Helios Core is a deterministic, auditable, and publication-ready environment for solar cell current-voltage (IV) characterization. It replaces subjective, manual workflows with standards-compliant, reproducible analysis, ensuring that identical inputs produce bitwise-identical outputs within defined scientific tolerances.
"If the same data is analyzed twice under the same conditions, the result must be numerically consistent within defined tolerances, fully explainable, and independently reproducible."
- Determinism Over Speed - Reproducibility prioritized over computational efficiency
- Transparency Over Convenience - No black-box mathematics; all equations visible
- Immutability Over Flexibility - Raw data is sacred and never modified
- Auditability Over Simplicity - Every calculation is logged and traceable
- One-Diode Model (Standard Shockley + Rs + Rsh)
- Two-Diode Model (Dual recombination pathways)
| Parameter | Symbol | Physical Meaning | Typical Range |
|---|---|---|---|
| Short-circuit current density | Jsc | Photogenerated current | 0-50 mA/cm² |
| Open-circuit voltage | Voc | Maximum voltage output | 0-1.5 V |
| Fill Factor | FF | Squareness of IV curve | 0.3-0.85 |
| Power conversion efficiency | PCE | Overall device performance | 0-30% |
| Series resistance | Rs | Contact/transport losses | 0-100 Ω·cm² |
| Shunt resistance | Rsh | Recombination losses | 10-10⁶ Ω·cm² |
| Ideality factor | n | Recombination mechanism | 0.8-2.5 |
- Preconditioning - Data normalization and validation
- Global Search - Differential Evolution with fixed seed
- Local Refinement - Levenberg-Marquardt optimization
- Post-fit Validation - Physical plausibility checks
- Hysteresis Index (HI) - Forward vs reverse scan comparison
- STC Normalization - IEC 60891 standard corrections
- Maximum Power Point (MPP) - Cubic spline interpolation
- Residual Analysis - Model mismatch detection
Raw CSV/TXT/XLS → Auto-detection → Multi-pixel segmentation → Immutable storage
- Smart Parser: Detects voltage, current, hardware profiles (Keithley, Keysight, Ossila)
- Provenance Tracking: SHA-256 hashes of all raw data
- Batch Processing: Multi-pixel substrates automatically segmented
- Manual data scrubbing/masking
- Interactive parameter adjustment
- Rapid preliminary fitting (~8 seconds/curve)
- Hypothesis testing and diagnostics
- Deterministic analysis only
- Fixed seeds and tolerances
- Manual interventions disabled
- Bitwise reproducible results (~60 seconds/curve)
- Primary IV and power curves
- Residual plots for model validation
- Batch statistical distributions
- Pixel spatial mapping
Supplementary Bundle (.zip)
├── report.pdf # Vector graphics, journal-compatible
├── data.csv # Normalized IV data
├── audit.json # Complete provenance and parameters
└── reproduce_analysis.py # Standalone reproduction script
- Three-Column Command Center:
- Inventory & Provenance - Dataset management
- Visual Analysis Hub - IV/Power plots + residuals
- Parameter Inspector - Extracted metrics + diagnostics
- Deterministic UI: No hidden state, no UI-side calculations
- Physics Audit Sidebar: Real-time diagnostic monitoring
- Deterministic Physics Engine: Fixed seeds, single-threaded BLAS
- File-based Persistence: SQLite + raw file storage
- REST API: Stateless, deterministic endpoints
- pvlib-python - Photovoltaic modeling (IEC-aligned)
- SciPy - Numerical optimization (Differential Evolution, LM)
- NumPy - Numerical primitives (float64 only)
- Pandas - Structured data handling
| Aspect | Control Method | Verification |
|---|---|---|
| Floating-point | float64 only, explicit casting | Bitwise hash equality |
| Randomness | Fixed seed (42) in Reference Mode | SHA-256 output hashes |
| Threading | Single-threaded BLAS (OMP_NUM_THREADS=1) | Environment variable lock |
| Convergence | Explicit tolerances (xtol, ftol, gtol) | Solver log validation |
- Synthetic Dataset Testing - 100+ curves with known parameters
