/Multislope-3I

Improved multislope ADC

Primary LanguageCMIT LicenseMIT

Multislope 3I

Introduction

This repository contains hardware and software files for the third multislope prototype board, designated 3I. This is the most complete and proper version of the PCB so far, including a proper 8-resistor network, an LM399/ADR1399 reference, composite integrator (with two SOIC-8 footprints), shielding on the integrator summing node and test points to easily probe crucial nodes. History and operation are documented extensively here.

Project Motivation

The second prototype PCB was completed hastily with not much thought put into either layout or choice of parts, so a better-designed (and hopefully better-evaluated) board was in order.

Usage

The KiCAD project files can be used to generate gerber files, which can be used to manufacture the PCBs. The code is written in C/C++ in the VScode development environment. Detailed instructions on how to compile the software will be added later. A .uf2 file is present in the project outputs which can be uploaded directly to the Pi Pico.

List of Files

  • Hardware: KiCAD PCB design files, BOM and schematic. Front side of the PCB
  • Software: C/C++ code, uf2 file for direct upload.
  • Resources: Infographics

Application Examples

  • The readings from the overall converter (PWM counts and residue) were noisy, and the raw readings from the MCP3202 residue ADC were noisy and contained unexplained banding. Linearity shows a parabolic curve, probably because of wrong constants or small mismatches in switch resistances.

Linearity

Linearity

MCP3202 Raw Data (Detail)

Noise

Cause

  • Reading noise was determined to be caused by poor layout and decoupling around the MCP3202 residue ADC.
  • Linearity bell curve could be caused by input resistor's PCR.

Notes

  • An interactive Excel document is provided in the Resources folder, with a slider to vary the residue constant to visually determine the best constant to reconcile runup and residue readings. Update: Excel file broke somehow, but you can still see the effect of moving the slider in the video linked below, timestamp 10:15.
  • Another version of the converter is in development, with improvements to speed, resolution and noise.

Links

https://www.youtube.com/watch?v=aNtOfKR7sto