Which are the *best* NRF24L01+ modules?

[Oitzu]

I ordered some surface-mounted by accident.

The reading say:
NRF M
24L01+
1436AH

They seem to work fine, but if somebody has an packet-loss testing sketch i would test it.

[NeverDie]

@Fabien said:

@NeverDie : Can you please post the sketch you use to test your NRFs modules. I just recieve 10 today from Itead. Same packaging that the ones you recieve. I just want to compare result in same conditions.

OK, sure. It started out as RF toy code, and then I just evolved it. It contains a lot of commented out code that I haven't bothered to delete. If that gets in the way of your understanding, just delete the code that's commented out. Aside from that, it's straightforward.

Here's the main transmitter code. After compiling and uploading, you should open a serial window on your computer to read the statistics it prints out:

/*
 nRF24Sender Demo for RFToy
 
 This demo shows how to use RFToy to make a
 wireless temperature sensor. This is the
 sender module which transmits the current
 temperature value to a receiver module. The
 demo uses the Mirf library.
 
 This demo uses a 100K resistor and 100K
 thermistor to form a simple temperature 
 sensor. Pin A1 is used to read the value. 
 The connection is:
 VCC->100K->A1->thermistor->GND
  
 Written by Jonathan Goldin @ Rayshobby LLC
 Nov 2014
 For details, visit http://rayshobby.net/rftoy
*/

#include <SPI.h>
#include <Mirf.h>
#include <nRF24L01.h>
#include <MirfHardwareSpiDriver.h>
#include <U8glib.h>

U8GLIB_SSD1306_128X64 u8g(U8G_I2C_OPT_NONE);	// I2C / TWI 

void setup(){
  Serial.begin(115200);
  Serial.println("Starting...");
   
  /*
   Set ce and csn pins
   */
  
  Mirf.cePin = 17;
  Mirf.csnPin = 16;
  
  Mirf.spi = &MirfHardwareSpi;
  Mirf.init();
  
  /*
   * Configure reciving address.
   */
   
  Mirf.setRADDR((byte *)"clie1");
  
  /*
   * Set the payload length to sizeof(unsigned long) the
   * return type of millis().
   *
   * NB: payload on client and server must be the same.
   */
   
  //Mirf.payload = sizeof(long);
  Mirf.payload = sizeof(long);  
  /*
   * Write channel and payload config then power up reciver.
   */
   
  /*
   * To change channel:
   * 
   * Mirf.channel = 10;
   *
   * NB: Make sure channel is legal in your area.
   */
   
    // we use channel 90 as it is outside of WLAN bands 
  // or channels used by wireless surveillance cameras 
  //Mirf.channel = 90;
  
  Mirf.config();
  
    
  //This register value is not remembered between power cycles.
  //It defaults to 0x0F.
  //It should be initialized each time if different than 0x0F.
  Mirf.configRegister(RF_SETUP,0x07); //0x0F is 2mbps, max Tx power 
                                      //0x07 is 1mbps, max Tx power
                                      //0x2F is 250kbps, max Tx power.
  Serial.println("OTA datarate set to  1Mbps.  Transmit Power set to Maximum.");
  
  // Read and print RF_SETUP
 byte rf_setup = 0;
 Mirf.readRegister( RF_SETUP, &rf_setup, sizeof(rf_setup) );
 Serial.print( "rf_setup = " );
 Serial.println( rf_setup, BIN );
  // OLED
  u8g.firstPage();
  do{
    uint8_t h;
    u8g.setFont(u8g_font_10x20);
    u8g.setFontRefHeightText();
    u8g.setFontPosTop();
    h = u8g.getFontAscent()-u8g.getFontDescent();
    u8g.drawStr(29,(u8g.getHeight()-h)/2,"Tx Sender");
  } 
  while(u8g.nextPage());
  Mirf.setTADDR((byte *)"serv1");
  
