This project implements a minimal analog front-end and software reconstruction algorithm to track a dipole vector immersed in saline solution. Drawing from the Einthoven triangle and its bipolar leads, maps electric potential to a real-time vectorial representation, similar how ECG devices take cardiac signals.
The conduction and acquisition of biopotentials are non-observable phenomena, mastering these concepts typically requires a deep dive into abstract physiological theory. This project provides a tangible, real-time bridge between theory and practice. By measuring potential differences induced by a hand probe within a conductive saline medium, this software calculates and graphically reconstructs a single dipole orientation and magnitude in real time. This approach transforms invisible electrical fields into an intuitive visual format.
The physical setup mimics a standard three-lead frontal plane EKG configuration (RA, LA, LL) within a controlled environment.
- The Medium: A 500mL container filled with a 0.9% saline solution or a salt-water mixture.
- The Electrodes: Three terminal electrodes, which can be made from peeled wire, are placed in an equilateral triangle around the perimeter of the container.
- Peripheral Spacing: Peripheral electrodes are ideally separated by 120 to 150 mm.
- Anode: A wire is placed at the center of the container with a small exposed portion. This allows the system to show positive and negative voltages at the center of the projection lines.
- Cathode (Probe): An oscilloscope probe serves as the hand movable electrode. A wire with a 300Ω series resistor can also be used for testing.
- Power: The system is powered with 5V DC relative to the central terminal. This configuration is specifically chosen to minimize cathode corrosion within the electrolytic solution.
While high-performance differential ADCs or microcontrollers like the NXP K64F are recommended for precision, this repository provides a setup for easy testing using a Raspberry Pi Pico. The signal is pre-amplified by instrumentation amplifiers, biased at mid-supply, and processed in software to remove offset and gain before being streamed as comma-separated numeric values for real-time plotting.
project/
│
├── firmware/
│ └── Vecto_pico.py # Micropython program for Raspberry Pi Pico
│ └── demo_signal_pico.py #Fast test of Pico board and graph program
│ └── VECTO2_MK64F12_Project.hex # Binary file for FRDM K64F board
├── images/
├── kicad/ # Circuit files
├── software/
│ └── vecto_graph.py # Real-time plotting via Python and Matplotlib
└── README.md
The system is designed to provide high-fidelity signal acquisition while maintaining a simple interface for real-time analysis.
- Analog Front-End; True differential sensing to capture signals before ADC quantization.
- Digitization; The Raspberry Pi Pico ADC serves as a digitizer. A more reliable and precise version using the NXP K64F microcontroller.
- Visualization; A Python sketch allows for interaction with the circuit and real-time signal plotting.
The following components are required to replicate the basic experimental setup:
| Qty | Component | Notes |
|---|---|---|
| 2 | INA128 instrumentation amplifiers | See schematic |
| 1 | Raspberry Pi Pico V2 | |
| 12 | Resistors (1% preferred) | See schematic |
| 4 | Decoupling and filtering capacitors | See schematic |
| 1 | Isolated 5V Lab power supply | Also tested with SWM12-5-N adapter |
| 1 | 1X Oscilloscope probe | Also tested with a wire with 300 ohm series resistor |
| 1 | 500mL 0.9% saline solution | Also tested with salt-water mixture |
| 1 | Low profile plastic container | Peripheral electrodes separated between 120 to 150 mm |
The following components are required to replicate the improved performance setup version based on (FRDM K64F board):
| Qty | Component | Notes |
|---|---|---|
| 1 | MAX6018 | Voltage reference |
| 1 | MAX4053 | Multiplexer |
| 1 | K64F Microcontroller | FRDM K64F Board |
| - | Decoupling and filtering capacitors | See schematic |
| 1 | power supply | SWM12-5-N adapter |
| 1 | ITX0505SA | DC-DC Converter |
| 1 | 1X Oscilloscope probe | Also tested with a wire with 300 ohm series resistor |
| 1 | 500mL 0.9% saline solution | Also tested with salt-water mixture |
| 1 | Acrylic plastic container | Peripheral electrodes separated between 120 to 150 mm |
- Analog Domain: INA128 powered from an isolated 5V source, Two operational amplifiers couple the LI and LII signals, by the Kirchoff’s voltage law LI+LIII=LII, which allows the system to calculate LIII, However the measurement of the third value allows the system to reduce the zero error.
- Digital domain: Pico powered from USB- Improves noise immunity
- Reference coupling: Single-point connection between:
- INA128
REF5 V mid-voltage - Pico 3.3V mid-voltage This architecture minimizes ground loops and digital noise injection and centers ADC measurements.
