reason is that the dimension of the low operation band RBWO is so large that it is difficult to manipulate in experiments. It should be mentioned that the HPM sources with frequency less than 2 GHz still have very important applications in many fields.

In this section, we report recent progress in compact L- and P-band RBWOs [42–46]. In our design, special attentions are paid on two aspects. One is to replace the hollow SWS for the coaxial SWS; another is to introduce a coaxial extractor at the end of the SWS. The quasi transverse electromagnetic (quasi-TEM) mode has no cutoff frequency in the coaxial SWS. Thus, the radius of the coaxial SWS becomes significantly smaller than that of the hollow SWS. Moreover, the peculiarities of the coaxial SWS are that the introduction of the inner-conductor can increase the space-charge limiting current, the interaction efficiency, and the coupling impedance is larger than that in the hollow SWS. Besides, the coaxial extractor at the end of the SWS is used to reduce the length and increase the efficiency.

The schematic of the compact L-band coaxial RBWO is given in Fig. 12 [42–44]. Compared with conventional RBWOs, there are the coaxial SWS and the coaxial extractor structure at the end of the SWS section. The coaxial SWS and the coaxial extractor are designed to reduce the size, realize the mode selection, and increase the efficiency of the device. In addition, it possesses the characteristics of frequency tuning. It shows that the resonance frequency decreases obviously with the increase of the inner-conductor radius. After removing the inner-conductor, its resonance frequency belongs to the S-band.

The experiment is carried out on a high-current electron beam accelerator capable of producing a 50 ns duration electron beam repetitively in the voltage range of 0.5–1 MV. When the diode voltage is 890 kV, the current is 7.7 kA, and the guiding magnetic field is 0.93 T, the radiated microwave with power of 2.1 GW and pulse duration of 41 ns is generated. Its efficiency is 31% and its main mode is TM₀₁ mode. Fig. 13 gives the microwave signal and the corresponding fast Fourier transform. Obviously, it can be seen that the generated microwave frequency remains approximately 1.61 GHz. Compared with the conventional RBWO (Φ10 × 70 cm), the volume of the compact L-band coaxial RBWO (Φ5.5 × 35 cm) decreases by 72.5%.

Furthermore, the mechanism and realization of a band-agile coaxial RBWO are presented. The operation frequency tuning can be easily achieved by merely altering the inner-conductor length. The key effects of the inner-conductor length contributing to the mechanical frequency tunability are investigated theoretically and experimentally.

As shown in Fig. 14, there is a specific inner-conductor length (L₀) where operation frequency jumps from one band to another band. When the inner-conductor length is larger than L₀, the operation mode is mainly the quasi-TEM mode, which is the basic mode of coaxial SWSs. Fig. 15(a) shows the electrical field distribution of the π mode of the quasi-TEM mode. Its resonance frequency is 1.68 GHz, which belongs to L-band. When the inner-conductor length is smaller than L₀, the operation mode is mainly the TM₀₁ mode, which is the basic mode of hollow SWSs. Fig. 15(b) shows the electrical field distribution of the π mode of the TM₀₁ mode, whose resonant frequency belonging to S-band is 2.43 GHz. In addition, the operation frequency is tunable within each operation band.

During simulation, the L-band microwave with frequency of 1.61 GHz is radiated when the inner-conductor length is 39 cm. Meanwhile, the S-band microwave with frequency of 2.32 GHz is radiated when the inner-conductor length is 5 cm. The frequency adjustment bandwidths of L- and S-band are about 8.5% and 2%, respectively. Moreover, the online mechanical tunability process is described in detail in Ref. 45. In the initial experiment, the generated microwave frequency remains approximately 1.59 GHz or 2.35 GHz when the inner-conductor lengths are 39 cm or 5 cm. In brief, this technical route of the band-agile coaxial RBWO is feasible and provides a guide to design other types of band-agile HPM sources.

The schematic of the compact P-band coaxial RBWO is given in Fig. 16 [46]. Compared with the L-band coaxial RBWO, there are only three periods SWS with both inner and outer conductor ripples. The double-corrugation SWS can reduce the period length of coaxial SWS for the same operating frequency of the RBWO and shorten the saturation time of the microwave signal. The length of the inner-conductor can be used to adjust the optimal phase difference between the forward fundamental harmonic and the–1st order harmonic. The coaxial extractor can also be used as a reflector to adjust the optimal ratio of the harmonic amplitudes and improve the axial distribution of a high-frequency field.