  Serial.write("Sending...\r\n"); 
  delay(200);
} // End of *Setup*


long temp;
  int temp1;
  int temp2;
  long timeTxSent;
  long timeRxReceived;
  long roundTrip;
  byte age1=52;
  byte age2=11;
  long txCounter=0;
  long matchCount=0;
  long differentCount=0;
  long lostCount=0;
  long cumulativeRoundTrip=0;
  long averageRoundTrip=0;
  boolean packetLost=false;
  float packetErrorRate=0;  //no errors yet, and maybe there never will be.
  float lostPacketRate=0;  //no packets lost yet.
  const int statusFrequency=500;  //How many iterations of main loop before printing status info.
  long minRoundTrip=9999; //value will be driven down when program runs
  long maxRoundTrip=0;  //value will be driven up when program runs
  
void loop(){
   
  
  
    
  
 
  txCounter++;
  packetLost = false;  //It can't be lost, because it hasn't even been sent yet.
  
  temp = txCounter;  //getTemp(resistance);
  temp1=temp;
  
  timeTxSent=micros();
  Mirf.send((byte *)&temp);  
  
  
  

  while(Mirf.isSending()){
  }

  /*
  Serial.write("temp=");
  Serial.print(temp,DEC);
  Serial.write("\r\n");
  Serial.write("temp1=");
  Serial.print(temp1,DEC);
  Serial.write("\r\n");
  Serial.write("Finished sending.\r\n");
  */
  //delay(10);
  unsigned long time = millis();
  while ((!packetLost) && (!Mirf.dataReady())){
    //Serial.println("Waiting");
    if ( ( millis() - time ) > 8 ) {
      //Serial.println("Timeout on response from Rx Echo Reflector!");
      lostCount++;
      packetLost=true;
    }
  }
  
  if (!packetLost) {
  Mirf.getData((byte *) &temp);  
  timeRxReceived=micros();
  temp2 = temp;  
  roundTrip =  timeRxReceived - timeTxSent;
  if (roundTrip < minRoundTrip) {
    minRoundTrip=roundTrip;
  }
  if (roundTrip > maxRoundTrip) {
    maxRoundTrip = roundTrip;
  }
  if(temp1 == temp2){  
    matchCount++;
    cumulativeRoundTrip += roundTrip;
    averageRoundTrip = cumulativeRoundTrip/(txCounter-lostCount-differentCount);
  } else {
    differentCount++;
    Serial.println("***DIFFERENT**");
      Serial.write("temp1=");
      Serial.println(temp1, BIN);
      Serial.write("temp2=");
      Serial.println(temp2, BIN);
  }
  
  if ((txCounter%statusFrequency)==0) {
    /*
    if(temp1 == temp2){  
      Serial.print("Match");
    } 
    else {
      Serial.println("***DIFFERENT**");
    }
    Serial.write(",");
    */
  }
  }
  
  lostPacketRate = 100*((float)(lostCount))/((float)txCounter);
  
  if ((txCounter%statusFrequency)==0) {
   
    Serial.print(txCounter);
    Serial.write(",lost=");
    Serial.print(lostPacketRate);
    /*
    Serial.write("%,T=");
    Serial.print(temp,DEC);
    //Serial.write(".  ");
    Serial.write(",T1=");
    Serial.print(temp1,DEC);
    //Serial.write(".  ");
    Serial.write(",T2=");
    Serial.print(temp2,DEC);
    */
    Serial.write("%,RT=");
    Serial.print(roundTrip,DEC);
    Serial.write(",minRT=");
    Serial.print(minRoundTrip);
    Serial.write(",maxRT=");
    Serial.print(maxRoundTrip);
    Serial.write(",aRT=");
    Serial.print(averageRoundTrip,DEC);
    Serial.print(",#lost=");
    Serial.print(lostCount);
    //Serial.write(",mat=");
    //Serial.print(matchCount);
    Serial.write(",diff=");
    Serial.print(differentCount);
    Serial.write("\r\n");
    delay(100); //give time for it to print out
    //txCounter = 0;  //restart gathering statistics
  }

  
  /*
  delay(1000);  // keep the 'sending' message displayed on OLED for 1 sec
  u8g.firstPage();
  do{
  } while(u8g.nextPage());

  delay(2000);  // wait for 2 seconds till next transmission
  */
}  //End of main loop. 

Here's the code for the receiver node. It doesn't need to be plugged into a computer:

/*
 nRF24Receiver Demo for RFToy
 
 This demo shows how to use RFToy to make a
 wireless temperature sensor. This is the
 receiver module which displays the received
 temperature value to OLED. The demo uses
 the Mirf library.
 
