Design a clinical-accuracy pulse oximeter using AFE4490 analog front end, dual-wavelength LED, and Beer-Lambert law oxygen saturation algorithm.
Design a clinical-accuracy pulse oximeter using AFE4490 analog front end, dual-wavelength LED, and Beer-Lambert law oxygen saturation algorithm. This comprehensive guide covers everything from design through implementation, testing, and deployment.
Beer-Lambert Law: light attenuation through tissue depends on absorber concentration. Oxygenated hemoglobin (HbO2) absorbs IR more than red. Deoxygenated Hb absorbs red more than IR. At 660nm (red): Hb absorbs 10× more than HbO2. At 940nm (IR): HbO2 absorbs more. PPG (Photoplethysmography): during systole, arterial blood volume in finger increases → more light absorbed. During diastole: less absorption. AC component = pulsatile blood. DC component = tissue + venous blood. SpO2 = f(R) where R = (AC_red/DC_red) / (AC_IR/DC_IR). Calibration empirical equation: SpO2 = -45.060 × R² + 30.354 × R + 94.845.
10 components required for this project.
| # | Component | Purpose | Qty |
|---|---|---|---|
| 1 | AFE4490 (Texas Instruments Pulse Oximeter AFE) | LED driver + transimpedance amplifier + ADC | x1 |
| 2 | Red LED (660nm, 5mW) | Oxygenated/deoxygenated hemoglobin discrimination | x1 |
| 3 | Infrared LED (940nm, 5mW) | Second wavelength measurement | x1 |
| 4 | Photodiode (BPW34, broadband) | Transmitted light detection | x1 |
| 5 | STM32L4 (low power MCU) | SpO2 algorithm processing | x1 |
| 6 | 0.96" OLED display (SSD1306) | SpO2 and HR display | x1 |
| 7 | Finger clip probe housing (3D printed) | Optical path and tissue contact | x1 |
| 8 | LiPo 100mAh + charger | Portable power | x1 |
| 9 | Band-pass filter (0.5–4 Hz) | Heart rate signal isolation | x1 |
| 10 | Arduino for prototyping (MAX30100 module first) | Algorithm prototyping before custom AFE | x1 |
Follow these 4 steps carefully.
Beer-Lambert Law: light attenuation through tissue depends on absorber concentration. Oxygenated hemoglobin (HbO2) absorbs IR more than red. Deoxygenated Hb absorbs red more than IR. At 660nm (red): Hb absorbs 10× more than HbO2. At 940nm (IR): HbO2 absorbs more. PPG (Photoplethysmography): during systole, arterial blood volume in finger increases → more light absorbed. During diastole: less absorption. AC component = pulsatile blood. DC component = tissue + venous blood. SpO2 = f(R) where R = (AC_red/DC_red) / (AC_IR/DC_IR). Calibration empirical equation: SpO2 = -45.060 × R² + 30.354 × R + 94.845.
AFE4490: complete analog front end — drives LEDs (programmable current 0–50mA) and processes photodiode signal. LED timing: alternating LED activation — RED → blank → IR → blank. Sample rate: typically 250 Hz (programmable). Transimpedance amplifier: converts photodiode current to voltage. Programmable gain (2kΩ–1MΩ feedback). 22-bit ADC. SPI communication. Output: 32-bit signed values for RED and IR channels. Signal chain: LED driver → tissue → photodiode → TIA → ADC → MCU. Key requirement: synchronize LED timing to measurement timing (only sample during LED-ON phase).
Extract AC and DC from raw samples: DC component = low-pass filter (moving average over 10s). AC component = bandpass filter (0.5–4 Hz — heart rate range). Peak detection: find peaks in red and IR AC signals. Calculate per-beat: AC_red = peak-to-peak red amplitude, DC_red = mean red. Similarly for IR. R ratio = (AC_red/DC_red) / (AC_IR/DC_IR). Look up SpO2 from calibration equation (or table). Average over 4–8 beats for stability. Heart rate: count peak-to-peak intervals, convert to BPM = 60 / period_seconds.
Clinical standard: SpO2 accuracy ±2% (ISO 80601-2-61). Accuracy factors: motion artifact (arm movement causes false AC signals — motion detection using accelerometer, reject noisy beats), ambient light (AFE4490 measures with LEDs off to subtract ambient light), sensor positioning (correct finger placement critical — too loose or too tight reduces signal quality), probe design (wavelengths must match calibration curves), skin pigmentation (melanin absorbs some light — recalibrate for different populations). This is an educational device — not for clinical use without regulatory approval (CE/FDA Class II medical device).
Core code for spo2_algorithm.c:
// SpO2 Calculation Algorithm // Input: Raw red and IR samples from AFE4490 at 100Hz #include <stdint.h> #include <math.h> #define SAMPLE_RATE 100 #define BUFFER_SIZE 100 // 1 second of data float red_buf[BUFFER_SIZE], ir_buf[BUFFER_SIZE]; int buf_idx = 0; // Simple DC extraction (moving average) float dc_filter(float *buf, int n) { float sum = 0; for(int i = 0; i < n; i++) sum += buf[i]; return sum / n; } // Find peak-to-peak amplitude (AC component) float ac_amplitude(float *buf, int n) { float max = buf[0], min = buf[0]; for(int i = 1; i < n; i++) { if(buf[i] > max) max = buf[i]; if(buf[i] < min) min = buf[i]; } return max - min; } float calculate_spo2(void) { float dc_red = dc_filter(red_buf, BUFFER_SIZE); float dc_ir = dc_filter(ir_buf, BUFFER_SIZE); float ac_red = ac_amplitude(red_buf, BUFFER_SIZE); float ac_ir = ac_amplitude(ir_buf, BUFFER_SIZE); if(dc_red < 50000 || dc_ir < 50000) return -1; // No finger detected float R = (ac_red / dc_red) / (ac_ir / dc_ir); // Calibration equation (device-specific, calibrate against reference) float spo2 = -45.060f * R * R + 30.354f * R + 94.845f; return fmaxf(70, fminf(100, spo2)); } int calculate_heart_rate(float *ir_buf, int n) { // Simple peak detection - count peaks in IR signal int peaks = 0; float threshold = dc_filter(ir_buf, n); bool above = false; for(int i = 1; i < n-1; i++) { if(ir_buf[i] > threshold && !above) { peaks++; above = true; } else if(ir_buf[i] <= threshold) above = false; } return peaks * (60 * SAMPLE_RATE / n); // BPM }Test Pulse Oximeter Design (SpO2) by verifying each subsystem individually before full integration.
Verify power voltages, check ground connections, use serial monitor for debug.
An interactive simulator will be available here — simulate circuits and run code in-browser without hardware.