Typical experimental results under the condition of diode voltage of 582 kV, current of 8 kA, and guiding magnetic field of 0.9 T are exhibited in Fig. 17. It can be seen that the generated microwave frequency is approximately 897 MHz. The microwave power is measured to be 1.5 GW, and the pulse duration is about 40 ns, similar to the simulation results of 42 ns. Since the beam power was about 4.7 GW, the efficiency was approximately 32%. There is approximately a 74% decline of the volume compared with that of conventional RBWO under the same power capacity conditions.

In our laboratory, the investigations of MILO are mainly focused on improving power efficiency in L-, S- and C-band [17–19,47]. The typical structure of improved L-band MILO is shown in Fig. 18. A novel beam dump, a one-cavity RF choke section and a field shaper cathode are introduced into the improved L-band MILO. Through the axially moving of the beam dump disk, the axial gap width between the downstream end of the cathode and the beam disk can be adjusted continuously, which can affect the load current and hence the dc magnetic field in the SWS. The one-cavity RF choke section is optimized to reflect the leaking RF power toward the diode and reduce the length of SWSs. A field shaper cathode is proposed for avoiding the cathode flares in the triple point region when the MILO operates at several GW level outputs. At 1.755 GHz, the L band MILO radiates microwave power of above 3.1 GW when the diode voltage is 550 kV and the current is 54 kA. The pulse duration is above 40 ns, and the power efficiency is about 10.4%.

In order to further improve the power efficiency of MILO, a complex MILO is presented and investigated theoretically and numerically. The complex MILO composed of MILO-1 and MILO-2 is shown in Fig. 19. The MILO-1 and MILO-2 stand in the red solid rectangle and in the red dash rectangle, respectively. The emission regions stand in the dotted lines on the surfaces of the cathode-1 and cathode-2. The basic principle of the complex MILO is described in Ref. [48]. In simulation, the microwave powers of the complex MILO, MILO-1, and MILO-2 are 7.2 GW, 4.4 GW and 2.8 GW, respectively, when the diode voltage and the current are 620 kV and 58.4 kA, respectively. Thus, the power efficiency of the complex MILO is 19.9%. The microwave frequency of MILO-1 is 1.76 GHz and that of MILO-2 is 1.78 GHz, respectively.

A tunable MILO is put forward and simulated. The frequency of the MILO is co-determined by five factors: the inner radius, the outer radius, the period, the thickness of the SWS vanes, and the radius of the cathode. Furthermore, the most decisive factor is the depth of the SWS vanes, namely, the difference between the inner and the outer radii. Keeping other parameters unchanged while changing the inner radius of the SWS vanes can change the microwave frequency. However, the inner radius of the SWS vanes also affects the electromagnetic field boundary condition, the impedance, and the quality factor of the device at the same time. Consequently, the tuning bandwidth is very small. On the other hand, changing the outer radius of the SWS vanes can acquire a great tuning bandwidth while the influence on the MILO operation is very small. In conclusion, our scheme is to change the outer radius of the SWS vanes to tune the microwave frequency. Fig. 20 is the schematic of the tunable MILO. The online mechanical tunability process is described in detail in Ref. [49].

When the MILO is driven by the 430 kV, 40.6 kA electron beams, HPM is generated with power of 3.0 GW and frequency of 1.51 GHz, and the power efficiency is 17.2%. Fig. 21 gives the frequency and normalized power versus the outer radius of the vane. It shows that the microwave frequency decreases monotonously when the outer radius of the vane increases monotonously. Moreover, the frequency can reach above 3 GHz, but then the output power is lower than half of the output power. The 3 dB tunable frequency range (the relative output power is above half of the peak output power) is 2.25–0.825 GHz when the outer radius of the SWS vanes ranges from 77 mm to 155 mm, and the 3 dB tuning bandwidth is 92%, which is sufficient for the aim of large-scale tuning and high power output.

The TKA is an effective scheme to amplify microwave at high frequencies with lower power injection and achieve a 1 GW level output. The TKA can operate at single mode of coaxial TM₀₁ mode as long as the outer and inner radii of the coaxial drift tube close enough to cut off the operation mode. The power capacity can be effectively increased by increasing the radii of the coaxial tube and the coaxial cavities. The enhancement of the coaxial tube can provide a large-radius intense relativistic electron beam with high injection electric power. However, due to the coaxial cavities are over-sized, the designs of the cavities are significant. For the introduction of the inner-conductor, the TEM mode and the low-order coaxial TE mode can propagate in the coaxial tubes among the cavities of the TKA, which would lead to the symmetric and asymmetric mode competition. In this paper, an X-band TKA with an asymmetric input cavity is presented. The isolation method of the TEM mode leakage and suppression method of the asymmetric mode competition are developed.

