Real-time steerable frequency-stepped Doppler backscattering (DBS) system for local helicon wave electric field measurements on the DIII-D tokamak

The use of millimeter wave Doppler backscattering (aka Doppler reflectometry) to diagnose low-frequency plasma turbulence has become standard in fusion research plasmas because of the high spatial (sub-centimeter) and temporal resolution (sub-millisecond) measurement capabilities.^1,2 Doppler backscattering (DBS) typically uses a millimeter wave beam to perform wavenumber resolved, local measurements of plasma density fluctuations $ñ$ that satisfy the Bragg scattering wavenumber relation, $kx̃=−2ki$, where the subscripts $x̃=ñ$ or $b̃‖$ refer to the probed fluctuation and i to the incident millimeter waves at the measurement location. Millimeter waves backscattered from the DBS beam by turbulent density fluctuations near millimeter wave cutoff surfaces carry local information about the turbulence.^3–5 Turbulence at different scale lengths has already been studied in DIII-D (e.g., Refs. 6 and 7, NSTX,^8,9 MAST-U^10–12), and other tokamaks using this technique. The E × B velocity and wave electric field (E) amplitudes are inferred from the measured Doppler shift in the quadrature spectrum and the scattered power provides a relative measurement of the density fluctuation level. Millimeter-wave diagnostics such as DBS have more recently also been applied to probe radio or high frequency waves in fusion plasmas. Various high-frequency waves, including ion cyclotron emission,¹³ high-frequency Alfvén eigenmodes,¹⁴ and lower hybrid waves,¹⁵ have been investigated.

The new DBS system described here is based on a previous prototype E-band DBS system, which demonstrated the ability to observe a broad range of radio frequency plasma waves.¹⁴ The new system has also demonstrated a capability to observe fluctuations near the frequency of externally injected high-power helicon waves (476 MHz). Previously, the observation involved direct scattering of the millimeter waves from broadband density fluctuations around the helicon frequency. These broadband helicon fluctuations were thought to be generated via the interaction of the injected helicon wave with turbulence. One possible interaction is modulation of the helicon antenna coupling by turbulence in the plasma scrape off layer (SOL) in front of the antenna.¹⁶ This paper will discuss observations of high-frequency fluctuations with the new system that result from a distinctly different mechanism: direct backscattering from turbulent density fluctuations modulated by the injected helicon wave. As explained below, this mechanism will allow the spatial distribution of the helicon wave to be routinely generated with minimal assumptions.

This paper also describes the integration of the DBS diagnostic with an electron cyclotron heating (ECH) overmoded waveguide propagation system,^17,18 which allows the launched beam to be steered both toroidally and poloidally. The integration utilizes a high-power overmoded waveguide switch, which allows either ECH heating power to be launched or DBS experiments to be performed on the fusion plasma. The waveguide switch selects either ECH or mm-wave to propagate to the plasma using a pneumatically controlled linear actuator to control a mirror. Insertion of the mirror allows the DBS system to couple to the plasma, whereas removal of the mirror allows free propagation of the megawatt-level ECH to the plasma. The development and demonstration of this technique for interfacing the DBS system with plasma establishes a model for potentially adding new diagnostic capabilities to future burning plasma facilities that employ ECH. Among the details to be discussed is the waveguide switch that allows either the ECH source or DBS diagnostic to couple millimeter waves into the overmoded waveguide transport system, and the quasi-optical technique utilized to couple the DBS source radiation into the waveguide switch.

This paper reports these new design improvements and describes results from recent helicon experiments in the DIII-D tokamak. Both frequency stepping and dynamic steering during single plasma discharges are used to efficiently extract a wealth of fluctuation data with the DBS system. These results establish the DBS system as a sensitive, flexible diagnostic with fine temporal and spatial resolution as well as broad spatial and wavenumber coverage to characterize a broad spectral range (f = 20–550 MHz) of plasma waves present in a single plasma discharge.

The primary aim of the upgraded DBS system is to measure helicon wave-field amplitude and spatial structure and any interaction with simultaneously measured turbulence. This supports objectives of the DIII-D research program to establish helicon waves as an efficient off-axis current drive system to achieve the goal of steady-state reactor operation.^19–21 Measurements of helicon wave-field amplitude and spatial structure with the new DBS will be used to validate theory and code-predicted (GENRAY²² and AORSA^16,23) performance of helicon wave heating and current-drive in terms of propagation and absorption inside the plasma.

Details of the improved system design are discussed in Sec. II. Section III describes scattering measurements in recent DIII-D plasmas, including results demonstrating frequency stepping and beam steering within individual discharges. Section IV discusses the physics importance and summarizes the significance of the helicon measurements achieved to support the helicon current drive program at DIII-D.

