The AXUV Bolometer and L Camera System on TCV
1.Diagnostic Description
1.1.Overview:
Twin camera systems for measuring the total radiated power and the Lyman-alpha emission are being installed on TCV. Each system consists of an array of pin-hole cameras that view the plasma from multiple angles on a single sector in order to allow tomographic reconstruction of the poloidal 2D emission profile. A total of 7 units containing one camera for bolometry and one for L emission are being installed on the top and bottom inner and outer ports, and the upper, middle and lower side ports of sector 3. The cameras in each unit are identical in geometry, with the exception that the L camera has a VUV absorption filter placed behind the pin-hole aperture. Incident radiation will be detected in each camera using a 20-channel AXUV photodiode linear array. Transimpedance pre-amplifiers installed within the camera housing will amplify the photocurrents from the diodes. In order to fully exploit the advantages of using AXUV photodiodes, a bandwidth as close to 100kHz as practicable is incorporated in the pre-amplifier design. The resulting double ended voltage signals are carried by shielded cable to a DTACQ computer for signal conditioning and digital acquisition. The digitised raw data is to be stored locally on the DTACQ computer for a given period, and a sub-sampled, filtered data set is to be recorded on the MDS database.
1.2.Diagnostic Hardware
1.2.1.AXUV-20EL Photodiode Arrays.
These photodiode arrays, made by International Radiation Detectors, feature an almost flat spectral response in photocurrent per watt in incident power, from photon energies of 10 eV to 6 keV and have a dynamic range of 7 orders of magnitude in photocurrent. These properties, plus the fact that they are insensitive to microwave radiation, make them suitable to use as bolometers on TCV.
Table 1: AXUV-20EL array properties
Photodiodes in array / 20Size of active area / 0.75 x 4.0 mm
Shunt Resistance / 300 M
Capacitance / 1 nF
Rise Time (10% to 90%) / 0.2 s
1.2.2.122-XN L Filters
The VUV absorption filter used for the L camera diagnostic is produced by Acton Research Corporation. An example of a spectral transmittance curve is shown in Figure 1a). Given the wide viewing angles of the camera fans, an important issue is the variation in transmittance with the angle of incidence. A calibration of the angle of incidence effect carried out by Acton Optics is shown in Figure 1b). General filter properties are summarized in Table 2.
Table 2: 122-XN filter properties
Thickness / 2.0 mmDiameter / 12.7 mm (0.5 in)
Peak wavelength / 120.0 nm
Bandwidth / 8.3 nm
Peak Transmittance / ~ 6.5 %
Substrate / MgF2
Table 3: Calibration values for 122-XN filters
Filter # / Peak transmittance [%] / Peak wavelength [nm] / FWHM1 / 6.7 / 118.5 / 8.8
2 / 7.2 / 118.5 / 8.8
8 / 8.6 / 119 / 8.8
4 / 7.0 / 118.5 / 8.6
5 / 8.4 / 118.5 / 8.9
6 / 8.2 / 119 / 8.8
7 / 8.5 / 119 / 8.9
Insert spectral transmittance curve here!
Figure 1: a) The variation of trasmittance with wavelength (viewed at 0 degrees). The variation of the peak transmittance (green) and the wavelength at the peak (blue) with angle of incidence ( [degrees]) for the 1 inch witness sample VUV filter calibrated by Acton Optics.
1.3.Mechanical
1.3.1.Camera hardware allocation
The cameras are installed on the inner and outer vertically viewing ports on the top and bottom of TCV sector 3 and the top, middle and lower horizontally viewing ports on the side of TCV sector 3. The designated camera numbers, port location and allocated hardware components are summarized in Table 4.