As presented in Fig. 22, an X-band TKA with an asymmetric input cavity is designed to avoid the complicated diode structure of the previous TKAs developed by the Naval Research Lab (NRL), etc. To keep the axial electric field azimuthally symmetric in the gap region, the input microwave is divided equally into two parts by a power divider, and then injected into a reentrant coaxial cavity by two transmission channels, as illustrated in Fig. 23. Fig. 23 also plots the distribution of the total electric field in the input cavity obtained by a finite-different time-domain (FDTD) software CST when a TE₁₀ mode, with a transient power of 1 W and a frequency of 9.375 GHz, is injected from the rectangular waveguide. The dominant mode (TE₁₀ mode) in the rectangular waveguide converts into the TEM mode in the reentrant coaxial cavity and the coaxial TM₀₁ mode in the gap region of the input cavity. The azimuthal uniformity (ratio of the minimum over the maximum) of the axial electric field in the center of the gap region reaches about 92%.

Through the investigating propagation of the buncher cavity of the TKA, it can be noted that an appropriate buncher cavity for a TKA should satisfy three qualifications: 1) suppression of TEM-mode leakage; 2) no self-oscillation happening; 3) a high fundamental current modulation depth of the beam after interacting with the buncher cavity.

Two specially designed TEM mode reflectors with different Eigen frequencies are located in front of the buncher cavity and the output cavity, respectively to suppress the TEM mode leakage further. An appropriate TEM mode reflector for a TKA should satisfy three qualifications: 1) effective reflection to the TEM mode at the operation frequency; 2) little modulation to the beam at the operation frequency; 3) no excitation of the Eigen modes of the reflector by the electron beam. Fig. 24 plots the structure and the distribution of the total electric field and the axial electric field at the operation frequency in the reflector when injecting a TEM mode with 1 MW power from the right port. There is no obvious microwave leakage from the left port and the axial electric field is nearly zero at the beam radial position.

In Fig. 25(a), the output microwave of the TKA is obtained by a three dimensional (3-D) PIC code when the frequency and the total power of the input microwave are 9.375 GHz and 80 kW, respectively. In the beginning, the TKA can amplify the input microwave normally, while a pulse shortening happens at time of 70 ns. A frequency component of 11.46 GHz is observed in the FFT of the electric fields in the buncher cavity and the input cavity. The distribution of the axial electric field in the transverse section of the second gap of the buncher cavity is monitored, as illustrated in Fig. 25(b). An asymmetric mode competition occurs in the device, and the competition mode in the transverse cross section appears as coaxial TM₆₁ mode. The analysis of the three cavities shows that the three-gap buncher cavity has a resonant mode of coaxial TM₆₁₂ mode with a resonant frequency of 11.46 GHz, which is same with the frequency of the competition mode in the TKA.

The analysis indicates that the asymmetric mode is excited and amplified in the buncher cavity for the coaxial waveguide cannot cut off the corresponding asymmetric TE mode. A feasible suppression method of the asymmetric mode competition is to restrain the propagation of the coaxial TE₆₁ mode in the coaxial waveguide by a cavity reflector. To avoid an extra self-oscillation introduced by the resonant modes of the reflectors, the reflector in front of the buncher cavity should be optimized for high reflection coefficients both to the TEM mode and the coaxial TE₆₁ mode. The reflection coefficient to the coaxial TE₆₁ mode at the competition frequency of the reflector achieves 95.7%. Meanwhile, the reflection coefficient to the TEM mode at the operation frequency maintains over 99%, as shown in Fig. 26.

As presented in Fig. 27, there is no pulse shortening on the output power of the TKA with the coaxial TE₆₁ mode reflectors. The beam voltage and current are 570 kV and 6.5 kA, respectively. The frequency and power of the input microwave are 9.375 GHz and 100 kW, respectively. The output power of the TKA achieve about 1 GW, there are no pulse shortening on the device. The gain and efficiency are about 40 dB and 28%.

The backward-flowing power in the coaxial waveguide of the improved TKA is observed, as illustrated in Fig. 28. The power leakage to the input cavity from the coaxial waveguide is lower than 50 kW, which is less than 0.35% of the 140 MW backward-flowing powers in the buncher cavity. Furthermore, the power leakage from the output to the buncher cavity is lower than 2.5 MW, which is less than 0.2% of the 1.23 GW backward-flowing powers in the output cavity.