 Written by Jonathan Goldin @ Rayshobby LLC
 Nov 2014
 For details, visit http://rayshobby.net/rftoy
*/

#include <SPI.h>
#include <Mirf.h>
#include <nRF24L01.h>
#include <MirfHardwareSpiDriver.h>
#include "U8glib.h"


U8GLIB_SSD1306_128X64 u8g(U8G_I2C_OPT_NONE);	// I2C / TWI 

void setup(){
  Serial.begin(115200);
  Serial.println("Echo Receiver.  Listening");

  Mirf.cePin = 17;  //???
  Mirf.csnPin = 16;  //???
  /*
   * Set the SPI Driver.
   */

  Mirf.spi = &MirfHardwareSpi;

  /*
   * Setup pins / SPI.
   */

  Mirf.init();

  /*
   * Configure reciving address.
   */

  Mirf.setRADDR((byte *)"serv1");

  /*
   * Set the payload length to sizeof(unsigned long) the
   * return type of millis().
   *
   * NB: payload on client and server must be the same.
   */

  Mirf.payload = sizeof(long);

  /*
   * Write channel and payload config then power up reciver.
   */


 // we use channel 90 as it is outside of WLAN bands 
  // or channels used by wireless surveillance cameras 
  //Mirf.channel = 90;
  
  Mirf.config();
    //This register value is not remembered between power cycles.
  //It defaults to 0x0F.
  //It should be initialized each time if different than 0x0F.
  Mirf.configRegister(RF_SETUP,0x07); //0x0F is 2mbps, max Tx power 
                                      //0x07 is 1mbps, max Tx power
                                      //0x2F is 250kbps, max Tx power.
  Serial.println("OTA datarate set to 1Mbps.  Transmit Power set to Maximum.");
  
  // Read and print RF_SETUP
 byte rf_setup = 0;
 Mirf.readRegister( RF_SETUP, &rf_setup, sizeof(rf_setup) );
 Serial.print( "rf_setup = " );
 Serial.println( rf_setup, BIN );

  
  u8g.firstPage();
  do{
    uint8_t h;
    u8g.setFont(u8g_font_10x20);
    u8g.setFontRefHeightText();
    u8g.setFontPosTop();
    h = u8g.getFontAscent()-u8g.getFontDescent();
    u8g.drawStr(19,(u8g.getHeight()-h)/2,"Echo Rx");
  } 
  while(u8g.nextPage());
  
}  //End of *Setup* procedure

void loop(){
  /*
   * A buffer to store the data.
   */

  byte data[Mirf.payload];

  /*
   * If a packet has been recived.
   *
   * isSending also restores listening mode when it 
   * transitions from true to false.
   */

  if(!Mirf.isSending() && Mirf.dataReady()){
    //Serial.print("Got packet: ");

    /*
     * Get load the packet into the buffer.
     */

    Mirf.getData(data);

    
    // Set the send address.
    Mirf.setTADDR((byte *)"clie1");

    /*
     * Send the data back to the client.
     */

    Mirf.send(data);

    /*
     * Wait untill sending has finished
     *
     * NB: isSending returns the chip to receving after returning true.
     */
     
    

    //Serial.println("Reply sent.");
  }
}

As background, here's a link to the RFToy:
http://rayshobby.net/rftoy/
Links to the RFToy library, as well as the hardware design, can be found there. It's all open source.

This shows the pin assignments: https://github.com/rayshobby/rftoy-hw/blob/master/RFToy.png

You'll need a pinout diagram to match-up those internal pins to the physical pins of whatever Arduino you're using. For instance, for an Uno, here's a pinout diagram which maps internal pins to physical pins:
http://marcusjenkins.com/wp-content/uploads/2014/06/ARDUINO_V2.png
That way you'll know how to properly wire-up your NRF24L01+ so that it works properly with the library code.