- Noise Robustness - Parameter stability under ±2% Gaussian noise
- Boundary Stress Testing - Edge case handling (high Rs, low Rsh)
- Residual Pattern Analysis - Systematic error detection
| Mode | Time per Curve | Accuracy | Use Case |
|---|---|---|---|
| Exploration | 7-10 seconds | ~99% of Reference | Interactive screening |
| Reference | 55-65 seconds | 100% deterministic | Publication analysis |
- Primary Parameters: Jsc, Voc, FF, PCE, Rs, Rsh, n
- Derived Metrics: Hysteresis index, temperature coefficients
- Quality Indicators: RMS residuals, convergence status
- Health Checks: Parameter bounds, physical plausibility
- Publication-ready plots (PDF, SVG, PNG)
- Interactive web visualizations
- Batch comparison overlays
- Residual analysis charts
- Python reproduction script - Standalone, dependency-managed
- Audit metadata JSON - Complete analysis provenance
- Determinism hashes - SHA-256 verification chains
- Configuration snapshots - Exact solver parameters
- Solar Cell Researchers - Perovskite, silicon, organic PV
- Device Fabrication Labs - Process optimization, quality control
- Academic Institutions - Teaching, student projects
- Industrial R&D - Prototype characterization
- Daily lab analysis - Quick screening of fabrication batches
- Publication preparation - Reference-grade figure generation
- Method validation - Comparison against established techniques
- Educational demonstrations - Transparent physics education
- Inter-lab comparisons - Standardized analysis protocols
- First IV analysis tool guaranteeing bitwise reproducibility
- Cryptographic hash verification of all results
- Cross-platform numerical consistency within 0.1%
- Real-time diagnostic monitoring
- Systematic error pattern detection
- Actionable scientific recommendations
- Exploration: Fast, interactive hypothesis testing
- Reference: Locked, reproducible publication analysis
- Raw data immutability
- Full parameter traceability
- Standalone reproduction capability
- Backend: Render.com (Python FastAPI)
- Frontend: Vercel (Next.js React)
- Storage: File-based (SQLite + raw files)
- Cost: $0/month (free tier)
- Web Application - Full interactive interface
- API Endpoints - Programmatic access
- Export Scripts - Standalone Python reproduction
- Two-diode model validation
- Temperature-dependent analysis
- Batch statistical reporting
- Transient photovoltage analysis
- Impedance spectroscopy integration
- Machine learning-assisted diagnostics
- Multi-technique characterization platform
- Collaborative analysis environments
- Cloud-based reproducibility archives
- Eliminates "analysis variability" between research groups
- Provides standardized methodology for IV characterization
- Enables true reproducibility in photovoltaic research
- Reduces time from measurement to publication
- Sets new standards for computational reproducibility
- Provides reference implementation for IV analysis algorithms
- Creates audit trail for scientific data processing
- Democratizes advanced analysis capabilities
- Upload CSV/TXT measurement file
- Explore data in Exploration Mode
- Switch to Reference Mode for final analysis
- Export complete publication bundle
- Repository: github.com/otobrixai/helios-core
- Documentation: Complete FRD, SAD, SOP specifications
- API: RESTful endpoints with OpenAPI documentation
- Testing: Comprehensive validation suite
Helios Core transforms IV characterization from an art into a science by providing:
- 🔬 Scientific Rigor - Deterministic, physics-based analysis
- 📊 Publication Readiness - Complete export bundles with reproducibility scripts
- ⚡ Practical Usability - Fast exploration with rigorous reference modes
- 🔍 Complete Transparency - No black boxes, all mathematics exposed
- 💰 Zero Cost - Fully functional on free-tier infrastructure
Helios Core isn't just software—it's a scientific instrument for the computational age, bringing the reproducibility crisis in photovoltaic research to an end, one deterministic analysis at a time.