The designed TKA is demonstrated by the experiment. The diode voltage and beam current are about 570 kV and 6.3 kA, respectively [50]. The amplitude of the guiding magnetic field is about 1 T. The input microwave to the TKA is generated by a klystron amplifier which can operate at the frequencies range from 9.355 GHz to 9.395 GHz. The cold test demonstrates that the loss of the input microwave reaches over 7 dB in the input cavity. Namely the power of the microwave modulating the beam is actually lower than 20% of the input power. The typical experimental results are obtained when the power and frequency of the input microwave are about 90 kW and 9.37 GHz, respectively, as presented in Fig. 29. The power of the radiation microwave is ∼240 MW and the gain is about 34 dB. The output microwave simulated according the experiment waveform of the diode voltage is also plotted in Fig. 29. Considering the loss of the input power in the experiment, the input power is reduced to 15 kW in simulation. The experiment pulse duration achieves 100 ns, which approaches to the simulation result. It can be noted that there is no asymmetric mode competition resulting in the pulse shortening. Therefore, the asymmetric mode competition is effectively suppressed in the TKA with an asymmetric input cavity. The differences at the up and down edges mainly result from that the quality factors of the cavities in the experiment are lower than those in the simulation. Fig. 30 presents the radiation microwaves and the corresponding FFT result. The frequency of the generated microwave is 9.37 GHz, which is exactly identical to the frequency of the input microwave.

In this paper, a novel coaxial TTO with low guiding magnetic field is proposed. The schematic of the Ku-band coaxial TTO is presented in Fig. 31. The device mainly consists of six parts, which are the annular cathode, coaxial TM₀₂ mode resonant reflector, three-cavity buncher, dual-cavity extractor, novel electron collector, and coaxial output waveguide, respectively. The buncher and extractor are separated by the drift-tube to obtain efficient beam-wave interaction. The coaxial TM₀₂ mode resonant reflector with low surface electric field is used to prevent the microwave from propagating into the diode region and pre-modulate the electron beam. When the device operates, the annular electron beam emitted by the cathode transmits axially due to the guiding of the external magnetic field. The electron beam passes through the buncher and obtains velocity modulation. After a given-distance drift tube, the velocity modulation of the electron beam transfers into density modulation. Then, the bunched electrons interact with the electric field in the extractor and high power microwave is stimulated. Finally, the beam is dumped onto the electron collector and the Ku-band HPM are extracted by the coaxial output waveguide.

One virtue of the proposed device is that with the coaxial inner conductor and the volume wave as operation mode, this novel device has high power capacity. The operation mode of our device is TM₀₁ mode. One special feature of the coaxial TM₀₁ mode is that by loading the resonant cavities on both of the coaxial inner and outer conductors instead on only one of them and optimizing their depths, the strongest axial electric field can be located in the middle of the cavities, which is defined as quasi body wave. Fig. 32 shows the electric distribution of the TM₀₁ mode in the cavities loaded on both surfaces of the conductors and only on the outer one. Compared with the surface wave, volume wave can effectively lower the surface field. Besides, by increasing all radial dimensions of the device at the same increment, the working frequency keeps unchanged. This means we can further improve the power capacity with larger cross-section area.

Another virtue of the device is that with the inner-conductor, it can have less space charge effects. This means it has the merits of larger space-charge limiting current and lower guiding magnetic field. Larger space-charge can make it has lower impedance. With lower impedance, this device can have a lower voltage. For example, if for the traditional device (∼100 Ω), its voltage is 700 kV, with the same input electric power, for our device (∼50 Ω), the input voltage is only 500 kV. Lower guiding magnetic field can decrease its power demand of the magnetic field coils, and make the HPM system more compact.

The third virtue of the device is that it employs a coaxial TM₀₂ mode resonant reflector. Fig. 33 shows the electrical field distribution in the novel reflector at the desired frequency. As shown in the figure, the position of the maximum electric field is close to the electron transmission line at r = 4.3 cm instead of the corner wall, as is the case in the traditional one, which could increase the power handling capacity and pre-modulation of the electron beam. Moreover, by analyzing the external Q-factor of the buncher, we find that its value increase from 42 to 49, after introducing the novel reflector, which would be more suitable for beam-wave interaction.