So, I did that just now and tested it on the UNO, and so for the UNO the simplified wiring directions are:
NRF24L01+Pin --- --> Uno Female Header Pin
GND (1) ----------------------------> GND
VCC (2) -----------------------------> 3.3V
CE (3) ------------------------------> A3
CSN (4) -------------------------------> A2
SCK (5) --------------------------------> D13
MOSI (6) --------------------------------> D11
MISO (7) --------------------------------> D12
IRQ(8) ---------------------------------> n/a

The RFToy has an OLED screen that gets written to. You can remove that code if you wish, but leaving it in does no harm, even if you don't have an OLED screen on your arduino. I modified the code so that it's only written to during the setup loop, so regardless it shouldn't interfere with any of the measurements taken in the main loop..

Hope that helps!

[Fabien]

I just finish some tests with your sketch. With ITead Modules, I have 0.05% packet loss at 1mbps and about 15 meters with 1 wall.
aRT is 1700. With NRF from electrodragon, I have same results aRT is a bit higher eg 2000.
But during my tests, I found some issues :

• With UNO I have more lost packets without extra capacitor. I think 3.3V output is not enough good. With Arduino Nano, it's ok
• Quite different results when touching my dupont wires ...
• When I let temp1 and temp2 output, there's a bug with synchro. It says it's always different but it's only dure to a shift with comparaison. If i just comment serial.output of temp1 an temp2 diff is always equal to 0

I will make a test with RF24 from TMRh20 (used in MySensors).
About Mysensors, I have a question : is there a configuration to retry packet sending and delay bewteen resend like radio.setRetries(2,15); ?

So for me the Itead modules are good and electrodragon too.

[hek]

@Fabien said:

is there a configuration to retry packet sending and delay bewteen resend like radio.setRetries(2,15); ?

No, there isn't currently. You can always run a gw.setRetries() after gw.setup() has been called,

[Fabien]

@hek : I just look in the library and see there is a retry RF24::setRetries(5,15); in MySensor::setupRadio and it is called by MySensor::begin

[hek]

Yes, thought you meant if there were any config options to set this...

[NeverDie]

@Fabien said:

I just finish some tests with your sketch. With ITead Modules, I have 0.05% packet loss at 1mbps and about 15 meters with 1 wall.
aRT is 1700. With NRF from electrodragon, I have same results aRT is a bit higher eg 2000.
But during my tests, I found some issues :

• With UNO I have more lost packets without extra capacitor. I think 3.3V output is not enough good. With Arduino Nano, it's ok
• Quite different results when touching my dupont wires ...
• When I let temp1 and temp2 output, there's a bug with synchro. It says it's always different but it's only dure to a shift with comparaison. If i just comment serial.output of temp1 an temp2 diff is always equal to 0

I will make a test with RF24 from TMRh20 (used in MySensors).
About Mysensors, I have a question : is there a configuration to retry packet sending and delay bewteen resend like radio.setRetries(2,15); ?

So for me the Itead modules are good and electrodragon too.

Interesting! Thanks for reporting the results. I'm sure Itead will be happy to hear about it. :smile: BTW, what lettering is printed on the NRF24L01+ chips? Also, from where did you procure the modules (Itead headquarters?) and how long ago?

Prior to this morning, I had been using only an RFToy (which is effectively an 8Mhz 3.3V Arduino Mini Pro), and differences between the number transmitted (temp1) and the number echoed back (temp2) were fairly rare. Because probably no one but me has an RFToy, I setup and tested using an Uno this morning to verify that I was giving the correct wiring instructions (above). I did notice a much greater rate of differences reported by the Uno than the RFToy, but I didn't pay attention to the details. From what you're saying, it sounds like it hasn't finished shifting all the bits from the Rx buffer before comparing temp1 to temp2? Am I understanding you right? If so, you could try adding, say, a 2ms delay before doing the comparison to allow the received data to fully settle before the comparison. It wouldn't affect the roundtrip timing, because it would be happening after the elapsed microseconds had been recorded. Without looking into it, I had thought that my use on the Uno of fairly long dupont jumper cables was injecting noise. On the RFToy, the connections are shorter, and no jumper wires are used.

What size capacitor (how many Farads) have you found works best? What type (e.g. electrolytic, ceramic, film, etc.)?

If you move the code over to TMRh20, would you mind posting it back to this thread? I think more people will be familiar with it than with the Mirf library. I only used Mirf because the RFToy code used it, and for simplicity that was my starting point.