Table 4: Allocation of hardware components and ports
Position / Port / Camera # / Pre-amp card # / L filter #Bolo / L
1:10 / 11:20 / 1:10 / 11:20
Inner top / T03A_1 / 1 / L01 / L08 / H07 / H01 / 1
Outer top / T03A_3 / 2 / L02 / L09 / H08 / H02 / 2
Top lateral / E03A_1 / 3 / L10 / L03 / H09 / H03 / 8
Middle lateral / E03A_2 / 4 / L11 / L04 / H10 / H04 / 4
Bottom lateral / E03A_3 / 5 / L12 / L05 / H11 / H05 / 5
Outer bottom / T03A_3 / 6 / L06 / L13 / H12 / H06 / 6
Inner bottom / T03A_1 / 7 / L07 / L14 / H14 / H07 / 7
1.3.2.Camera Mechanical Design
Each camera houses two AXUV-20EL photodiodes. These are mounted on a 22pin DIP feed through which is embedded (and vacuum sealed) on a vacuum flange, such that the photodiode arrays are aligned in the poloidal direction. Slot apertures in the toroidal direction are positioned at a prescribed distance in front of each diode facing the plasma. The positions of the photodiode centres and aperture parameters in the poloidal plane for each camera may be found in the table below. These parameters are identical for the L photodiode and the bolometer photodiode in each camera. The photodiode for L has a VUV filter mounted in front of it. On the outer side of the vacuum flange, 2 pre amplifier cards with 20 channels each connect to the feedthroughs from the diodes. These cards are isolated from the torus earth. For all cameras, the photodiodes are positioned with their centres aligned on axis (i.e. along a common diagonal of the camera cross section), although the slot apertures of each camera are generally not aligned on axis (depending on the camera, according to the last column in the table).
1.4.Electronics: Pre-amplifiers
A total of 40 transimpedance amplifiers per unit (i.e. 20 per camera) are mounted on 4 boards within the camera housing, which will plug directly into the vacuum feedthrough pins. The space available for these boards is limited (a cylindrical volume 6cm high and 9cm diameter) by space constraints for the installation of these cameras on TCV – especially the inner top and bottom vertically viewing cameras. The amplifiers on these boards are supplied with +/-15V DC.
The anode pin for each consecutive photodiode in the array is situated on alternate sides of the AXUV chip, and one pre-amplifier card directly connects to the 10 channels on each side. Hence the pre-amplifier channel number and the photodiode position number are related as indicated in Table 5.
Table 5: Mapping from Pre-amp channels to Photodiode channels for bolometers and L cameras
Pre-amp / 1 / 2 / 3 / 4 / 5 / 6 / 7 / 8 / 9 / 10 / 11 / 12 / 13 / 14 / 15 / 16 / 17 / 18 / 19 / 20Diode / 1 / 20 / 2 / 19 / 3 / 18 / 4 / 17 / 5 / 16 / 6 / 15 / 7 / 14 / 8 / 13 / 9 / 12 / 10 / 11
One of the key features of this diagnostic will be high temporal resolution enabled by the use of AXUV photodiodes, which have a 10-90% rise time as low as 0.2s. In order to fully exploit this capability, the bandwidth of the signal amplification and conditioning electronics is kept as close to 100 kHz as possible. That said, the attainable bandwidth is constrained by the finite gain-bandwidth product attainable by current technology.
+++ pre-amp frequency response and calibration here! (P. Lavanchy +++
Figure 2 shows an example of the estimated photocurrents in each channel for a SNL diverted discharge (#17843) for the bolometer and Lsystems. These values are estimated by scaling the L line integrated emissivity from a b2eirene simulation with experimental measurements using a prototype AXUV bolometer and L camera. The important points to note are the large dynamic range from channel to channel, and the difference by over 2 orders of magnitude between the expected signal levels for the two systems.
Figure 2: Example of estimated photocurrents for each channel of the bolometer and L camera systems.
2.Camera Geometry Calibration
2.1.Motivation
During the design phase for this diagnostic (described in the fiche de projet), so called “phantom” datasets simulating 2D profiles of plasma emission were used to generate simulated chord measurements. Tomographic reconstructions from these measurements were then compared with the original 2D profiles in an effort to optimise the camera configuration. The susceptibility of the quality of the reconstructed image to systematic errors in the measurements produced by chord misalignments was evaluated for various camera designs. The figure of merit used to estimate the quality of the reconstruction was the reduced 2 evaluated between the original and reconstructed data. Chord misalignments were introduced by adding random values to the camera parameters of a given maximum amplitude with respect to the camera tolerances. The value of 2 was noted for a series of reconstructions of the same phantom dataset in which this amplitude was scanned.
