All about sparks in EDM (and links with the CLIC DC spark test) Antoine Descoeudres, Christoph Hollenstein, Georg Wälder, René Demellayer and Roberto Perez Centre de Recherches en Physique des Plasmas (CRPP), Ecole Polytechnique Fédérale de Lausanne (EPFL) Charmilles Technologies SA, Meyrin 1/ 36
Outline of the presentation I. Introduction II. III. IV. Experimental setup and diagnostics Some results about the EDM spark Links with the DC Spark Test 2/ 36
Electrical Discharge Machining (EDM) EDM = successive removal of small volumes of workpiece material, using the eroding effect of electric discharges on electrodes electrode Pre-breakdown dielectric liquid Breakdown Discharge workpiece V I End of the discharge ~ 200 V ~ 10 A ~ 20 V Post-discharge ~ 100 μs time 3/ 36
die-sinking machines Two types of EDM machines The electrode keeps its form (asymmetry wear/erosion) Production of injection moulds wire-cutting machines The electrode is a travelling wire Production of steel cutting dies and extrusion dies 4/ 36
Examples of parts machined with EDM 5/ 36
Motivations and purpose of the work We want improvements in machining accuracy (μ-machining) and surface roughness improvements in material removal rate and reduction of wear reliable numerical models to optimize the performances of EDM But important lack of knowledge about basic EDM phenomena complex phenomena experimental difficulties stochastic nature empirical optimization numerical models with empirical parameters We need a better fundamental understanding of the EDM discharge and of its interaction with the electrodes Systematic investigation of the EDM plasma 6/ 36
Outline of the presentation I. Introduction II. III. IV. Experimental setup and diagnostics Some results about the EDM spark Links with the DC Spark Test 7/ 36
Machining device EDM machine motor dielectric circuit control vertical displacement EDM pulse generator pump dielectric shower electrode optical fibre workpiece 8/ 36
Diagnostics Electrical measurements Imaging Light intensity Optical emission spectroscopy current probe electrode (Cu, C, W, Zn) EDM pulse generator G V voltage probe to photomultiplier spectrograph camera plasma optical endoscope fibre dielectric (water, oil, liq-n 2 ) workpiece (steel) 9/ 36
Optical Emission Spectroscopy OES = analysis of the emitted light with a spectrograph (dispersion of the light by a grating) spectrograph spectrum optical fibre CCD camera gratings computer Characteristic lines identification of emitting atoms and ions in the plasma Relative intensities of Cu lines (+ LTE) electron temperature measurements Stark broadening and shift of the H α line electron density measurements 10 / 36
Experimental difficulties Small size (gap 10 100 μm) In a liquid environment Ultra High Vacuum Weak light emission for spectroscopy Short duration (~ μs, ~ ns for breakdown phenomena) Electrical interferences Poor reproducibility of the discharges Few diagnostics available, and difficult to apply The list is almost the same with our DC sparks 11 / 36
Outline of the presentation I. Introduction II. III. IV. Experimental setup and diagnostics Some results about the EDM spark Links with the DC Spark Test 12 / 36
Imaging of the process (Cu / steel, 50 μs, 8 A, water) 13 / 36
Plasma imaging Evolution of the plasma light intensity breakdown phase / discharge phase Typical plasma image (8 μs, 4 A, water) Excited region broader than the gap Slight growth with time Diameter increases with the discharge current (Cu / steel, 100 μs, 24 A, oil) 14 / 36
Beginning of the discharge: fast imaging 50 to 100 ns V/200 [V] 1 I / 8 [A] 200 to 250 ns 0 100 to 150 ns 250 to 300 ns 150 to 200 ns exposure -0.5 0 0.5 1 time [μs] (6 A, water) the plasma develops very fast (< 50 ns) afterwards : stability 15 / 36
End of the discharge and post-discharge The plasma disappears as soon as the current is shut down Weak light emission during the post-discharge Images obtained directly after a discharge : Corresponding spectrum blackbody fit 2300 K blackbody : incandescence of the eroded particles 2300 K : molten metal 16 / 36
Typical spectrum Dielectric water : H, O oil : H, C, C 2 liquid nitrogen : N Cu / steel, water H α Electrode material copper : Cu graphite : C, C 2 tungsten : W zinc : Zn, Zn + Cu Fe Cr Cu O Cu O C Workpiece material steel : Fe, Cr, C (12 μs, 12 A) Characteristic atomic lines : dielectric cracking and contamination 17 / 36
Effect of the discharge on-time (Cu / steel, water, 12 A) Increase in the H α FWHM and shift Increase in the continuum Extremely high electron density at the beginning of the discharge 18 / 36
Time-resolved spectroscopy t = 0 : breakdown Continuum due to the merging of spectral lines (12 A, water, time res. 200 ns) The plasma is very dense during the first microsecond 19 / 36
Time-resolved spectroscopy of H α : electron density (16 A, water, time res. 1 μs) n e reaches 2 10 18 cm -3 at the beginning n e decreases with time (plasma expansion) The plasma is created from a LIQUID! 20 / 36
