Performance Assessment of the C-Mod Multi-Spectral Line Polarization MSE (MSE-MSLP) Diagnostic

Transcription

1 Performance Assessment of the C-Mod Multi-Spectral Line Polarization MSE (MSE-MSLP) Diagnostic Poster CP th Annual Meeting APS-DPP November 16-2; Savannah, Georgia S. Scott (PPPL), R. Mumgaard, M. Khoury (MIT/PSFC) The accuracy of the Alcator C-Mod Motional Stark Effect (MSE) diagnostic is limited primarily by partially polarized background light that varies rapidly both in time (1 ms) and space factor 1 variations are observed between adjacent spatial channels. ITER is likely to operate in a similar regime. Visible Bremsstrahlung, divertor molecular D2 emission, and glowing invessel structures generate unpolarized light that becomes partially polarized upon reflection. Because all three sources are broadband, the background light can be measured in real-time at wavelengths close to the MSE spectrum, thereby allowing the background to be interpolated in wavelength rather than in time. A 1-spatial-channel, 4-wavelength MSE-MSLP system has been developed using polarization polychromators that measure simultaneously the MSE pi- and sigma- lines as well as two nearby wavelengths that were chosen to avoid both the MSE spectrum and all known impurity lines on each sightline. Initial performance evaluation indicates that the background channel measurements faithfully track the background light in the pi- and sigma- lines. The improvement in accuracy of pitch-angle measurements and increased diagnostic flexibility over a wide range of plasma conditions will be reported.

2 Executive summary The Multi Spectral Line Polarization MSE diagnostic functions as designed. It accurately measures the partially polarized MSE background even in circumstances where the background changes rapidly in time and space: RF turn on, turn off L/H transitions P rad and VB spikes Ramp in density A few channels at the plasma edge have residual problems during large H L transitions. It provides a much more accurate measurement of the background light than traditional beam modulation.

3 C-Mod s MSE view dump is an ICRF antenna Developed a polarization sensitive camera to image light reflected from antenna Reflection is highly polarized, spatially complex Any light in the MSE wavelengths can become partiallypolarized MSE background MSE view footprints backlit onto antenna Polarization camera at MSE location Image reflected light Total light Polarization angle Unpolarized light source Polarized light Polarization fraction ICRF antennas R. T. Mumgaard Science meeting, Aug 4 th 214

4 MSE requires surprisingly high signal to background, and/or accurate estimates of the background, to achieve reasonable accuracy (e.g. =.1 to.2 o ) I s = intensity of polarized emission from beam I b = intensity of polarized background Standard beam on/off time interpolation measured signal + background estimated background (interpolation) Intensity of uncompensated polarized background = f I p ) b s S b = I s / I b = tan 1 (I 2 1 / I 2 2 ) = 28.6 o ( f S b beam measured background sin( 2( s b ) ) e.g. even with S b 3, we need to measure background to 1% accuracy to get.1 o accuracy on polarization angle! Imperfect estimation of the partially polarized background light is the dominant source of error for MSE on C Mod, and to date, limits its use to moderate density, L mode plasmas. time

5 Background is often dominated by visible Bremsstrahlung emission Seen on first pass and reflection Polarization fractions =.1-.1 Similar in all sightlines/wavelengths Perhaps just correlate with Bremsstrahlung diagnostic and subtract it? But this is not usually the dominate source. R. T. Mumgaard Alcator C-Mod MSE, HTPD- June 4 th /32

6 Glowing RF antennas, limiters, divertor can dominate MSE polarized background Polarized light post-disruption Structures only visible upon reflection Polarization fractions =.3-.5 Broadband emission Only visible in a few sightlines at a time Depending on location of source Due to complexity of view dump Varies rapidly in time R. T. Mumgaard Alcator C-Mod MSE, HTPD- June 4 th /32

7 Correlates very well with edge Dα But is not Dα Correlates with stray light from machine protection cameras Correlates with quasi- continuum from divertor spectrometers Polarization fractions =.1-.5 Changes quickly (<ms) Polarization angle depends on active divertor Similar behavior observed in all sightlines/wavelengths R. T. Mumgaard Alcator C-Mod MSE, HTPD- June 4 th /32

8 A multi-year effort has successfully characterized four sources of background light that affects MSE Visible Bremsstrahlung: ever present, sometimes dominates Small polarization fraction,.1 to.1 Similar in all MSE sightlines and wavelengths Molecular H 2 /D 2 ever present except in helium plasmas, sometimes dominates Moderate/high polarization fraction,.1 to.5 Similar behavior observed in all sightlines / wavelengths Most emission from active divertor polarization angle changes when USN LSN Can change on millisecond time scale during e.g. L H mode transitions Glowing invessel structures infrequently dominates High polarization fraction,.3.5 Glowing RF antennas, limiters, divertor. Also post-disruption Can vary rapidly in time (~1 ms) and space (factor >1 difference in adjacent channels) Emission from runaway relativistic electrons seen only during dedicated runaway expts Just a curiosity. Possibly gives info on energy distribution of runaways. see: R. A. Tinguely, GO4.6, Analysis of Runaway Electron Synchrotron Radiation in Alcator C-Mod.