The system reported here is upgraded in several ways relative to the prototype system,¹⁴ resulting in improved sensitivity and spatial and wavenumber coverage, especially for helicon wave measurements. Figure 1 shows the overall structure of the new DBS. The new system retains the quadrature configuration of the prototype, with the ability to change the operating frequency of the millimeter wave probe beam. However, the millimeter wave source has been modified to allow for real-time tuning over the entire 60–90 GHz E-band range. In addition, several modifications to the receiver circuit improve the system’s sensitivity, including hardening against the pickup of stray 476 MHz helicon radiation. Significantly, the new DBS also differs from the prototype in how it interfaces with plasma. The new DBS system is quasi-optically coupled to a waveguide switch, which integrates the DBS into a corrugated waveguide. This waveguide normally transports ECH power to a launcher, or steering assembly, for directing the power into the plasma.^17,18 The DBS uses the toroidal and poloidal aiming capabilities of the ECH launcher as well as the ECH optics focusing capability. The ECH launcher has a much wider range of steering angles than the steering assembly used by the prototype DBS.¹⁴ 

An overview of the system in Fig. 1 shows many of the modifications to the millimeter wave circuit, including changes to the millimeter wave source and the enclosure of the receiver circuit with Faraday shielding. The range of millimeter wave operating frequencies is expanded to the full E-band (60–90 GHz) by replacing a pair of phase-locked dielectric resonator oscillator (PLDRO) sources featured in the prototype with a remotely programmable frequency synthesizer (Mercury DS-3002). The synthesizer can produce frequencies over a range of 0.1–20 GHz with low phase noise. The synthesizer is used to produce frequencies in the 10–15 GHz range that are multiplied to the 60–90 GHz range by active ×6 frequency multipliers (Eravant, SFA-603903620-12SF-E1). Laboratory tests show high-frequency stability and improved signal-to-noise compared to the prototype system¹⁴ when using the synthesizer.

In contrast to the prototype system, the new DBS system uses active ×6 frequency multipliers instead of ×4 multipliers. The change in the multiplication factor in the multipliers was made because the programming circuitry of the synthesizer is anticipated to be vulnerable to neutron radiation from DIII-D plasmas. To mitigate this danger, the synthesizer is also placed ∼5 m from the tokamak, outside a concrete shield wall that largely contains the D-D fusion neutrons. This minimizes the possibility of neutron damage to the programmable circuitry, as well as random bit-flips of the program memory and logical control circuitry of the synthesizer. The synthesizer output is transmitted to the multipliers via a low-loss (∼0.15 dB/ft) 30-feet coax RF cable (model: Teledyne True Blue 420) through a penetration in the shield walls. The cable is well-shielded against the pickup of any stray helicon radiation from the high-power antenna that leaks into the area around the tokamak.

The programmability of the synthesizer also allows for efficient exploitation of DIII-D discharges, which can last up to 10 s. The operating frequency can be stepped during a discharge, allowing probing of multiple radial locations during a single discharge (as compared to multiple repeat discharges with a fixed source like in the prototype system).¹⁴ In sweep mode, the synthesizer can step in increments as small as 1 Hz through a chosen range, with a 1-ms settling time after each step. The fixed dwell time, which applies to all steps, can be set within a range from 0.5 ms to many hundreds of milliseconds. The synthesizer accepts an external trigger that controls the timing of the program execution. This trigger is taken from the DIII-D timing system, which coordinates the discharge control and data acquisition systems, ensuring synchronized DBS operation. Finally, the synthesizer is remotely programmed via a USB relay and a USB interface.

As with the prototype DBS system, the receiver circuit separates low-frequency and high-frequency fluctuation signals using diplexers at the millimeter wave mixer in-phase (I) and quadrature (Q) signal output ports. New amplifiers (Analog Module 351A-1-50-NI) are used for lower noise with more uniform gain and higher saturation voltage throughout the required frequency range (f ≲ 10 MHz). Prior to amplification, the high-frequency signals are down-shifted to a lower frequency by mixing with a 500 MHz crystal oscillator in a Mini-circuits low-frequency mixer. As with the previous system, all signals from the DBS system are recorded with a fast-sampling digitizer (100 MS/s) with high input bandwidth, f < 100 MHz, limited to f < 50 MHz by external anti-alias filters.

A second, more significant, upgrade is to improve the rejection of unwanted external pickup of 476 MHz radiation from the high-power helicon antenna. A 1.5-mm thick aluminum Faraday shielding box (Fig. 2) now encloses the millimeter-wave quadrature mixer, diplexers, crystal oscillator, and RF mixers. Coax cabling along with RF filters to reject 476 MHz are used for all signals and power supply cables penetrating the shield box. For the low-frequency signals, a 36 MHz low pass filter is used (Mini-Circuits SLP-36+), while for the down-shifted high-frequency signals, a 1.9–31.5 MHz high pass filter is used (Mini-Circuits ZABP-16+). A second RF pickup rejection technique is the insertion of 8–24 GHz bandpass filters (Mini-Circuits ZHSS-8G-S+) at the source side of active multiplier inputs in the millimeter-wave circuit (Fig. 1). The active multipliers have coaxial inputs that could otherwise allow 476 MHz pickup along the input line to enter the multiplier and potentially modulate the output LO and probe beam millimeter waves. Laboratory tests were conducted to validate the RF shielding effectiveness by using a mockup setup where a dipole antenna connected to a high-power waveform generator radiated a 476 MHz signal, simulating the electromagnetic interference expected during experiments. The initial RF pickup in the detection channels was measured using a spectrum analyzer without any shielding or filtering. Sequential improvements were then implemented, including the installation of 8–24 GHz bandpass filters to suppress unwanted signals, enclosing the receiver circuit in a 1.5-mm-thick aluminum Faraday shield to block stray RF radiation, and adding additional high-pass and low-pass filters at signal penetration points to prevent interference from 476 MHz radiation and harmonics. After each modification, the signal levels in the detection channels were re-measured, confirming that these combined shielding and filtering arrangements effectively reduced RF pickup by ∼34 dB.

