I'm using nRF24L01 to send ECG samples from an Arduino to Raspberry Pi, the type of sample is double after denoising, and I stock them in a byte[8] to send 8 by 8 sample. When I print the sample before stockage that show the good results, but if I print data stored in the byte the signal show a lot of fluctuations. In Raspberry Pi, I can read correctly byte data like transmitted, but if I send them in a vector of double or int in reception (python) don't give the same frame transmitted.

#include <IIRFilter.h>
#include <FIRFilter.h>

#include <SPI.h> // Call SPI Library 
#include <nRF24L01.h> // Call NRE Version Library
#include <RF24.h> // RF24 Header File

#define CE_PIN  9
#define CSN_PIN 10
const uint64_t pipe = 0xF0F0F0F0E1LL;
RF24 radio(CE_PIN, CSN_PIN);

const int ECG_samplefreq  = 200;//sampling frequcy
const int DC_offset = 511;

const double b_lp[] = {0.0026, 0.0155, 0.0388, 0.0517, 0.0388, 0.0155, 0.0026};
const double a_lp[] = {1.0000, -2.3797, 2.9104, -2.0551, 0.8779, -0.2099, 0.0218};

IIRFilter lp(b_lp, a_lp);
unsigned int i = 0;

void setup() {
  radio.begin();//Intializing NRF24L01 Module
  radio.openWritingPipe(pipe);// For Transmitting Operation, Must OpenWritingPipe
  pinMode(7, INPUT); // Setup for leads off detection LO +
  pinMode(8, INPUT); // Setup for leads off detection LO -

void loop() {
  const static unsigned long ECG_interval  = round(1e6 / ECG_samplefreq);  //sampling period = 5000 ms => 200 sample by 1 seconde
  static unsigned long ECG_prevmicros = micros();
  byte  s[8] = {0};

  if ((digitalRead(7) == 1) || (digitalRead(8) == 1)) {

  else {
    for (int i = 0; i < 8; i++)

      if (!Serial) {
        ECG_prevmicros = micros();
      else if (micros() - ECG_prevmicros >= ECG_interval)
        int value = analogRead(A5);
        double filtered = lp.filter( value - DC_offset);

        filtered = round( filtered + DC_offset);
        s[i] =  filtered
        Serial.println( s[i]);
        ECG_prevmicros += ECG_interval;

    radio.write(&s, sizeof(s) ); // sending in real time ECG sample. 8 sample by transmission . 8 => 4 oct of one sampe = 32 oct by transmission.


the problem thant i put in s[i] a double of four octs. im searching how can i send them whithout byte array, and receive the correctly. So thankful for your help

  • I am not a math genius but when I get you right you are trying to send 8 double values as a series of 8 x 8 bytes. My calculator tells me the result is 64 but you can only transmit 32 bytes at once using nRF24L01+. So maybe you should provide us some code and describe its function to us so we will be able to understand your problem and provide some help. – Kwasmich Apr 3 '19 at 11:13
  • Even if the OP is trying to send one double value per message (8 x 8 bytes), does one really need all that precision? Perhaps they can round off the double values to fit more data values within a 32-byte data packet. Show the code and explain a little more what is going on. – MichaelT Apr 3 '19 at 19:10

I see multiple flaws in your code:

  • radio.write(&s, sizeof(s)); does not what you think it would. The correct statement would be radio.write(s, sizeof(s)); without the ampersand.
  • You are using double. Not using floating point numbers at all would be preferable.
  • Your s is of type byte. Thus it only holds 8 bytes. So your floating point calculations are being truncated to the range [0, 255]. From your problem description this is not what you are trying to achieve, I think.
| improve this answer | |
  • Yes you're right. I mislaid in the storage in the byte variable s[8]; I changed the type of samples to int32_t; similarly the type of byte emission vector to int32_t s[8]. the problem now at the raspberry level I get a 32-bit list instead of 8 samples int32_t – Amina Elattaoui Apr 5 '19 at 10:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.