INA128 gain is defined by:
- Gain ≈ 2.6
- Optimized for ±0.3 V input signals
- Ensures ADC full-scale utilization without saturation
Two independent programs facilitate the implementation of the system.
- vecto_pico.py Streams three values ( LI. LII, LIII) obtained by the ADC.
- demo_signal_pico.py A sketch for rapid testing of communication and visualization program. Load the sketch to the Raspberry Pi Pico and a demo signal are streamed by the USB port to the visualization program.
- Two ADC channels sampled independently
- Oversampling with averaging
- Offset removal is managed via a shared reference.
from machine import ADC
import time
# --- Configuration ---
VREF = 3.3
VMID = 1.4 #VREF/2 A bit less improve zero error.
GAIN = 2.6 #The analog gain in the instrumentation amplifier.- Analog and digital supplies are isolated except for a shared reference point
- Very high input impedance → suitable for low-power sensors
- To change the signal register speed, press ‘m’ in the keyboard in the graphic program.
- The time scale is the same, that allows to draw waveforms more easy, for example a wave of 100mS period can be draw in 1S.
Each serial line contains three comma-separated floating-point values: Raspberry Pi Pico: DI, DII, DIII. K64F: DI, DII, DIII,Calculated AvR, Calculated AvF, Calculated AvL, Sample_time (mS).
Example:
Raspberry Pi Pico:
0.01234, -0.01022,1.81200
K64F:
0.01234, -0.01022,1.81200,0.02545,0.78459,0.89156,0.84100
The program continuously streams a previously recorded ECG signal to the host graphical application through a USB serial port, allowing the firmware and host program to be tested without difficulty.
# --- Parameters ---
SCALE_FACTOR = 0.000435 #This value converts the table values to mS
TARGET_POINTS = 680 #Adjust the time scale, The target points parameter keeps the ECG signal within a normal scale, typically ranging between 600 and 700 points. This is because the table contains a fixed number of points, which must be adapted to the amount of data that can be effectively represented graphically.
The delay time parameter allows you to slow down the signal animation to observe vector behavior more clearly.
- Animation Speed: Adjusting the delay does not change the actual time scale of the signal.
- Accuracy: For example, if a signal period is 0.75s, the scale will always display 0.75s regardless of the animation speed. This ensures that any captured or "printed" signal maintains its physical temporal accuracy.
# --- Parameter ---
# Adjust this value (e.g., between 5-50) to slow the animation
utime.sleep_ms(5)- Flash the Pico: Use Thonny to upload the demo_signal_pico.py sketch to your Raspberry Pi Pico.
- Configure the Host: Open the host script vecto_graph.py in a text editor and update the serial port number to match your Raspberry Pi Pico’s port.
- Updating Parameters: To apply new changes to the Pico, close the host program, update the code, and re-upload the sketch to the board.
- Pause: Press the "P" key while the signal is running to pause the animation.
- Inspection: Move your mouse over the signal profile. A crosshair will track your pointer, and the vector graph will update dynamically to reflect the values at that specific point.
- Measurement: Click anywhere on the three signals to set a reference point.
- The bottom-right corner of the screen will display two numerical values.
- Click a second time at a different location to calculate the amplitude and time lapse (Δt) between the two selected points.
- High-baud serial acquisition and circular buffering for real-time plotting.
- Data history retention for 5 seconds with a scrollable timeline.
- Interactive controls: Pause/resume via keyboard ('p') and grid-based signal inspection.
- Multi-channel and vector visualization.
Using pyserial, numpy, matplotlib, and FuncAnimation. It continuously reads the serial output and maintains a fixed-length real-time window for inspection. The script:
- Continuously reads Pico serial output
- Maintains a fixed-length real-time window
- Allows retrospective inspection of stored data
- Supports interactive control via keyboard
Parameter Observed Input range ±0.3 V CMRR ~90 dB (INA128) Effective resolution ~10–11 ENOB ADC sampling stability Good (with averaging) Noise sensitivity Dominated by reference stability
- Pico ADC is not differential
- Overall ENOB limited by RP2040 ADC
- INA128 is not rail-to-rail amplifier.
- Requires careful reference grounding Adding DC-DC converter before the polarization electrodes and a precision voltage reference improves the noise reduction the K64F firmware includes 60Hz Notch filter.
The second part of this project consists of: Python host-side excecutable code. Bioamplified version. Zero correction and central terminal software. This will be documented separately.
- Profesor Diana Patricia Amador PhD.
- Profesor William Ricardo Rodríguez PhD.
- Hernan Bernal Mechanical Designer.