The millimeter wave circuit couples (Figs. 1, 3, and 4) into an electron cyclotron heating system to take advantage of the high-power millimeter wave beam steering assembly, or launcher, for steering the DBS probe beam. The launcher system^17,18 contains a paraboloidal fixed mirror that focuses radiation from a 60.325 mm corrugated overmoded waveguide onto a steerable flat mirror. The focusing mirror is designed so that radiation propagating in the HE₁₁ mode in the waveguide, after emerging from the waveguide aperture and reflection from the focusing mirror, forms a Gaussian beam in the far field. At 110 GHz, the focusing mirror produces a 13 cm diameter spot size beam at a distance of 1 m from the radiating waveguide aperture,²⁴ which is well inside the core plasma.

The steerable mirror is driven by a two-axis motorized system for poloidal and toroidal beam steering. The poloidal steering angle can range from 29° above horizontal to 43° below horizontal. The toroidal steering angle can range from 34° right of the inward major radial direction to 28° left when viewed from above. The mirror angles can be set before each plasma discharge and held fixed throughout, in a mode of operation similar to steering for results reported in Ref. 14. However, a control system for the motor that sets the poloidal steering angle can also be programmed to change the angle throughout the discharge, as quickly as 40–50°/s, allowing significant angle changes during the course of a discharge. With this steering capability, the DBS system can cover a fluctuation wavenumber range of nearly |k_θ| = 0–25 cm⁻¹. This in-shot steering capability is demonstrated in experiments reported here. The toroidal steering angle is also set by a motor, but the control system for the motor can only set the angle before each discharge and hold it steady throughout. To prepare for an experiment, ray tracing is performed with the GENRAY²² or TORAY²⁵ millimeter wave ray tracing codes to determine the optimum steering angles for the target wavenumber, using a model density profile and equilibrium magnetic field of the anticipated discharge (typically based on a similar discharge that is being recreated).

As mentioned earlier, to interface with the ECH waveguide propagation system, the millimeter wave circuit is quasi-optically coupled via a T-shaped waveguide switch²⁶ into a circular corrugated waveguide (Figs. 1, 3, and 4) leading to the plasma. This corrugated waveguide normally feeds power directly from a gyrotron to the ECH launcher. It has a 31.75 mm inner diameter at the location of the switch, tapering up to the 60.325 mm inner diameter waveguide of the launcher assembly ∼1 m from the launcher. All legs of the switch are also corrugated, with a diameter of 31.75 mm. Pneumatic actuators are utilized to insert a gold-coated metal mirror into the junction of the switch allowing a 90° reflection of the DBS probe radiation into the corrugated waveguide leading toward the plasma. Withdrawal of the mirror leaves the corrugated waveguide open for high-power gyrotron ECH radiation to pass into the plasma. Interlocks control the operation of the switch, and an operational safety protocol is implemented to protect both the ECH and DBS systems. It is important to mention that with this arrangement, the DBS diagnostic can operate only when the waveguide is not used by the ECH gyrotron. The waveguide switch²⁷ design ensures negligible leakage from the ECH input leg to the DBS input leg when in position for ECH operation.

The quasi-optical assembly for coupling the millimeter wave circuit to the switch consists of a scalar horn, a pair of lenses, and a fully adjustable mirror (Figs. 3 and 4). DBS launch radiation from the horn is focused by a hyperbolic high-density polyethylene (HDPE) lens toward a second hyperbolic lens, which focuses the radiation to the entrance aperture of the waveguide switch to facilitate high-quality coupling to an HE₁₁ mode. Testing indicates that this assembly effectively couples millimeter waves into the HE₁₁ mode throughout the full E-band range.

