Development of multi-band and high-speed visible endoscope diagnostic on EAST with catadioptric optics

1.   Introduction

Visible light diagnostic is a crucial instrument for monitoring the plasma position, the plasma surface interaction, and especially for the Edge Localized Mode (ELM) eruption process in H-mode plasmas [1, 2]. It is indispensable for researching ELM-filament, which is a common and serious problem in magnetic confinement fusion research. ELM-filaments could cause severe heat and particle localized deposition, leading to material erosion for both the first wall and divertor target plates, impurity sputtering, and high particles recycling, which are particularly serious in International Thermonuclear Experimental Reactor (ITER) [3, 4]. A previous study showed that the pedestal temperature of ITER can be as high as 4 keV. For example, for a 10 MA plasma discharge in ITER, the energy loss due to a single ELM event can be up to 20 MJ [5], and it is far beyond the bearing capacity of the materials in the first wall.

A variety of visible light diagnostics for tokamak have been developed in the past 30 years [6–9]. In the Experimental Advanced Superconducting Tokamak (EAST), the early visible light systems consist mainly of tangential stainless tubes [1]. An industrial camera was fixed at the end of the tube. The disadvantage of this method is that the camera was subjected to strong electromagnetic interference and high-level radiation from high-energy particles and x-rays, which led to a breakdown or even damage to the camera.

Furthermore, high-speed cameras require more space, so transmitting the light from the inside of the device vacuum chamber to the outside of the device is a practical solution. A wide-angle visible light system in visible range has been developed, while this system only can cover one band. This means that the system can only monitor the trajectory of a kind of particle at once. In order to monitor and thus compare the images of different bands and the trajectories of different impurities simultaneously, a more advanced system needs to be designed and developed. To achieve this goal, a new multi-band and high-speed visible light endoscope diagnostic has been developed and successfully tested on the EAST tokamak since 2021.

The results of the experimental test shown in the following contents indicate that this multi-band visible endoscope diagnostic with the high-speed camera of NAC ACS-1 M60/pco.dimax basically could meet its design requirement. This multi-band and high-speed visible light diagnostic will play an important role in image monitoring of different impurities and the identification of plasma boundary position to better achieve plasma feedback control.

2.   Design requirements

The visible light diagnostic is designed for the identification and feedback control of the tokamak plasma boundary, the transport law of plasma impurity, and other related transient processes, such as plasma breakdown and dust plasma. Configuration reconstruction based on electromagnetic measurement diagnosis is difficult to adapt to the harsh working environment of the vacuum chamber in future fusion reactor. Therefore, it is necessary to look for alternative diagnostic methods for the reconstruction of plasma boundary. The most important application of the EAST visible light diagnostic is to identify and control the plasma boundary through visible camera, and provide a viable control scheme for future fusion devices. By comparing the difference in visible light radiated by different impurities at the boundary, the most suitable identification spectral line can be found to obtain the most ideal plasma boundary position. Therefore, the optical system is required to include different bands, covering the entire plasma cross-section and having a high spatial resolution. Combining the above factors and requirements, the endoscope is designed to cover the spectral range from 380 to 750 nm with a high spatial resolution, which is < 5 mm near the object distance. The resolution here refers to the spatial resolution corresponding to each pixel. The endoscope should be coupled to a set of three visible light cameras to monitor three different bands (corresponding to different impurities) at the same time. In addition, the system should be reasonably protected against the coating of the first mirror from wall cleaning and/or lithiation.

The filter is installed by plugging, which can be replaced manually without affecting the imaging quality of the optical path. The vacuum glass shall be easily replaced without dismantling the optical path. In case of accident, it can also be sealed by connecting the manual vacuum baffle valve with the standby KF40 vacuum flange to prevent vacuum leakage of the device. The position constraint relationship of each flange of the optical path is shown in figure 1 to ensure that the endoscope system can be built into the device smoothly. After the optical path being installed in place, the baffle connecting rod is located below the lens barrel. When standing outside of the window and looking in the window, the cylinder rotates clockwise to open the baffle, and the reverse rotation can close the baffle. The maximum rotatable angle of the baffle is 90°. The camera support connection interface flange shall be reserved in the optical path. The camera support shall be firmly connected with the optical path. The support can be adjusted in XYZ three-direction to ensure that the center of the camera sensor can be aligned with the center of the image plane of the optical path.

