[ SYSTEM_STATUS: ACTIVE // SPECTRAL_ANALYSIS: ACTIVE // 2026 ]
[ 01 // OPERATIONAL_FRONT_G4 ]

METAL_DETECTOR
SPECTRAL_G4 // CORE_AI

Pushing the limits of traditional VLF detection. Spectral-G4 harnesses the power of the STM32G4 processor and external 18-bit SAR converters to analyze the phase of three harmonics simultaneously. This is more than a detector – it's a mobile geophysical lab integrated with local AI.

ADC_18BITSAR_PRECISION_ARCHITECTURE
MULTI-FREQ7.8 / 23.4 / 39.0 kHz
LOCAL_AINEURAL_DISCRIMINATION
DETECTION_PROTOCOL // 2026

Spectral signature analysis allows for precise discrimination between ferrous and non-ferrous metals even in extremely difficult ground conditions. The system operates on raw I/Q vectors.

  • [01] PHASE_SENSITIVE_DETECTION (PSD)
  • [02] VECTOR_GROUND_REJECTION
  • [03] SPECTRAL_SIGNATURE (AI)
STATUS: SYSTEM_READY // Spectral-G4 Pro_Version
[ 02 // CORE_SPECIFICATIONS ]

CORE_SYSTEM_
COMPONENTS

FULL TECHNICAL DOCUMENTATION AVAILABLE AT: [ SPECTRAL_G4_REPOSITORY ]

MODULE_01 // MULTI-
FREQUENCY_ENGINE

F_CORE // Simultaneous analysis of 7.8kHz, 23.4kHz and 39.0kHz.

PROCESSING:IQ_DEMODULATION // 170MHz DSP
CAPABILITIES:

Detecting phase shifts for each harmonic separately. Allows building a precise spectral signature of the object (Spectral_ID).

STATUS: SYNC_STABLE // FREQ_LOCKED
MODULE_02 // ANALOG_FRONT-END (AFE)

ADC_ULTRA // 18-bit PulSAR for extreme dynamic range.

RESOLUTION:2^18 STEPS // SAR_ADC_INTEGRATION
CAPABILITIES:

Eliminates the need for analog ground compensation circuits. The entire signal is digital, allowing for lossless Software-Defined filtering.

STATUS: AFE_CALIBRATION // OPTIMAL_SNR
MODULE_03 // NEURAL_CLASSIFIER

AI_BRAIN // Target classification via neural networks.

ARCHITECTURE:X-CUBE-AI // ON-DEVICE_INFERENCE
CAPABILITIES:

Instead of simple VDI thresholds, AI analyzes the full vector trajectory during coil movement, recognizing signature shapes of difficult targets.

STATUS: MODEL_LOADED // INFERENCE_ACTIVE
[ 03 // HARDWARE_CONFIGURATION ]

G4_UNIT_RESERVATION

[ HARDWARE_CONFIGURATION ]

G4_SET // FOUNDER_PACKAGE

CO-CREATOR // Receive a prototype in exchange for supporting development.

COST_ESTIMATE:PREFERENTIAL_PRICE (FOR_CONTRIBUTORS)
PACKAGE_INCLUDES:

Access to 'Series 0' prototype at production cost. Direct contact with the creator, influence on functionality.

TECHNICAL_MESSAGE:

Ideal for enthusiasts who want to help bring this gear to life and be the first to receive it.

FALLBACK // ENGINEERING_CONTACT:
takzen.app@gmail.com
[ 04 // SPECIALIZED_DEPLOYMENTS ]

MULTIDIMENSIONAL_
OPERATION

STATIC_DETECTION

Pure vector compensation without motion filters. Allows for target detection at zero coil speed, ideal for precise pinpointing and analysis of deep signals.

DYNAMIC_DEPTH

Multi-stage band-pass filters designed on STM32G4. Maximum dynamic range while eliminating ground mineralization influence.

HARMONIC_ANALYSIS

Not just the base phase. We analyze the 3rd and 5th harmonics, allowing for unique discrimination of objects with similar VDI but different material structure.

[ 05 // PERCENTAGE_SPECIFICATION ]

G4_ADVANTAGE //
Signal_
Characteristics

GROUND_STABILITY_01

Vector Ground Balance system eliminates soil influence with 0.01-degree phase accuracy, ensuring stability in difficult terrain.

MICRO_SENSITIVITY_02

Thanks to the 18-bit SAR converter, the system detects microvolt-level signals previously hidden in quantization noise.

SPECTRAL_LOGIC_03

Agentic signal analysis. AI decides in real-time whether a signal is worthy of attention or just EMI interference.

REACTIVITY_04

Processor latency below 1ms. Every signal reaches the audio directly, without delays typical of legacy signal processors.

[ 07 // ENGINEER_BIOGRAM ]

PROJECT_CREATOR // Krzysztof_Takzen

I am Krzysztof. My goal is to redefine the modern metal detector by integrating low-level DSP with sovereign AI models. Spectral-G4 is the fruit of my search for the perfect balance between analog sensitivity and digital precision.

I specialize in Embedded Systems, STM32 microprocessor architecture, and digital signal processing. This project is a 'Code-First' manifesto in the hardware world – rigorous engineering and algorithm transparency.

As part of the Spectral-G4 project, I fulfill my mission through:

  • >> Development of an open DSP platform for hobbyists
  • >> Implementation of neural networks on microcontrollers
  • >> Designing ultra-low-noise analog signal paths
Instruktor Krzysztof
ENGINEER_IDENTIFICATIONACTIVE
HARDWARE_KEY_AUTHMATCH_99.8%
SYSTEM_ACCESS_LEVELLEAD_DESIGNER
AUTH_STATUSVERIFIED
[ 08 // PROJECT_RESOURCES ]

REPOSITORIES //
DOCUMENTATION

CODE_SHOWCASE:
FIRMWARE // STM32G4

Public demonstration repository (Showcase). Contains selected DSP algorithm fragments and project structure, demonstrating code style and engineering quality.

HARDWARE: ARCHITECTURE //
SYSTEM_OVERVIEW

Public technical documentation including block diagrams, interface specifications (SPI/I2S), and analog signal path characteristics. Full schematics exclusive to partners.

[ STATUS: DOCUMENTATION_IN_PROGRESS ]
[ 10 // COMMUNICATION_CHANNEL ]

DIRECT_OPERATIONAL_LINK