For compactness, the quasi-optical assembly is laid out on a breadboard with a mirror between the two lenses to redirect the path of the millimeter wave beam. The lenses and mirror are mounted on stages with various degrees of translatability to permit alignment of the components. The quasi-optics and switch are aligned on the DBS breadboard in the laboratory before the breadboard is positioned to integrate the switch into the ECH waveguide at the DIII-D facility. The alignment process ensures precise microwave beam coupling into the ECH waveguide using a He–Ne laser for initial alignment and a 1D millimeter wave beam profiler for verification. The He–Ne laser, launched from a fixed mount at the intended ECH switch position, aligns the quasi-optical system by verifying the placement of collimating and focusing lenses, the flat mirror, and the scalar horn antenna opening. Since the laser does not pass through the HDPE lens, alignment is performed before lens installation to ensure proper beam path positioning. The switch is then temporarily mounted in place of the laser to confirm the target location before being removed to fine-tune the quasi-optical system, ensuring the focused beam waist aligns with the ECH waveguide input aperture. A 1D beam profiler is used to measure the beam profile both vertically and horizontally at various distances, confirming collimation and proper focusing for efficient microwave beam propagation in the HE₁₁ mode.²⁸ This ensures minimal losses and optimal system performance across the 60–90 GHz operating range of the DBS system. At 63 GHz the beam diameter is 1.9 cm, while at 88.8 GHz, it is 1.6 cm. The beam diameter is measured between points of the intensity profile that have 1/e² of the peak intensity. After the switch is mounted, beam profiles for different operating frequencies are measured vertically and horizontally at different distances from the output leg aperture in the laboratory. Gaussian profiles are verified for a broad range of frequencies (63–87 GHz). After the switch is integrated into the ECH waveguide at the DIII-D facility, horizontal and vertical profiles are again obtained inside the DIII-D vacuum vessel. For the in-vessel measurements, the 78 GHz signal couples into the waveguide switch, propagates through ∼ 3 m of corrugated waveguide to the launcher, and is launched into the vacuum vessel by the ECH beam focusing and steering system. Figure 5(a) shows a beam profile at 78 GHz measured in the laboratory at 10″ away from the waveguide switch output with the aligned quasi-optical system and switch mounted on the breadboard. Profile measurements of the beam emerging from the launcher within the DIII-D vessel are shown in Fig. 5(b). The beam is poloidally steered to be horizontal and toroidally steered in the negative major radial direction. The profile is measured at a distance of 10.5″ from the steering mirror. Deviation from a Gaussian structure in Fig. 5(b) is due to reflections from the detector setup, table, and tokamak walls. These profiles indicate reasonably good coupling from the quasi-optics to the corrugated waveguide and steering system into the plasma.

The quasi-optical coupling scheme for the new DBS delivers several benefits. One benefit is that the millimeter waves propagate in the corrugated waveguide as an HE₁₁ mode²⁹ with a low loss between the waveguide switch input and the ECH launcher. Another benefit is that the radiation propagating in the HE₁₁ mode benefits from the launcher focusing mirror design in that it creates a reasonably good beam profile in the far field [Fig. 5(b)]. A third benefit is spatial filtering of radiation transmitted from the plasma to the millimeter wave circuit, which enforces spatial localization of the measurement and limits noise from plasma electron emission. Different waveguide modes produce different antenna patterns in plasma with different spatial footprints. Scattered millimeter waves from the DBS probe beam, as well as electron cyclotron emission from the plasma, that match the antenna patterns of other modes besides the HE₁₁ can couple to these modes in the corrugated waveguide and propagate back toward the millimeter wave circuit. The quasi-optical assembly preferentially couples the HE₁₁ mode into the millimeter wave circuit, minimizing contamination from higher-order modes. Beam profile measurements [Figs. 5(a) and 5(b)] confirm that the launched beam maintains a nearly pure Gaussian profile across the entire frequency range, closely matching the expected HE₁₁ mode characteristics. This ensures effective spatial localization of the DBS measurement. Although noise reduction due to modal suppression has not been directly measured, the beam profile scans demonstrate minimal mode contamination, supporting the conclusion that unwanted waveguide modes are effectively suppressed.

The choice to couple the DBS system to the ECH system via the waveguide switch poses mechanical and electrical challenges. The first challenge is to electrically isolate the DBS system from the ECH waveguide. This is accomplished using quasi-optical coupling. However, this creates a mechanical challenge in the process, by mechanically decoupling the DBS system from the switch. The waveguide undergoes thermal expansion and contraction during experimental operations which could translate the switch by up to ∼0.5 cm, potentially misaligning the quasi-optical assembly with the switch. The solution adopted here is to attach the switch to a 0.5″ thick plate of G-11 fiberglass laminate that also connects to the optical breadboard on which the DBS system and quasi-optical assembly are mounted (Figs. 3 and 4). The insulating character of the G-11 plate ensures that electrical isolation is maintained. The breadboard is mounted via rails on a support frame that is free to translate together with the switch during thermal expansion and contraction of the waveguide and vacuum vessel (Fig. 4). The rails are necessarily angled away from horizontal for correct alignment with the waveguide. To avoid burdening the switch with the weight of the breadboard and DBS system, the breadboard and switch are attached to the support frame via a cable and pulley system to a spring underneath the breadboard (not visible in Fig. 4) that attaches to the support frame.

A new method of polarization control has also been implemented for the new DBS system using a waveguide rotary joint (Spinner BN 636282), which coaxially joins two rectangular waveguide sections and allows them to independently rotate freely around a shared axis (Figs. 1 and 3). Millimeter waves in fundamental mode are transmitted between the waveguide sections, effectively spatially reorienting the wave electric field direction when the waveguide sections are rotated to have different orientations. Typical insertion loss within this rotary joint is ∼1 dB (0.2 dB variation over 360° of rotation). One port of the rotary joint attaches to the millimeter wave circuit while the other port attaches to the scalar feed horn antenna (Quinstar QSH-E2500) via a rectangular to circular waveguide transition. The rotary joint can be arbitrarily rotated to select the angle of the rectangular port of the transition feeding the scalar horn, allowing the polarization of the radiation emerging from the horn antenna to be adjusted appropriately relative to the edge pitch of the plasma magnetic field to couple to either X-mode or O-mode. The orientation of the rotary joint is set using a motor driven by a remotely controllable power supply. The polarization can be rotated by 90° in less than a minute, making changes between X and O-mode between discharges straightforward. A camera is used to monitor the position of the rotary joint while the motor drives the rotation.