Figure 3: Degradation in the reconstructed image quality as the level of chord misalignment produced by systematic errors in the camera design parameters are increased. The four cases shown correspond to scenarios with either 7 or 9 cameras and photodiode arrays containing 16 or 20 channels.
As shown in Figure 3, the quality of the reconstructed image for each camera configuration begins to increase dramatically as the tolerance values are approached. It was therefore considered necessary to calibrate the camera geometry ex-situ in order to minimise this source of systematic error.
2.2.Experimental Set Up
It was considered that the most straightforward way to calibrate the camera geometry was to take measurements of an effective point source of light as its position was scanned in front of the cameras. This was done by mounting the camera onto a bracket and using a precisely controlled 2D translation stage to scan the position of a light source in front of the camera aperture, as illustrated in Figure 4. The light source consisted of a 1mm diameter pin-hole attached to a light sphere that was illuminated by a 100 watt halogen lamp. The light sphere and pin-hole were securely mounted onto the translation stage. The halogen lamp was powered by a DC power supply, and the light output was monitored by a witness photodiode that viewed the interior of the light sphere via an optical fibre bundle. A chopper was placed in front of the pin-hole to enable noise reduction by lock-in amplification using a reference TTL signal from the chopper.
Measurements were made by incrementing the horizontal position of the light source in millimetre steps, pausing for about 1 second each step. The light received by each photodiode channel was recorded continuously at 1kHz for about 15 minutes as the light source was scanned, using a DT100 acquisition computer. For this purpose, two special cables were made with 37 way D-type connections on one end and 25 way D-type connections on the other. A modification to one of the cables was made in order to acquire extra signals needed for the calibration. These consisted of the chopper reference signal, the witness signal of the light source and another TTL signal that indicated intervals when the translation stage was stationary during its scan.
The acquired raw data from each experiment is saved onto CD-ROM for each camera as a series of Matlab binary files (.mat files). The file name conventions are as follows. Each file has the prefix “x_scan”, followed by the numbers “12” (i.e. pre-amp cards 1 and 2, for the bolometer array) or “34” (i.e. pre-amp cards 3 and 4, for the lyman-alpha array), followed by the letter “i” (if the camera was oriented upside-down), and / or the letter “n” (if the slit aperture was removed). For each camera, the light source was scanned from left to right (as viewed by the camera) for the bolometers and right to left for the L cameras. This means that x increases with time in the files with “12” or “34i” and decreases with time in the files with “34” or “12i”. The set up of the translation stage and camera position are summarised in the following table.
Camera # / Optical axis [mm] / Perp. distance to slit [mm] / Camera inverted? / Scan range [mm] / No. of points in scan1 / 198.8 / 156.7 / No / 0 - 280 / 401
2 / 198.8 / 156.7 / Yes / 0 - 280 / 401
3 / 198.0 / 112.9 / No / 0 - 400 / 401
4 / 198.0 / 112.9 / No / 0 - 400 / 401
5 / 198.0 / 112.9 / No / 0 - 400 / 401
6 / 198.8 / 156.7 / No / 0 - 280 / 401
7 / 198.8 / 156.7 / Yes / 0 - 280 / 401
Figure 4: Photograph of the experimental set up showing the translation stage, point light source and camera mounted onto a bracket.
2.2.1.Precise camera and translation stage alignment.
The axis of the camera flange was used as a reference line defining the z-axis, such that the horizontal and vertical rails of the translation stage define the x and y axes. A blank flange was mounted onto the bracket port, and the orientation of the bracket was adjusted to ensure that the perpendicular distance between the flange and the x-y plane remained constant to within 0.02mm (using a compression micrometer gauge). It was found that this was not reproducible using copper gaskets for the CF 100 and cf 150 ports, so rubber o-rings were used instead in order to enable metal to metal contact between the flange face and the port. Reproducibility to within 0.02 mm was then obtained.