Time-resolved spectroscopy: electron temperature No ionic spectral lines Two-line method with Cu lines: Cu lines T e 0.7 ev (8'100 K) cold plasma 21 / 36
Spatially-resolved spectroscopy Spatial sampling with endoscope + fibre bundle Spatial resolution : ~ 20 μm (typical gap : ~ 100 μm) to copper electrode (+) Cu line Cr line to steel electrode (-) plasma light analysis in different zones Plasma contamination: Cu line (from electrode) Cr line (from workpiece) (100 μs, 6 A, water) spatial asymmetry of the contamination 22 / 36
Spatially-resolved spectroscopy Electron temperature vertical profile 510.5 521.8 515.3 T e homogeneous Electron density vertical profile n e slightly higher in the center (50 μs, 12 A, water) 23 / 36
Plasma coupling parameter Γ Γ = 2 2 Z e 4πε ak 0 B T n T 1/3 n Γ = "Coulomb interaction" "thermal energy" EDM Γ << 1 : ideal plasma Γ 1 : weakly non-ideal plasma Γ > 1 : strongly coupled plasma EDM : n e 10 18 cm -3 T e 0.7 ev Γ 0.33 T EDM plasma: weakly non-ideal (dense & cold) 24 / 36
Summary: physical properties of EDM plasmas Composition: dielectric cracking + electrodes contamination T e 0.7 ev (cold) n e 10 18 cm -3 (dense) p 10 bar Weakly non-ideal Fairly ionized Small dimensions High electric fields Intense during the first μs Numerous terms in the schematic! energy balance Relatively insensitive to most of the discharge parameters 25 / 36
Outline of the presentation I. Introduction II. III. IV. Experimental setup and diagnostics Some results about the EDM spark Links with the DC Spark Test 26 / 36
Are we talking about the same thing? EDM current constant duration is chosen (2μs 1ms) low voltage brkd spark energy < 0.01 J DC spark test discharge of a capacitor high voltage brkd, high current spark energy ~ 1 J V I V I ~ 200 V ~ 10 A ~ 20 V ~ 10 kv ~ 300 A ~ 100 μs time ~ 2 μs time 27 / 36
Are we talking about the same thing? EDM both plasmas are sparks DC spark test sparks in LIQUID sparks in VACUUM And what about the breakdown mechanism? 28 / 36
Breakdown mechanisms in vacuum FE current outgassing heating of the emission site presence of vapour material fatigue vapour is ionized by FE current material break-up avalanche of electrons BANG! Vapour ( = a medium more dense than vacuum ) is needed for the propagation of the avalanche 29 / 36
Outgassing in the DC spark test gas is released in a few attempts before breakdown Example: Release of hydrogen gas, Mo electrodes Ion Current [A] 8x10-9 7x10-9 6x10-9 5x10-9 4x10-9 3x10-9 2x10-9 Ion Current [A] 4,6x10-10 4,5x10-10 4,4x10-10 4,3x10-10 4,2x10-10 4,1x10-10 4,0x10-10 2700 2750 2800 2850 2900 2950 Relative Time [sec] 341,755 MV/m 372,037 MV/m 389,341 MV/m 1,2x10-7 1,0x10-7 8,0x10-8 6,0x10-8 4,0x10-8 Pressure H 2 [mbar] 1x10-9 2,0x10-8 0 0 500 1000 1500 2700 2800 2900 3000 Relative Time [sec] (from Trond) 30 / 36
Breakdown mechanism in dielectric liquids (and, to some extent, in high-pressure gases) A single electron avalanche can not propagate far away in a liquid needs a medium less dense than a liquid : streamer breakdown! Initiation : a vapour bubble (pre-existing, or by FE) formation of a streamer (= thin weakly ionized channel) the streamer bridges the gap creates highly ionized channel, heats up surrounding gas, shockwave primary avalanche in this low-density region, bubble growth streamer growth and propagation (10 2-10 4 m/s) back streamer ( return stroke ) = enormous electron avalanche, reverse ionizing front ~ 10 7 m/s BANG! 31 / 36
Streamers in dielectric liquids in oil time in μs gap 1.3 cm 82 kv sharp needle J.C. Devins et al., J. Appl. Phys. 52 4531 (1981) 32 / 36
Streamers in dielectric liquids structure and propagation speed depend on polarity (NB: can start from the cathode or the anode!) J.C. Devins et al., J. Appl. Phys. 52 4531 (1981) ~ 100 m/s ~ 1-10 km/s Common fact about breakdowns in vacuum, gas or liquid : the electron avalanche takes place in a gaseous medium 33 / 36
Light emission at breakdown EDM 500 ns light peak DC spark test 50 ns light peak (sometimes!) Also a dense plasma at the very beginning? Rapid melting / vaporization of a small protrusion? 34 / 36
Light emission at breakdown EDM 500 ns light peak DC spark test 50 ns light peak (sometimes!) continuous emission due to high density??? line emission by excited species 35 / 36
Conclusion Electron avalanches leading to breakdown need a gaseous medium to propagate ( low-density region ) This is true in vacuum, gases, dielectric liquids and solids EDM discharges produce dense plasmas, especially during the first microsecond They are created from a liquid First light measurements suggest that plasmas of the DC spark test could also be dense immediately after the breakdown The melting of a protrusion scenario is probable in this case These plasmas have some similarities in the early stage of their life (probably cousins, but not twin brothers) 36 / 36
37 / 36
Streamer propagation 38 / 36