The frequency stepping and steering capabilities of the DBS are demonstrated in several different DIII-D discharges. This has allowed the turbulence, Alfvén wave type instabilities, and externally launched helicon waves to be studied as a function of location and probed wave number. The radial location of DBS measurements is limited by the cutoffs. In the case of an X-mode DBS beam polarization, the cutoffs depend on both the plasma density and B-field value. For example, with B = 2T, in X-mode, the range of density is n_e0 ∼ 4 × 10¹² cm⁻³ to > 10¹⁴ cm⁻³ for f_DBS = 60–90 GHz. Figure 6 shows the turbulence measurements obtained while stepping the operating frequency in a constant density, L-mode discharge in DIII-D with plasma current and toroidal magnetic field of I_P = 1 MA and B_T = 2T, respectively. Neutral beam heating with ∼3 MW is utilized and a peak density, n_e0 ∼ 2 × 10¹³ cm⁻³ is maintained. Measurements are obtained using the synthesizer programmed to cycle sequentially through DBS launch frequencies of 60, 66, 72, and 78 GHz (with 400 ms dwell time at each frequency). The polarization of the millimeter probe beam is X-mode, and the steering mirror is fixed to a toroidal angle of 0° and a poloidal angle relative to normal incidence of ∼1.5°, which is suitable for probing lower k fluctuations. This allows the DBS beam to probe from the edge of the plasma (ρ ∼ 1) to the inner core (ρ = 0.45) and measure the Doppler shift of the turbulence, which is predominantly caused by plasma E × B rotation. Here, ρ is a flux surface coordinate giving the square root of normalized toroidal flux enclosed by the flux surface, and ρ = 1 corresponds to the last closed flux surface. The quadrature spectrum of turbulent density fluctuations measured by the low-frequency channel as the system steps sequentially between operating frequencies is shown in Fig. 6(a) while line average plasma density remains approximately constant [Fig. 6(b)]. The quadrature spectrum is the spectrum of the complex valued representation of the electrical field of the scattered radiation, E = I + iQ, where I and Q are the low-frequency in-phase and quadrature signals. As the synthesizer steps from low to higher operating frequencies, the DBS probed location moves from edge to core, and a peak in the spectrum is observed to shift to higher frequencies as ρ decreases. The peak frequency is the Doppler shift frequency [Fig. 6(a)] f_D, where 2πf_D = k · v, k is the probed wave number, and v is the turbulence lab frame phase velocity. Turbulence lab frame phase velocity is often dominated by the E×B velocity,^3, v_E × B=E × B/B² for DIII-D plasmas³⁰ with strong torque from beam injection such as considered here. Raytracing with the GENRAY code²² shows [Fig. 6(c)] the ray trajectory for each frequency, from which the measurement location (the location of deepest penetration) is determined. The ray tracing shows an increase in the probed wavenumber as radial location ρ decreases [Fig. 6(d)]. The plasma radial electric field (E_r) is estimated from the probe wavenumber [as predicted by GENRAY, Fig. 6(d) as well as DBS measured Doppler shift in frequency (f_D from Fig. 6(a)]. The E_r profile shows a typical³⁰ behavior characterized by elevated values at mid-radius, gradually decreasing to zero at the plasma edge. This is consistent with typical plasma conditions and E_r measured by other diagnostics^30,31 at DIII-D. This is attributed to the interplay between ion orbit loss and the radial force balance expressed through Ohm’s law.³⁰ The strong ion orbit loss in the outer mid-radius region creates an inward-pointing radial electric field to confine the remaining plasma, leading to the observed peak at ρ ∼ 0.6. The radial electric field then drops to zero at the very edge (ρ > 0.9) because of returning particles from the scrape-off layer. Note that the difference in magnitude of E_r shown here and in Ref. 30 can be attributed to the poorer L-mode confinement in the present study, as opposed to the better H-mode confinement discussed in the referenced work.