Figure  1.  The diagram and sectional view of window interface J where the diagnostic locates.

DownLoad: Full-Size Img PowerPoint

3.   Field of view (FOV) simulation

Coordinate system needs to be established for FOV simulation analysis. Taking the center of the EAST device as the origin of the coordinate system, as shown in figure 2, X direction passes through the origin and is perpendicular to the large flange surface of the window J, and Z direction is perpendicular to the horizontal plane and points to the top of EAST tokamak.

Figure  2.  The definition of datum coordinate system.

DownLoad: Full-Size Img PowerPoint

As shown in figure 3, the basic parameters for FOV simulating include viewpoint P, observation direction PC, horizontal FOV ∠B₁PB₂ and vertical direction FOV ∠D₁PD₂. It should be noted that PC is perpendicular to B₁B₂ and D₁D₂, and B₁B₂ parallels to the horizontal plane. The observation direction PC is defined by the azimuth (θ, φ) of the unit vector n on line CP. Figure 4 shows the specific parameter values of the corresponding horizontal FOV.

Figure  3.  Definition of basic parameters of field of view (left) and optical axis direction (right).

DownLoad: Full-Size Img PowerPoint

Figure  4.  Distribution of horizontal field of view in vacuum chamber (left) and vertical field of view (right).

DownLoad: Full-Size Img PowerPoint

The first wall of EAST can be approximately obtained by rotating the polar section of the first wall around the Z-axis, as shown in figure 5. The last closed magnetic surface of the plasma is calculated by EFIT, which is also a rotation surface around the Z-axis. These two rotation surfaces can be drawn by the RevolutionPlot3D function of Mathematica (version 11.3). With the two graphic display options of ViewVector and ViewAngle, the observation effect of the light path along the PC direction at the viewpoint P can be simulated. In order that the horizontal FOV and the vertical FOV can be designed independently, the ImageCrop function is further chosen to cut the 3D graphics obtained by the RevolutionPlot3D function, to obtain the final simulation effect.

Figure  5.  Polar section model for field of view simulation.

DownLoad: Full-Size Img PowerPoint

The viewpoint P and the observation direction $\mathit{n}$ are controlled by the ViewVector, and the size of the FOV is adjusted by the scale of image clipping in the ImageCrop function. The above scale is determined by the sizes of the FOV in the vertical direction and the horizontal direction, respectively corresponding to $\angle {\mathrm{D}}_1{\mathrm{PD}}_2$ and $\angle {\mathrm{B}}_1{\mathrm{PB}}_2.$ The blind spot range is affected by the size of the small hole of the front reflector. The small hole is on the main axis and focus of the ellipsoid. The beam near the axial FOV (parallel to the optical axis of the ellipsoidal mirror), after being reflected by the ellipsoidal mirror, falls on the central opening position (entrance pupil position) of the front plane reflector, and cannot be reflected into the rear end optical path for imaging. Therefore, objects within this range cannot be imaged, that is, a certain range of blind spots are formed on the phase machine target surface. To further optimize the position of the blind spot, $\angle {\mathrm{B}}_1\mathrm{PC}$ and $\angle {\mathrm{B}}_2\mathrm{PC}$ are further controlled independently to ensure that the blind spot is outside the last closed magnetic surface.

Based on the above design requirements and basic parameters for FOV, the FOV simulation diagram is obtained, as shown in figure 6. The FOV can cover the whole section of small circular, and the blind spot is located outside the plasma, which provides an important means for boundary identification and comparative study of the spatial distribution of different impurities.

