Dense plasma focus — coaxial drive, focused pinch
A coaxial-electrode device whose discharge lifts a current sheath over an insulator, accelerates it down the coaxial gap and rolls it into a dense micrometre-to-millimetre-scale pinch at the anode tip, where instability-driven ion beams and hot spots produce the neutron yield. SCPN-DENSE-PLASMA-FOCUS-CORE owns the device truth of the dense_plasma_focus configuration.
What the configuration is
The plasma focus ends in a pinch, but its identity is the coaxial rundown and the focus formation: a capacitor bank of kilojoules to megajoules drives a current sheath along the annulus between anode and cathode, the sheath sweeps up the fill gas like a snowplow, turns the corner at the anode tip and collapses onto the axis. The pinch lasts tens of nanoseconds; its neutrons come largely from a fast deuteron beam driven by the pinch's own instabilities striking the surrounding gas, which is why the family's twelve-machine table lists beam quantities beside pinch ones. The Lee model describes all of this with a normalised circuit and a handful of mass and current factors fitted to measured waveforms.
What the core owns
- the analytic device physics models: closed-form and 0-D models from the plasma-focus literature evaluated on the validated configuration (no solver code, no phase integration, no FUSION seam);
- the device boundary: plant and experiment truth, shot lifecycle and configuration policy for coaxial-electrode focus devices; sheath-dynamics and focus-phase semantics (rundown, roll-over, radial collapse, pinch and disruption phases, beam-target contribution declarations) as device truth;
- the declared configuration: anode radius, cathode radius and anode length; bank energy, peak current, fill pressure and whether the fill is deuterium;
- diagnostic semantics, reference frames and clock identity declarations; the CONTROL adapter specification; the registry binding (version 1.0.0) and the domain manifest.
What it explicitly excludes
- Generic axial-current Z-pinch devices: SCPN-Z-PINCH-CORE — the DPF ends in a pinch, but its identity is the coaxial rundown and focus formation, not a preformed static column; theta pinch: SCPN-THETA-PINCH-CORE.
- Beam-target fusion systems with externally accelerated beams: SCPN-BEAM-TARGET-CORE (electrostatic and hybrid family); the DPF's internal beam-target contribution is device truth here, not a beam-facility claim.
- Solver mathematics and validation evidence: SCPN-Fusion-Core, until an exact surface passes the reactor family migration gate; no solver code exists in, or was copied into, the device core.
- Typed signal semantics and comparability: SCPN-Phase-Orchestrator (review-only output, never actuation). Control admission and action formation: SCPN-Control. Machine protection: independent systems retain the final veto. Portfolio presentation and gating: SCPN-Studio.
Level-0 physics and its anchor
Bank normalisation and fill state (Lee 2014 eqs. 4–6, 9, 43), axial characteristics (eqs. 5–7, eq. 1 at rest), radial characteristics, slug relations and rule-of-thumb geometry (eqs. 14, 15, 24–28, 32, 34; ICTP Tables 2–3), pinch-phase closed forms (eqs. 39–48, both self-absorption branches), the fast-ion-beam chain (TECDOC eqs. 5–6, items (a)–(k)) and two neutron instruments (Lee eq. 50 beam-target with a declared cross-section; the empirical \(9\times10^{10} I^{3.8}\) law, refused outside 0.1–1 MA). Numerics through the pinned shared kernel library (logarithm, exponential, real power), bit for bit.
Anchored on Table 1 of Saw & Lee (IAEA-TECDOC-1829, 2017), the twelve machines fitted with the code, four of them in the tests: PF1000 (486 kJ, 1332 µF, 27 kV, anode 11.6 cm × 60 cm, peak 1.85 MA), NX3 (14.5 kJ), INTI (3.4 kJ), PF400J (0.4 kJ). The \(E_0\) and rise-time columns are reproduced within 2.5 % and 2 %; the terminal snowplow speed overestimates the table's \(v_a\) column by 7–14 % on every row and the tests assert that sign; the beam chain reproduces three rows within 3 % and PF400J within 12 % (its two-digit pinch radius enters as \(r_p^{-2}\)); the drive parameter reproduces the \(SF\) column for PF1000 and NX3 within 1 %, and the INTI row is excluded because its printed \(SF\) does not follow from its own printed inputs. The tabulated pinch temperature and density are outputs of the integrated code and are not reproduced by the closed forms (factors of about 3 and 5 in a hand check), which the record states.
Load the four machines in the explorer →
Non-claims
- Every number is a closed-form evaluation of published relations on a synthetic configuration and a declared pinch state; no phase of the Lee model is integrated, no shot is simulated, no current waveform is fitted.
- The anchors reproduce numbers printed in the sources, which are themselves outputs of the source's fitted code; they are not correlations with experimental data.
- No yield, gain, reactivity, confinement or breakeven statement is made; the beam-target and scaling-law values are consistency instruments at the declared inputs, and the thermonuclear term is not implemented.
- No value describes, approximates or validates any real machine; electrode-geometry, fill-gas and repetition-rate choices are configuration facets, not separate claims.
Capabilities and evidence
Evidence maturity computational_prototype, five implemented capabilities, each with its evidence record in VALIDATION.md: the device configuration model, the diagnostic and clock semantics model, level-0 device physics (the closed forms on this page, with a Rust crate mirroring every kernel in identical operation order, float64 bit patterns compared, and a per-point evaluation benchmark), the device 3D model and the device CAD model. No parameter set or channel describes any real machine or diagnostic.