A second discharge (Fig. 7) illustrates the measurement of mid-frequency fluctuations (f ∼ 5–5.5 MHz) that resemble global Alfvén eigenmodes (GAE), as described in Refs. 32 and 33. The discharge for this case is a high-power (∼14 MW) neutral beam-heated L-mode plasma with B_T = 1.7 T and I_P = 800 kA. The beam power is delivered by a combination of neutral beams with different injection geometries (see e.g., Ref. 32) injecting at different times. For this discharge, the DBS system operated in O-mode polarization and was programmed to step using 1.2 GHz increments starting from 60 GHz at t ∼ 370 ms, with a dwell time of ∼200 ms at each frequency. The high-frequency modes are detected as early as t = 2000 ms when the operating frequency is 69.6 GHz and continue to be detected nearly continuously until t ∼4000 ms when the operating frequency is 80.4 GHz. The line average plasma density is ∼5.2 × 10¹³ cm⁻³ throughout the period where these modes are observed. GAEs are global modes with long radial wavelengths (typically comparable to the minor radius) and their associated density perturbation modulates the millimeter wave index of refraction.³³ This in turn modulates the optical path length of the millimeter wave radiation from the launcher on its path into the plasma and then back to the launcher after scattering from turbulence. Figure 7(a) shows the spectrum of the phase fluctuations (⁠$ϕ̃$⁠) from the low-frequency signal, which are proportional to path length fluctuations, for a later portion of the discharge, with a slightly higher line average density ∼5.3 × 10¹³ cm⁻³, from t = 3400–4000 ms. The optical path length of the radiation is given by ϕ/k₀, where k₀ is the millimeter wave vacuum wave number, and ϕ is related to in-phase and quadrature signals by $I=Acosϕ$ and $Q=Asinϕ$⁠. The phase fluctuations are calculated from the AC component of the phase change obtained through IQ detection, where the in-phase (I) and quadrature (Q) components are used to compute the instantaneous phase as Φ(t) = tan⁻¹(Q/I). The AC component is extracted by removing the low-frequency DC offset from the computed phase. To generate the spectrogram, the phase signal is segmented into 2-ms overlapping records (75% overlap), mean-subtracted, and conditioned with a Hanning window before applying a Fast Fourier transform (FFT). The absolute value of the FFT provides the power spectrum of fluctuations. The final spectrogram is built from these spectra, with time dependence given by the central time of each record and is further smoothed using boxcar averaging over 10 kHz in frequency and 1 ms in time. Since the Doppler shift causes phase accumulation, phase runaway effects can introduce discontinuities at the start and end of each record, contributing to the spectral content. As noted in Holzhauer et al.,³⁴ this is a recognized feature in Doppler reflectometry. The application of a Hanning window helps mitigate these discontinuities, as discussed in Bendat and Piersol (Sec. 11.5.2.1).³⁵ Because of the poloidal aiming and a peak plasma frequency comparable to the operating frequency for this period, the DBS probe propagates above the magnetic axis. At 79.2 GHz, it makes the closest approach at ρ ≈ 0.4, where it probes density fluctuations with k_⊥ ≈ 9 cm⁻¹. Figure 7(b) shows the line average density, and the total neutral beam power, along with the contribution from a particular neutral beam (denoted “30L,” located 30° toroidally). Notably, the observed mode activity is destabilized exclusively when the “30L” is injecting, consistent with the wave-particle resonance with the beam ion that drives GAEs.³² Resonance between a beam ion and a wave depends on the wave frequency and wavenumber, as well as the beam ion velocity. Each beam sources a distribution with a narrow range of pitches, v_b‖/v_b, where v_b is the beam ion velocity and v_b‖ is the velocity component parallel to the magnetic field. The differing injection geometries of the various beams affect the pitch of the ions they source and thus the waves that the ions can potentially resonantly drive. As discussed in Ref. 32, the 30L beam sources a distribution likely to drive waves such as observed here. The frequencies of the modes vary over time within each period of 30L injection and from one period to the next. This is unsurprising, given the evolving plasma conditions, which affect the GAE dispersion relation. However, consistent with the global nature of GAEs, when the operating frequency steps, plasma conditions change little during the settling time and the same mode frequencies are observed before and after the step. This can be seen, for example, at t ≈ 3810 ms in Fig. 7, when the operating frequency steps from 79.2 to 80.4 GHz during a period of 30L injection.

An example of a discharge with helicon injection is shown in Fig. 8. This discharge is characterized by a plasma current (I_P) of 1 MA and a toroidal magnetic field (B_t) = 1.8 T; the focus of the frequency stepping tests is directed toward simultaneous helicon wave and turbulence measurements, as depicted in Fig. 8. This is a neutral beam-heated L-mode plasma. Measurements of density fluctuations in the helicon range of frequency are obtained during a period of constant plasma density with pulsed helicon injection spanning 1190–1260 ms. Approximately 800 kW helicon power (measured at klystron output), modulated ON and OFF with a 50% duty cycle at a rate of ∼194 Hz, is fed into the helicon antenna [Fig. 8(d)]. Two different Doppler backscattering (DBS) operating frequencies (f_DBS) are launched successively (both in X-mode) during the discharge. The received signal in the helicon range of frequencies is measured using the down-conversion receiver circuit, illustrated in Fig. 2. The high-frequency quadrature spectrum [Figs. 8(b) and 8(c)] shows modulation correlated with the modulation of the injected helicon power, with a sharp peak at the helicon frequency and broadband fluctuations appearing when the helicon power is ON and only background noise being present when the helicon power is OFF. The sharp peak, which is attributable to the pickup of stray RF radiation outside the plasma at the helicon source frequency by the measurement circuit (Fig. 2), is discussed below. Notably, the intensity of the broadband fluctuations decreases abruptly around t ≈ 1225 ms, mirroring a sudden drop in the intensity of the low-frequency turbulence [as depicted in Fig. 8(a)] simultaneously observed in the quadrature spectrum of the low-frequency signal from the DBS. This drop is attributed to a change in the probe frequency from 69 to 75 GHz within the DBS system occurring at that time. Ray tracing indicates that with the shift of the operating frequency at t ≈ 1225 ms, DBS shifts from measuring density fluctuations with k_⊥ = 2.6 cm⁻¹ in the outer plasma core at ρ = 0.7 to k_⊥ = 3.2 cm⁻¹ at the mid-radius of the plasma at ρ = 0.5.