9 But we can exploit the fact that all three sources of background light are quasi broadband Measure the polarized background light at wavelengths near the MSE and lines In real time (no time interpolation), simultaneously with the p and s measurements. With the exactly the same optics & fibers as the and light. Then interpolate in wavelength rather than in time to estimate polarized background emission.

10 R. T. Mumgaard Alcator C-Mod MSE, HTPD- June 4 th /32 background Estimated V U Q I Measured V U Q I Beam V U Q I = MSE red of V U Q I Wavelength interpolate MSE blue of V U Q I Valid if sources are indeed quasi-continuum. Doesn t require changes to upstream optics, just a detector swap. Measure the Stokes vectors at wavelengths adjacent to MSE on same sightline Use same PEM technique Wavelength interpolate Stokes vectors at each time to estimate MSE background

11 Passbands for the red and blue background filters were chosen to avoid lines of known C-Mod impurities N Kr Mo III Ca II I Ar II Fe I Ne I N IMo Ca I Kr IIII 3 N I Ca I Kr II Mo I N II W Ca I I N I N I N I N I Intensity Ar II[arb] II 3 N I N I Fe I N I N I Ne Ar I II Fe I Ar I Ca III Ca III H C II C II Fe I C I Ne I Kr II Fe I Ne Ar IIMo I Kr I Fe I Mo I Ca I Ca III N II W N I N I N I N I N I W N I I Brightest lines (Striganov and Sventitskii) H C II C II Ne I N II Ar II Ne I Ar II Kr I Fe Ne I I Ar II Mo I Mo I Ar II Ar II Ar II Ar II W He I I Fe Ne I Mo I Ar I Ar II Wavelength [nm] Blue# nm 1.8nm FW.5M All lines (From NIST) MSE Red# nm 1.5nm FW.5M Red# nm 1.7nm FW.5M He I Ar II Fe I Wavelength [nm] MSE Linestyle indicates the relative brightness of the lines compared to the brightest line for that species in the range from nm. Intensity > 5% max Intensity Intensity < 5% max Intensity Intensity < 25% max Intensity

12 Fiber from existing MSE upstream optics Spherical field mirrors APDs for each spectral channel ~1m Relay lenses Interference filters Condenser lens Interchangeable filter ovens allow quick, easy filter changes without need for realignment Cavity layout allows many spectral channels on same exact sightline, highly photon efficient Small AOI required to preserve bandpass filter performance, 3 used R. T. Mumgaard Alcator C-Mod MSE, HTPD- June 4 th /32

13 Very high throughput: 9mm 2 sr, filled at NA=.39 Acceptable filter performance: ~.5nm FWHM Thermal tuning with ovens Easy to create and maintain cavity alignment Water cooling for APD detectors Easily replicated, cost effective Machine independent Assembled, cover removed Filtered image transported to each detector R. T. Mumgaard Alcator C-Mod MSE, HTPD- June 4 th /32

14 A full 1-channel system was fabricated Cost ~$27k per polychrometer (sightline)

15 The MSE-MSLP diagnostic is rack-mounted; individual polychrometers can slide open for maintenance

16 A typical Alcator C-Mod plasma poses challenges for MSE background estimation NL_4 [1^19 m^-2] nebar (1^19 m^-3) Te (GPC) Te (TS) VB emission time (sec) 1. Ip (MA) Wtot (kj) Hα varies rapidly halpha 4 VB emission varies in time 6 2. injections Prad_2pi PRF (MW) ILL_ _b_1.pdf time (sec)

17 Measuring polraized background during beam off periods and then interpolating to beam on periods is not sufficiently accurate MSE π line, sightline total signal, beam on time-interpolated background I2Ω2 (mv) 1-1 total signal, beam off ersatz beam net signal due to beam off on off on off on off on off time (sec) ILL_ _fft_ss8_v2b_pi.pdf

18 During beam off periods, we measure the ratio of the MSE π and σ intensity to the two background intensities. During beam on periods we multiply this ratio by the measured background intensity to estimate the background intensity experienced by the π and σ channels sightline 8 raw polarized signal from background filter at nm 8 I2Ω2 (mv) 6 4 Normalization periods green: backtround intensity normalized to MSE-π intensity (purple periods) and scaled by background intensity (white periods) 2 black: raw polarized signal from MSE p-line at nm plotmslp_ pdf time (sec)

