Heart Rate with the MAXIM 30102 sensor.

Measuring heart rate with the MAX30102, using the SpO2 configuration. The MAX30102 is a nice sensor made by MAXIM INTEGRATED and it is really useful and fun. In this simple project, I used it to measure the heartbeat, throught the SpO2 configuration, with 400 Sa/s, every 32 samples they are averaged so we have 12.5 Sa/s. The rate is processed and the output is visible on the monitor.

Introduction.

My university gave us a special Arduino shield for Arduino Uno, which is called UniGEng ExpBoard 2. The shield covers completely every single pin on the board so it's way easier to use and we don't have to use breadboards, jumpers and other components that would slow our project study. Here's the pin scheme of the shield:

The word "ossimetro" means oximeter which is the sensor MAX30102. As every project that involves registers, we will use the library Wire of Arduino, that works with A4 and A5 pins. Here's shown the board, the sensor is on the bottom left:

The sensor is the one on the bottom left. So, I won't show you any kind of circuit. It'll be just code and some datas I got from the project. I will show you how to filter datas, store them and process them.

Reading the registers of the MAX30102 with Wire library.

If we read the datasheet that the company published on the web thorught this link, we can see there are some specific istructions to communicate with the sensor, using the Wire library. The Wire library helps us because the sensor uses a I2C protocol. The I2C protocol is usually read with the Wire library because with the A4 and A5 pins both the SDA (data line) and SCL (clock line) are way easier to manage. At page 16 of the datasheet we can see how to communicate with the sensor, I implemented two easy functions to make reading and writing faster:

byte Read(byte reg, byte dati){
  Wire.beginTransmission(MAXIM);
  Wire.write(reg);
  Wire.endTransmission(false);
  Wire.requestFrom(MAXIM,dati);
  if(Wire.available()){
    return Wire.read();
  }
  else{ 
    return 0;
  }
  Wire.endTransmission();
}

void Write(byte reg, byte N) {
  Wire.beginTransmission(MAXIM);
  Wire.write(reg);
  Wire.write(N);
  Wire.endTransmission(); 
}

Once we have them we can configure the sensor to get the better signal we can from our body.

Sensor and output configuration.

If we look the datasheet we can read there are some registers that have bits dedicated to the configuration of the device and the data output (page 10 of the datasheet). Inside the setup() function we have five lines they use the function Write() to configure the data output and the sensor. I found that the best configurations (I bet you can do it better!) are the following:

• 0x0A SpO2 configuration: 01101111
  • Max range of the A/D converter, max samples frequency with 18 bits of resolution (400 Sa/s), max pulse width in order to get max number of bits.
• 0x08 FIFO Configuration: 11110000
  • Max possible average of samples, the sensor takes every 32 samples and make the average of them, then it sends it to the output. The fourth bit it's linked with the rollover of the FIFO, which means that the sensor doesn't wait for the user to read it but it just refresh it every time he can.

The last changes I decided to make in the registers was to toggle the 2nd most significant bit in the 0x02 Interrupt Enable 1 in order to read in Fifo_output function real values (sometimes the sensor was giving random numbers like 4.43..e+9 so i thought it was pretty weird).

What should I expect from the plotter?

These is an example of what you can get from the data registers:

what we can see is that we have an oscillating signal where the impulse is the process that comes from diastole and systole movements. To study the zero crossing we can use a moving average, with much samples as we need, in my case I chose 10, but you can decide how many as you like.

  ARRAY_RED[i]=DATA_R;
  if(i < 10) i++;
  else i=0;
  unsigned long SUM_R=0;
  for(int j=0; j<10; j++) {
    SUM_R += ARRAY_RED[j];
  } 
  SUM_R=SUM_R/10;
  

If you plot both the variables, the main signal and the moving average here's what you should get:

if you make the difference between the two variables here's what we get:

then we can study the zero crossing as the period with the algorithm we prefer, here's what I used (doesn't work well with high BPMs):

if (n==1) DeltaTime=millis();
  if(peaks>0){
    if(count && ((millis()-T)>333)){
      T=millis();
      n++;
      count=false;
    }
  }
  else {
    count=true;
  }
  
  if(n==15){
    unsigned long period=millis()-DeltaTime;
    SettingTime=millis()-SettingTime;
    float frequency=((float)(n-1)/(float)period)*1000*60;
    if (frequency<30){
      Serial.println("PROBLEMS WITH THE POSITION, TRY TO PLACE IT BETTER.");
    }
    else{
      Serial.print("BPM: ");
      Serial.println(round(frequency));
    }
    n=1;
    // Serial.println(SettingTime);
  }
  

Conclusion

The MAX30102 is a great sensor, indeed it costs around 10 euros (Amazon.com). Tho it's not so easy to develop a firmware which can use perfectly the sensor. My task at school was to use the SpO2 mode of the sensor even tho there's a dedicated mode (heart mode) which can be maybe easier than the SpO2 one. Anyway, I'm satisfied of the work and you can use the sensor with this firmware to measure the heart rate of a person at rest.