Measuring the height of the translation stage above the optical table as the x-position was scanned enabled the horizontal alignment of the x-axis rail to be checked. It was found that variations in the height remained below +/- 0.15 mm across a 400 mm horizontal scan. This variation appeared to be the result of a slight dip in the table surface (between 150 and 300 mm), rather than a misalignment of the axis.
2.2.2.Reproducibility of the translation stage programmed position.
In order to check the reproducibility of the translation stage programmed position in the x direction, the translation stage was moved repeatedly between two programmed positions (x1 and x2). A reference post was set up close to x1 so that variations in the position each time the stage returned could be measured using the compression micrometer gauge. Over a total of 60 measurements in which the translation stage was reset three times, and two different values of x1 and x2 used, the standard deviation of measured variations in x1 was less than 5m.
2.2.3.Establishment of the optical axis
The blank flange had a diode laser mounted on axis, however the direction of the laser beam was not exactly along the axis of the flange, nor was the orientation of the laser adjustable. However, rotating the flange in its port mounting marked out a circle on the x-y plane. The centre of this circle is the intersection of the flange axis with the x-y plane, to be used as the point (0,0) in all measurements.
In practice, rotation of the flange and measurement of the laser dot position was carried out in a series of discrete steps. As mentioned previously, a photodiode views the inside of the light sphere via an optical fibre. This enabled the pin-hole set up to be used as a position detector, since the interior of the light sphere would be illuminated when pin-hole was moved to intersect the laser beam. For each measurement the position of the pin-hole was coarsely adjusted to the laser dot position, and then a series of measurements were taken on a 3mm x 3mm grid in order to obtain the position of the centre of the laser dot more accurately. In principle three measurements on a circle are necessary to find it’s centre. Taking eight measurements resulted in 56 combinations of three points, and hence 56 estimations of the circle centre from which the mean and standard error were taken.
Figure 5: Intensity measurements for the optical axis for the CF100 and CF150 flanges (left and right resp.)
Once the optical axis was established, the perpendicular distance from the centre of the flange and the light source was measured. To do this, the laser diode was removed and a brass rod with a similar diameter to the laser and a 3mm long, 0.6mm diameter cylindrical tip machined on axis at one end was inserted through the centre of the blank flange. The light source was moved to the centre position, and the tip of the rod was inserted into the 1mm diameter aperture. The position of the flange on the rod was recorded using a cylindrical sleeve, and the distance from the end of the rod to the sleeve was measured using a micrometer.
2.2.4.Anisotropy of the light source
Since the viewing angle of the pin-hole changes as it is moved, the light emitted through the source should be as isotropic as possible, so that variations in the measured source intensity can be attributed to the optical path length and the projected angles of the source and detector. The light sphere was attached to the pin-hole for this purpose. However, it was found that the light sphere only produced “approximately” isotropic illumination, and only up to angles of +/-45 degrees, as shown in Figure 5a). This figure was produced by removing the camera slit aperture, leaving the diodes open to view the light source from all angles as it was scanned. The light source was chopped to isolate its contribution from the background. A reference signal produced by the chopper was also acquired, and a numerical phase coherent average with respect to this signal was taken, as discussed in the next subsection. The traces shown are produced by dividing the measured intensity of the light source by cos()4, where is the angle to the light source from each diode centre. This is done in order to account for the change in optical path length and projected angles of the source and detector surfaces with . If the light source really is isotropic, each trace should be uniform (the variation of the profiles from trace to trace is not understood!). As can be seen, apart from the sudden drop in intensity at 45 degrees, there is another drop between +/-3 degrees. The light sphere was inspected, and it was found that a small port directly opposite the pin-hole aperture is clearly visible inside the sphere. The central drop in intensity is presumably the image of this in homogeneity in the light sphere surface. While it was clear that the source anisotropy had to be taken into account in the calibration, it was decided that a smooth variation in anisotropy was preferable to the discovered abrupt variations. Therefore a piece of semi-transparent tape was placed over the pin-hole to blur the image of the light source. The effect of the tape is shown in Figure 5b). In order to account for this anisotropy profile of the light source, a scan had to be made for each camera with the slit removed before making the scan with the slit installed. The measurements with the slit in place were then divided by the anisotropy profile. The result was compared with the calculated angular etendue of each camera.