Figure  6.  Field of view simulation diagram based on the model above.

DownLoad: Full-Size Img PowerPoint

Through FOV simulation and analysis, the parameters of the optical system are designed as follows: horizontal FOV (H): $\angle {\mathrm{B}}_1\mathrm{PC}=6°,$$\angle {\mathrm{B}}_2\mathrm{PC}=39°;$ vertical FOV (V): $\angle {\mathrm{D}}_1\mathrm{PC}=\angle {\mathrm{D}}_2\mathrm{PC}=32°;$ observation direction (azimuth of $\mathit{n}$): $\theta =90°,\phi =-65°;$ object distance: $u=1750\mathrm{mm};$ object depth of field: $800–4800\mathrm{mm};$ relative aperture: $D/f=1:4;$ spatial resolution: $\le 5\mathrm{mm}$ at object distance of 1750 mm.

The final performance parameters of the optical path shall be compatible with the alternative cameras given in table 1, mainly involving the size of the image plane and spatial resolution. It is required that the size of the image plane is not larger than the sensor size of NAC ACS-1 M60, and the optimal spatial resolution is designed to be less than 5 mm at object distance $u=1750\mathrm{mm}.$ The approximate calculation formula of spatial resolution is $\mathrm{∆}=\frac{u}{f}\delta ,$ where $f=16\mathrm{mm}$ is the focal length of the optical path, $\delta =22\mu \mathrm{m}$ is the pixel size of NAC ACS-1 M60 camera, thus the $\mathrm{∆}=2.4\mathrm{mm}\mathrm{at}u=1750\mathrm{mm},$ which is less than 5 mm. At the maximum depth of field of $u=4800\mathrm{mm},$ the corresponding resolution $\mathrm{∆}\approx 6.6\mathrm{mm}$ with NAC ACS-1 M60 and $\mathrm{∆}\approx 3.3\mathrm{mm}$ with pco.dimax.

Table  1.  Main technical parameters of visible camera for optical design.

Camera	Resolutions	Pixel size ($\mu {\mathrm{m}}^2$)	Frame rate (fps)
pco.dimax	$2016\times 2016$	$11\times 11$	1279
NAC ACS-1 M60	$1280\times 896$	$22\times 22$	54 000

 | Show Table

DownLoad: CSV

4.   Optical design

Fully superconducting tokamak EAST is the first fusion device with modern divertor configurations in the world [10–12]. The major radius and minor radius of the EAST device are 1.7 m and 0.4 m respectively. According to system requirements, the FOV of the endoscope is designed to be $45°$ and $64°$ in the horizontal direction and vertical direction, respectively. Consequently, the FOV is large enough to cover the entire cross-section of the EAST device, which includes the inner and outer walls, the upper and lower divertors.

4.1   System structure scheme

The overall structure of the optical system is shown in figure 7. In the optical path, the first one is the front-end reflection system, which includes a parabolic mirror and a plane mirror. After passing through the reflection system, the optical axis rotated 180°–65°=115°, as shown in the partial enlarged detail of the optics header in figure 7 (the header in this figure should be turned $90°$ on the right side); and then, after passing through a collimating lens made of fused quartz material, the beam is collimated into parallel light; next it pass through a vacuum glass of isolating window which is also made of fused quartz. Then there are two color separation plates, SP1 reflection: B region (380–500 nm), and transmission: G and R region (500–750 nm); SP2 reflection: G region (500–580 nm), and transmission: R region (580–750 nm).