A comparison of the broadband helicon fluctuations in Fig. 8(b) with similar fluctuations in Fig. 8(c) indicates that the spectrum of broadband fluctuations is not symmetric, leading to an explanation for the cause of the broadband fluctuations. A consistent frequency shift toward more negative values is observed in the broadband fluctuations in the positive and negative helicon frequency range that mirrors a similar shift in the turbulence spectrum. This pattern indicates that the high-frequency fluctuations are features of the spectrum of radiation scattered from turbulence rather than radiation scattered from helicon waves. They are sidebands caused by imaging of the spectral feature in the turbulence range of frequencies to frequencies around ±476 MHz. Two mechanisms are proposed in Ref. 16 for producing the sidebands. One is path length modulation by density or magnetic field strength perturbations associated with the helicon wave that affects both the probe beam and the probe beam radiation returning along the path to the DBS after backscattering from turbulence. The density and magnetic field strength perturbations perturb the index of refraction of the millimeter waves. This mechanism is not expected to play a role in this case since the path length modulation accumulates over the path of the millimeter wave radiation and cancellation occurs when the path crosses through full wavelengths of the perturbation to the index of refraction. The helicon waves have a relatively short wavelength compared to the extent of the region over which path length modulation is accumulated along the millimeter wave ray. From GENRAY ray tracing for the injected helicon wave, the helicon wavelength is predicted to be ∼4.7 cm when it passes through the DBS measurement region for this case. The extent of the region with significant helicon power along the millimeter wave ray is not well known, but the helicon antenna has a poloidal extent of 20 cm,²⁰ which sets a minimum.

It is expected in this case that the sidebands are caused by the second mechanism proposed in Ref. 14, a rapid modulation of the turbulence Doppler shift by the helicon wave E-field in the region where turbulence is measured. Electrons are magnetized for waves at the helicon frequency, and the extremely low-frequency turbulence cannot dynamically respond on the time scale of the helicon wave period, so the electron density fluctuations that scatter the millimeter wave probe beam are essentially frozen into the electron fluid and advected by an oscillating v_E×B resulting from the helicon E-field. (Ions are essentially unmagnetized, so they are not advected directly by E × B drift, but quasi-neutrality ensures ion density fluctuations are driven to match the advected electron density fluctuations.) The spectral analysis of Fig. 8 uses long time records (∼1/3 ms), so the rapid oscillation of the Doppler shift of the scattered spectrum manifests numerically as high-frequency sidebands, which are the images of the spectrum in the low-frequency range of turbulence.

A careful analysis of fluctuations in the helicon frequency range (Fig. 9) shows that the broadband fluctuations observed during helicon injection undergo an ∼75% reduction in mean power when the probe frequency increases from 69 to 75 GHz. Figure 9(a) shows the mean power fluctuations within the range 475–477 MHz, excluding a sharp peak as seen in Fig. 9(b) at ∼ 476 MHz and also excluding background noise. [Note that background noise is not subtracted from the spectra in Fig. 9(b).] This decrease occurs as the measurement location shifts away from the outer plasma to the plasma mid-radius even though the peak helicon power to the antenna during each pulse remains nearly constant throughout the period shown. The reduction in the high-frequency broadband is consistent with the interpretation of the broadband features as sidebands of the low-frequency turbulence spectrum since the simultaneously observed turbulence also undergoes a substantial reduction at the same time. However, helicon amplitude at the location of measurement may also change as the measurement location changes, which would play a role. The amplitude of the Doppler shift modulation is expected to affect the ratio of the power in the sideband to that in the turbulence. An analysis of these measurements to infer the helicon amplitude requires further development of a synthetic diagnostic model and is left to future work.

The sharp peak is excluded from the above analysis because it is due to pickup by the measurement circuit (Fig. 2) of stray RF radiation outside the plasma at the helicon source frequency. The contribution of pickup is confirmed by blocking the beam from entering the plasma for a similar discharge. The peak is still observed in the measured spectrum, even as the spectral power is reduced to noise levels throughout the rest of the spectrum. Note that the measured frequency of the peak, which can be seen in Fig. 9(b), is shown to be ∼476.1 MHz, slightly different from the true value of 476 MHz produced by the helicon source. This is an artifact of how the measured frequencies are adjusted to account for the downshifting in the measurement circuit using the 500 MHz crystal oscillator (Fig. 2). There is an uncorrected small temperature-related deviation of the crystal oscillator frequency from the rated value of 500 MHz. The mean power of the pickup peak (with background noise subtracted) is shown separately in Fig. 9(a). Notably, the mean power of the peak remains unchanged, independent of the external DBS launch frequency (f_DBS). In principle, a peak at the helicon source frequency could come from the millimeter wave radiation directly backscattered from the helicon wave injected into the plasma. However, the insensitivity of the power in the peak to measurement location, in this case, indicates that it is entirely attributable to RF pickup since a change in measurement location should also result in a change in any millimeter wave power scattered directly from the injected helicon wave.