Moreover, the proposed device employs a novel electron collector instead the tradition one. The novel collector is away from the extractor and beam interception occurs over an inclined plane, which has larger dump area compared with a vertical plane for the traditional collector (shown in Fig. 34). Larger dump area means less electron density. This can lower the probabilities of the primary or secondary reflection electron and plasma formation, which easily cause asymmetric mode competition. In addition, there is an electron absorption cavity together with the electron dump. It plays an important role in absorbing the reflection electrons and preventing them from propagating to the extractor region. Hence, the novel collector is in favor of suppressing the asymmetric competition mode.

To investigate the proposed Ku-band TTO with low guiding magnetic field and validate its performance, we carried out a two-and-one-half-dimensional (2.5 D) PIC simulation on the proposed device, when the diode voltage is about 420 kV, the diode current is about 8.3 kA and the guiding magnetic field is about 0.7 T Fig. 35 shows the output microwave power versus time and the frequency spectrum. From the figures, we can see that with diode voltage of 420 kV, diode current of 8.3 kA and the guiding magnetic field of 0.7 T, the output power is 1 GW, and the microwave frequency is 14.25 GHz.

The proposed Ku-band TTO with the designed parameters derived above was fabricated. For investigating and verifying its performance, a primary experiment on the Ku-band device guided by the magnetic field was performed at an intense relativistic electron beam accelerator, which is capable of producing a 50 ns duration electron beam in the voltage range of 0.3–1 MV. When the diode voltage, diode current, and guiding magnetic field are 500 kV, 10 kA, and 0.7 T, respectively; the typical wave forms of the diode voltage, beam current, and the radiation microwave are shown in Fig. 36(a). The integrated power is 1 GW, corresponding to the conversion efficiency of 20%. The received microwave signal with its FFT is given by Fig. 36(b). The main frequency 14.3 GHz was observed in the experiment, which is in good agreement with simulation one [51].

The research progresses on Cherenkov and transit-time HPM sources at NUDT of China are presented. The main contents and innovative work are as follows.

1. Recent experimental results of three kinds of long pulse HPM sources operating in S, C and X-bands are reported. The S-band source generates 1.8 GW power with pulse duration of 110 ns in 20 Hz repetition mode. The X-band source generates 2 GW microwave power with pulse duration of 110 ns in 30 Hz repetition mode. It is suggested that explosive emissions plasma formed on the surface of SWSs due to RF breakdown are the crucial factor limiting pulse duration. In addition, there are a coaxial SWS and a coaxial extractor in the compact RBWO, which are designed to reduce the size and increase the efficiency. At L-band, a 2.1 GW microwave is obtained and the efficiency is 31%. At P-band, a 1.5 GW microwave is obtained and the efficiency is 32%. There is approximately a 75% decline of the volume compared with that of conventional RBWO under the same power capacity conditions.

2. Investigations on the MILO are focused on the enhancement of the power efficiency. At 1.755 GHz, a microwave with power of 3.1 GW and efficiency of 10.4% is obtained. To further improve the power efficiency of MILO, a complex MILO is presented. In simulation, the power efficiency of the complex MILO is 19.9%. A tunable MILO is put forward and simulated. The 3 dB tunable frequency range is 2.25–0.825 GHz and the tuning bandwidth is 92%.

3. The X-band TKA and Ku-band coaxial TTO with the high power capacity and power efficiency are reported. An asymmetric input cavity is designed to avoid the complicated diode structure of the previous TKAs. By a reflector with high reflection coefficients both to the asymmetric mode and the TEM mode, the asymmetric mode competition is effectively suppressed. At 9.37 GHz, a microwave with power of 240 MW and gain of 34 dB is obtained. In the Ku-band coaxial TTO, the quasi body wave occurring in the middle of the inner and outer conductors is chosen as the operation mode to enhance the power capacity. The inner-conductor is introduced to increase the space-charge limiting current and reduce the impedance. The third virtue is that it employs a coaxial TM₀₂ mode resonant reflector. In addition, the device employs a novel electron collector instead the tradition one. At 14.3 GHz, a microwave with power of 1 GW and efficiency of 20% is obtained.

The authors wish to thank Dr. Baoliang Oian, Dr. Ting Shu, Dr. Chengwei Yuan, and Dr. Huihuang Zhong for their assistants in the research. This work is supported by the National Natural Science Funds of China under under Grant No. 11505288, Provincial Natural Science Foundation of Human and Scientific effort project of NUDT.