Figure  7.  The Overall structure of the optical system (the header in this figure should be turned $90°$ on the right side). The vacuum glass: the relay lens, which concentrate the light collected by the head so that the optical path can be elongated to the required position. SP1 and SP2: chromatic beam splitter, that reflects the blue band (380–500 nm) and the green band (500–580 nm) into the corresponding spectral optical path respectively and allow the light of other bands to pass through and transmit to the next optical path. The filter in blue/green/red channel: usually a narrow-band wave plate, that transmits the visible light of a specific spectral line, such as C Ⅲ, He Ⅱ, Li Ⅱ, and ${\mathrm{D}}_{\alpha }.$

DownLoad: Full-Size Img PowerPoint

The main mirror is a paraboloid mirror, R= -133.533 mm, k=-1; luminous aperture: 86×64 mm², off-axis: 20.1406 mm. The size of the plane mirror is 82×68 mm². A small irregular cone is opened on the plane mirror. The results show that the aperture of the system is as small as 4 mm. In order to ensure a certain margin, the diameter of the small hole is set as 5.2 mm. According to the hole diameter of 5.2 mm, the FOV occupied by the blind spot is about 4.3°.

The turning angle of the plane mirror is determined according to the installation position and observation position of the lens. The turning angle of the plane mirror in the figure is 32.5°. After passing through the reflection system, the optical axis rotated 180°–65°=115°. In the front-end reflection system, the distance between the paraboloid mirror and the plane mirror is 70 mm. The distance between the small hole (plane mirror) and the collimating lens is 1277.238 mm; the luminous aperture of the collimating lens is 83.5 mm. After passing through the collimating lens, the beam is a parallel light. The distance between the small hole and the sealing window is 1664.351 mm; the diameter of the sealing window is 72.5 mm, the thickness is 12 mm, and the material is fused quartz. After passing through the sealed window, it is the color splitter for light splitting and the imaging mirror group of B, G and R bands. The optical structure of the three mirror groups is the same, the length of the lens barrel (front film to rear film) is 617.801 mm, and the focal length is about 182.99 mm. The structures of B, G and R bands are shown in figure 7.

4.2   Image quality analysis

The main technical specification data of the NAC ACS-1 M60 for visible light diagnostic applications are as follows. The resolution (H$\times$V) and pixel size are 1280 $\times$ 896 pixels and 22 $\times$ 22 $\mu {\mathrm{m}}^2$ respectively. The Nyquist frequency is 22 c mm^-1 (cycle per mm) and the dynamic depth of field is 8/10/12 bit. In addition, the spectral range is from 290 to 1100 nm, and the electronic shutter is 1.1 $\mu \mathrm{s}.$

Since the Nyquist frequency of the camera NAC ACS-1 M60 is 22 c mm^-1, the 22 c mm^-1 system transfer function is investigated for the MTF curve [13–15]. The MTF of the designed optical system for each band are analyzed with ZEMAX and shown in figures 8–10.

Figure  8.  The MTF of the designed optical system of B band.

DownLoad: Full-Size Img PowerPoint

Figure  9.  The MTF of the designed optical system of G band.

DownLoad: Full-Size Img PowerPoint

Figure  10.  The MTF of the designed optical system of R band.

DownLoad: Full-Size Img PowerPoint

The maximum distortion of the system is 22.5% and the relative illuminance of the system for 0.55 μm is shown in figure 11.

Figure  11.  Relative illuminance in horizontal (a) and vertical directions (b) of the system.

DownLoad: Full-Size Img PowerPoint

4.3   Depth of field analysis

The designed object distance of the system is 1750 mm. At 800 and 4800 mm object distances, the B band MTF curves are shown in figures 12 and 13 separately.

Figure  12.  B band MTF curve with object distance of 800 mm.

DownLoad: Full-Size Img PowerPoint

Figure  13.  B band MTF curve with object distance of 4800 mm.

DownLoad: Full-Size Img PowerPoint

4.4   Analysis of dichroic mirror size and offset

45° dichroic mirror SP1: the footprint of dichroic mirror SP1 is shown in figure 14 below. It can be seen that the luminous aperture is oval and the size is 70 mm × 100 mm.

Figure  14.  The footprint of dichroic mirror SP1.

DownLoad: Full-Size Img PowerPoint

45° dichroic mirror SP2: the footprint of dichroic mirror SP2 is shown in figure 15. It can be seen that the luminous aperture is oval and the size is 77 mm × 103 mm.