The ECH launcher steering capability is illustrated in Fig. 10. A high-power, neutral beam-heated H-mode plasma (I_P = 1.1 MA and B_T = 2.02 T) is selected as the target discharge for the test. Using a fixed frequency of 73.8 GHz with X-mode polarization, a wavenumber scan is performed at a measurement location of ρ ≈ 0.97. The steerable mirror is programmed before the discharge to scan through a range of poloidal angles at a constant 0° toroidal angle to investigate low-k fluctuations, as illustrated in Fig. 10. The time-dependent quadrature spectrum of the low-frequency signal in the top panel [Fig. 10(a)] shows turbulent density fluctuations, while the requested and actual mirror steering angles (measured relative to normal incidence) are shown in the bottom panel, along with the line average density. Significant variations can be observed in the intensity and width of the spectrum as the mirror position is adjusted, particularly when the mirror is steered to probe very low k. Figure 10(b) shows the requested poloidal and actual poloidal steering angle of the mirror. The value shown is relative to the angle needed for normal incidence, which is determined from ray tracing. Density is constant throughout the period shown in the figure, as can be seen in Fig. 10(b). Two full scans spanning a nearly 8° range of poloidal angles (k_⊥ ∼ 3–7 cm⁻¹) are shown in the figure during the period t ≈ 1800–2550 ms, and the spectrum of fluctuations can be seen to exhibit a similar dependence on poloidal angle in each scan, suggesting that the turbulent spectrum is approximately stationary during this period. A wavenumber power spectrum for three different probed wavenumbers is plotted in Fig. 10(c) showing an often-observed trend of decreasing fluctuation power with increasing k.³⁶ This power spectrum has been calculated from the quadrature spectrum integrated over ±1 MHz frequency band. As can be seen, there is a delay of ∼40 ms between the request and the achieved mirror angle in the motor feedback system. This delay is attributed to a combination of communication latency and the inertia involved in initiating mirror movement from a static position.

New frequency stepping and real-time steering capabilities have been demonstrated in DIII-D plasmas for a DBS system that measures turbulence and RF waves simultaneously. The system has been integrated into a steerable ECH launcher and uses a programmable synthesizer-based millimeter-wave source. This integration addresses technical challenges and sets a model for future burning plasma facilities with ECH. The steering capability allows for probing plasma fluctuations across a wide range of wavenumbers within a single discharge. Frequency stepping tests revealed challenges in spatial distribution measurements, suggesting future work on synchronizing steering and frequency stepping to maintain constant wavenumber measurements. The system will also enhance DBS measurement capabilities for DIII-D, extending the capability for turbulence measurements with wave-number matching (i.e., k_‖ ∼ 0 at cutoff) to high k (k_⊥ > 15 cm⁻¹)³⁷ allowing for poloidal correlation studies with other millimeter wave scattering systems that are poloidally separated but at the same toroidal location.³⁸ In addition, results presented here reveal broadband helicon fluctuations caused by a mechanism different from that previously reported in Ref. 14. High-power helicon wave measurements showed high-frequency broadband features in the density fluctuation spectrum near the helicon frequency, believed to be sidebands of turbulence modulated by the helicon wave.

This work was supported by U.S. DoE Grant Nos. DE-FC02-04ER54698, DE-SC0020649, and DE-SC0020337. The authors would like to thank Larry Bradley and the DIII-D Diagnostic team for their engineering as well as technical support during the installation of the DBS diagnostic setup. This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

The authors have no conflicts to disclose.

S. Chowdhury: Conceptualization (equal); Data curation (lead); Formal analysis (equal); Methodology (equal); Validation (equal); Visualization (equal); Writing – original draft (lead); Writing – review & editing (equal). N. A. Crocker: Conceptualization (lead); Formal analysis (equal); Funding acquisition (lead); Investigation (equal); Methodology (equal); Project administration (lead); Resources (lead); Software (lead); Supervision (lead); Visualization (equal); Writing – review & editing (equal). W. A. Peebles: Conceptualization (equal); Formal analysis (equal); Investigation (equal); Supervision (equal); Visualization (equal); Writing – review & editing (equal). R. Lantsov: Visualization (equal). T. L. Rhodes: Supervision (supporting); Visualization (supporting); Writing – review & editing (equal). L. Zeng: Visualization (supporting). B. Van Compernolle: Writing – review & editing (supporting). S. Tang: Resources (supporting). R. I. Pinsker: Resources (supporting). A. C. Torrezan: Data curation (supporting). J. Squire: Resources (supporting). R. Rupani: Project administration (equal); Resources (equal). R. O’Neilll: Resources (supporting). M. Cengher: Resources (supporting).

The data that support the findings of this study are available from the corresponding author upon reasonable request.