19 Measured polarized signals in the two background channels typically follow one another very well, confirming that the background light spectrum is broadband and slowly varying in time sightline 8 background #1 background # I2Ω2 (mv) PRF (MW) time (sec) follows_details_ pdf

20 In this noise-only shot, the wavelength-interpolation accurately reproduces the observed background light in the MSE π-line during ersatz beam-on periods sightline 8.3 normalization periods I2Ω2 (V).2.1. net signal is near zero time ILL_ _ss8_mslp_v2b_pi.pdf

21 H --> Lmode transitions are the most challenging phenomena when estimating MSE polarized background: rapid changes in VB emission and Hα 2 15 NL_4 [1^19 m^-2] PRF (MW) nebar (1^19 m^-3) Prad_2pi Te (GPC) Ip (MA) 2.5 Te (TS) VB emission time (sec) 8 Wtot (kj) halpha time (sec) ILL_wf_ _c_1.pdf

22 In this shot, wavelength interpolation accurately estimates the polarized MSE-p background emission even thru H --> L and L --> H transitions sightline 8 normalization period H -->L MSE-π I2Ω2 (mv) 4 2 back1 back2 net MSE-π is close to zero L -->H time-binned net MSE-π ILL_mslp_ _pi_ss8.pdf Prad [arb] 3 2 VB Hα 1 PRF time Caveat: post-facto, discovered imperfect background estimation on sightlines 9 and 1 on this shot, see discussion on upcoming slides. Size of transition may matter.

23 Surprisingly, on a similar-looking H --> Lmode transition, MSE-MSLP provides a less-accurate measurement of the MSE-π background for plasma-edge sightlines in the first few ms following the transition NL_4 [1^19 m^-2] nebar (1^19 m^-3) Prad_2pi Te (GPC) Te (TS) 2. Ip (MA) Wtot (kj) VB emission.25 halpha PRF (MW) wf_ _e_2.pdf time (sec) time (sec)

24 During the H--> Lmode transition, both background measurements see a smaller rise in signal than does MSE-π 6 both background channels see smaller rise in background signal than does MSE-π MSE-π 4 normalization period sightline 8 back-1 I2Ω2 (mv) 2 back-2 effect yields a spurious net signal at time of transision ersatz beam on off on off [arb] Ratio of VB to Hα intensity changes by a factor of ~2.7 during the transition. Prad VB Hα PRF ILL_mslp_ss8_ _pi.pdf time

25 The shortfall in background measurement at the H --> L transition is larger at the plasma edge, and absent in the plasma core sightline 1 (outer plasma edge) MSE-π I2Ω2 (mv) ersatz beam on off back-1 back-2 uncompensated net signal time on off ILL_mslp_ss1_pi_ pdf

26 The shortfall in measured background light is slightly more pronounced in the MSE σ-line than in the π-line sightline 8 MSE-σ I2Ω2 (mv) 5 uncompensated net σ-signal time ILL_mslp_ss8_ _sigma.pdf

27 The shortfall is observed on other shots sightline 1 H -->L MSE-π 15 I2Ω2 (mv) 1 L -->H H -->L 5 ersatz beam on uncompensated net π-signal off time on off on ILL_mslp_ss8_pi_ pdf

28 A single-channel prototype of MSE-MSLP (fabricated 212 at PSFC) was installed on ASDEX-Upgrade in October. Operational in 1 week. Credit: Alexander Bock

29 ITER Relevance Several existing MSE diagnostics (DIII D, NSTX, TFTR) have/had little or no problem with polarized background light. High current heating beam intense and light Viewing dumps and/or non reflective walls. Some existing MSE diagnostics (ASDEX U, JET) have only a modest problem with background light. But MSE on ITER will experience the same set of problems that it faces on Alcator C Mod: Hot walls, potentially glowing Low current beam (heating beam) low S/N Lack of viewing dump reflections leading to polarization High density, long sightline intense VB & molecular D2 MSE MSLP has been adopted for use on ITER MSE system: operational experience on C Mod will inform design and expected performance.

30 The future Repair the diagnostic neutral beam (December 215). Operate MSE MSLP during FY16 C Mod run campaign with DNB and document performance to provide guidance for ITER MSE. Find a good home for MSE MSLP following FY16 campaign. Tune algorithms that compute ratio of background signal to MSE and. Can use e.g. I 2 2, (I 2W12 +I 2w22 ).5, or total signal intensity. Possibly use post shot data from Intershot Calibration System. Explore whether constant of nature ratios are better than time varying ratios. Explore use of measured intensity ratio VB/H to identify time periods when MSE MSLP is not sufficiently accurate.