Figure  15.  The footprint of dichroic mirror SP2.

DownLoad: Full-Size Img PowerPoint

After the light passes through two 45° color separation plates, it will be offset. Based on the fused quartz material and thickness of 10 mm, the offset dt=3.15 mm.

5.   Test and results in EAST tokamak

The optical system was installed and tested on the EAST tokamak device during the April 2021 experimental campaign. There are only two NAC ACS-1 M60 cameras in hand. They were employed to image the optical red and green channels for the testing, and the blue channel was mounted with a pco.dimax high-speed camera instead of M60. The pixels size of pco.dimax is half of M60, i.e. $11\times 11\mu {\mathrm{m}}^2.$ It should be noted that the maximum frame rates of pco.dimax and NAC M60 are 1 kfps and 120 kfps, respectively.

Based on the dichroic mirrors, the system is divided into three imaging mirror groups with the same structure and different bands, which are B (380–500 nm), G (500–580 nm) and R (580–750 nm) bands. The different wave bands correspond to different impurities radiation. On EAST device, R band corresponds to D_α, G band corresponds to Li Ⅱ and B band corresponds to C Ⅲ and He Ⅱ primarily. It can be seen from the figure 16 that there are significant differences on the low field side of different wave bands.

Figure  16.  Visible image of the optical system for three bands: (a) blue channel (b) green channel (c) red channel.

DownLoad: Full-Size Img PowerPoint

Figure 16 shows a visible transient image of shot #111 798 from 2021 campaign. For the image acquisition, the region of interest is 1200×1704 for blue channel with a frame rate of 2 kfps and 592×848 for green and red channels with frame rates of 10 kfps, respectively. The exposure times for blue, green and red channels are 80 μs, 15 μs, and 10 μs, respectively. It should be noted that the red channel is equipped with a filter of ${\mathrm{D}}_{\alpha }$ with a diameter of 50 mm, FWMH of 1.5 nm, and a central wavelength of 656.3 nm. For the temporary test, the above filter diameter is much smaller than the actual required diameter of 86 mm. From the comparison between figures 6 and 16, it can be seen intuitively that the imaging effect of the final closed magnetic surface given by the FOV simulation is basically consistent with the plasma image by the camera. In order to simplify the calculation, the effects of optical image distortion are not considered in the simulation analysis. In addition, it can be seen that the blind spot of the FOV given by the simulation analysis is completely outside the last magnetic plane, but the actual result is that the plasma boundary is very close to the blind spot. One possible reason is that the optical head lens does not meet the required installation accuracy. More detailed analysis results will be presented in the future. The tests on EAST confirm that the visible endoscope system with the high-speed camera is available for monitoring the transient events of the ELMs crush during H-mode operation.

In fact, various interesting and complex phenomena will be observed using this high-speed camera system, such as the spatial distribution and evolution of different impurities in the same FOV. Another phenomenon is the identification of plasma boundary. It should be noted that although the boundary on the low field side is not very clear to our naked eyes sometimes, the outermost boundary has been sufficiently identified through image recognition and machine learning. We have compared it with the EFIT data at the same time and found a good agreement which will be presented detailed in a separate paper in the future. The optical system will be used to contrastively study the spatial distribution of different impurities and related physics, including the recognition of plasma boundary.

6.   Conclusions

The new high-speed and multi-band visible endoscope diagnostic has been implemented on the EAST tokamak to contrastively investigate the spatial distribution and evolution of different impurity spectrums in the same wide-angle FOV. The FOV of the optical system includes the inner and outer walls, the upper and lower divertors of EAST. The spatial resolution is less than 5 mm near the object distance $u=1750\mathrm{mm}$ with the camera NAC ACS-1 M60. The experimental results confirm that it can be applied to the recognition of plasma boundary and related physical research. The method and results of the recognition of plasma boundary basing on the data from this diagnostic will be given in detail in another paper in the future.