clutgen 0.1.1

CLUTGen, a flexible ADC LUT generator for embedded systems.

---

C Lookup Table (LUT) Generator

CLUTGen automates the creation of Look-Up Tables for embedded systems, specifically for converting raw ADC (Analog-to-Digital Converter) readings into calibrated physical units (like temperature, pressure, or distance).

Features

• Curve Fitting: Automatically find your fit from many algorithms.
  • Linear Interpolation
  • Cubic Splines
  • Polynomial Interpolation (up to 7th degree)
  • Piecewise Interpolation
  • Inverse Distance Weighting
• Visual Preview: Generates plots to compare calibration data against the calculated LUT.
• C Code Generation: Outputs production-ready .c and .h files with Doxygen-style comments.

Requirements

• Python 3.10+
• numpy, matplotlib

Usage

# From the root of the repository
pip install .
# Generate LUTs with cubic splines fit for every configuration toml in the given paths.
clutgen -s -i ./examples/calibration/*.toml ./examples/temperature_copy.toml

YAML Configuration Template

Create a CSV file mapping your POIs with the given schema, where raw is the measured ADC raw value, and calibration is the desired output for the LUT at that given raw value:

raw,calibration

With your table done, create a configuration TOML, e.g.:

# temperature.toml

name = "temp_sensor"                  # Used for variable names (e.g., temp_sensor_lut)
description = "ambient temperature"   # Used for documentation and plot titles
table_resolution_bits = 12            # Resolution (e.g., 12 bits = 4096 entries)
lut_type = "int16_t"                  # C type (int8_t, uint16_t, etc.)
 
samples_csv = "./data/temperature_samples.csv" # Path to CSV/Excel with 'raw' and 'calibration' columns. Path is global or relative to this file.
interpolation = "polynomial" # Optional, forces a given interpolation method to this LUT.

Currently available interpolation methods are:

Interpolation Argument	Description
linear	Connects points with straight lines; assumes a constant slope between data points.
splines	Uses smooth, low-degree curves (like cubics) to join points without sharp transitions.
polynomial	Fits a single high-degree equation to the data; good for global trends but prone to oscillations.
piecewise	Functions as a zero-order hold (ZOH); it holds the value of a point constant until the next one is reached, resulting in a step-like sequence.
idw	Inverse Distance Weighting; calculates unknown values based on the weighted average of nearby known points.