PYNQ-Z2 Mixed-Signal Oscilloscope

GitHub: https://github.com/Ross0907/pynq-z2-mixed-signal-oscilloscope

Overview

The PYNQ-Z2 Mixed-Signal Oscilloscope is a custom-developed FPGA-based instrumentation platform combining a dual-channel 16-bit oscilloscope, logic-analyzer connectivity, and an arbitrary waveform/function generator.

The project aims to develop an accessible, reconfigurable alternative to conventional closed instrumentation systems, bringing together analog circuit design, FPGA processing, high-speed PCB engineering, and software-defined measurement.

Beyond conventional waveform acquisition, the architecture is designed to support future integration with AI-assisted laboratory automation.

Hardware Architecture

The acquisition subsystem incorporates the AD9655 dual 16-bit ADC, ADA4817-2 high-speed amplifiers, ADA4927-2 fully differential driver, and TMUX1574 analog multiplexers.

Two acquisition configurations are provided:

DUAL: Simultaneous acquisition at 62.5 MSPS per channel.

FAST: Single selected-channel acquisition at 125 MSPS.

Each analog input incorporates protection, selectable attenuation, AC/DC coupling, amplification, and switchable anti-alias filtering.

An AD9102-based arbitrary waveform generator provides a separate signal-generation subsystem.

The custom four-layer PCB integrates analog front ends, dedicated clocking, power conversion, local regulation, FPGA interconnects, logic-analyzer connectivity, and three BNC connectors.

Design and Engineering Validation

The complete analog acquisition path was evaluated using Cadence PSpice, considering both channels, acquisition configurations, ADC loading, switching, filtering, and input coupling.

Pre-fabrication simulation results include:

ParameterResultFAST -3 dB bandwidth41.37–41.47 MHzDUAL -3 dB bandwidth22.05–22.07 MHzFAST Nyquist rejection41.74–42.22 dBDUAL Nyquist rejection36.19–36.28 dBAC coupling corner3.44–3.47 Hz

The engineering workflow additionally incorporates KiCad PCB design, post-layout extraction, signal-integrity analysis, and power-integrity verification.

These figures represent simulation results, not measurements from fabricated hardware.

Extending the Platform: AI-Assisted Instrumentation

A major future development direction is integration with ngscopeclient and the Model Context Protocol (MCP).

The proposed architecture connects the FPGA acquisition backend to ngscopeclient through a custom libscopehal-compatible instrument driver. An MCP server would subsequently expose structured instrument-control and measurement capabilities to compatible AI assistants.

This would enable natural-language interaction with physical measurement hardware, including:

Configuring acquisition, triggering, and waveform generation.

Capturing and analyzing analog and digital waveforms.

Performing automated frequency, amplitude, FFT, and noise measurements.

Decoding digital communication protocols.

Comparing acquired measurements against design specifications.

Assisting with circuit debugging and generating reproducible test reports.

The longer-term objective is to evolve the oscilloscope into an AI-accessible laboratory instrument capable of participating in automated experimental workflows.

The MCP/ngscopeclient integration is a proposed extension; it is not presented as an already validated implementation.

Project Vision

This project demonstrates how custom mixed-signal hardware, FPGA-based acquisition, open-source instrumentation software, and AI interfaces can be combined into a modular platform for electronics research, education, debugging, and future autonomous laboratory development.

Complete project development and engineering documentation are maintained on GitHub.