Is this the electrodragon NRF24L01+ module that you tested? http://www.electrodragon.com/product/nrf24l01/

[Fabien]

No I use this one : http://www.electrodragon.com/product/nrf24l01b-tw-si24ri-nrf24l01-fully-compatible/

[hek]

@Fabien
Gotta love their product description...

"The main advantage of it is cheap."

[Fabien]

@hek : Yes but I havn't notable difference with the one from itead ! And you know it is clone.
@NeverDie : The first 2 chip from Itead : 1443IA , not a dot but a square and too small to see if there is a hole inside the +. I think i don't investigate more with others sketch because with my dupont cable error rate can increase a lot when just touching a cable.
I just setup a Iboard with PA+LNA from electrodragon and all my sensebender with clone from electrodragon too, and everything is working fine, up 20 meters without any problems.
For security I will change my NRF with the ones from Itead and will keep my PA+LNA for milight gateway emulation.

[NeverDie]

@Fabien said:

@hek : Yes but I havn't notable difference with the one from itead ! And you know it is clone.

Move your test nodes further apart, and keep increasing the distance until you start losing a higher percentage of packets. Most likely, as the separation distance and/or RF impairments increase, at some point one module or the other is going to perform much better than the other--that is unless they're using the same type of NRF chip. At least to date that's been my experience with the very few datapoints I've collected so far.

[Moshe Livne]

@NeverDie If/when you get contact with nordic rep, it would be great if they can provide a sketch that does the testing. they probably have one...

[NeverDie]

@Moshe-Livne said:

@NeverDie If/when you get contact with nordic rep, it would be great if they can provide a sketch that does the testing. they probably have one...

What kind of testing?

[Moshe Livne]

@NeverDie

1. Authentic module testing (they might be reluctant to give something like this as it can probably be used to make the copies closer to the original - but worth the try)
2. QA testing - packet drop, etc at different speeds

[NeverDie]

@Moshe-Livne said:

@NeverDie

1. Authentic module testing (they might be reluctant to give something like this as it can probably be used to make the copies closer to the original - but worth the try)
2. QA testing - packet drop, etc at different speeds

OK. I haven't yet been contacted by Nordic, and I don't know how that conversation will unfold. If there' is room for Q&A, then I'll be sure to ask them your questions as well as relay their answers back to you.

[GIEL]

@NeverDie

NORDIC has on their Website the option to raise questions (MyPage) which are handled by their Tech Support Team. Normally they reply fast.

[NeverDie]

@GIEL said:

@NeverDie

NORDIC has on their Website the option to raise questions (MyPage) which are handled by their Tech Support Team. Normally they reply fast.

Thanks! I tried it just now, so hopefully I will get a response soon.

Today I ordered some RFM69's. Unit cost is higher, but shipping cost is so much lower that the total cost is actually lower. If they test out better, then I may just go that route instead. In fact, if they test out better. then is there any reason to prefer the NRF24L01+?

[NeverDie]

I received the NRF24L01+'s that are on the red PCB's (above), and when I saw they were using the now notorious 1242AF chips, I had little hope. However, I tested them at 1mbps over the same challenge distance as the others, and so far they're doing very well: I transmitted over 200,000 packets, and there were only 0.03% lost packets. Average round trip time was 2.2ms.

As before, I'm using the RFToys to do the testing. The modules seem more finicky about their orientation than others that I've tested, and moving things just a little can make for much, much worse results.

I bought them from MDFly on ebay.

[NeverDie]

I received more blob modules, and they continue to impress. me. I sent 160,000 packets, and there's only .22% packets lost with an average roundtrip time of 2.2ms. Also, the module doesn't seem particularly sensitive as to its orientation. For that reason, I like them more than the red modules.

[Yveaux]

I did some current consumption testing on my modules and the results were quite surprising.
What should happen is that the current consumption rises during transmission, then stays high until transmission is finished.
Most modules however show very deep spikes in current consumption during transmission.
This behavior does not seem to be chip related, more module related (the green ones perform best in this respect).
It could be caused by the board layout and/or components used.
My HF knowledge is very limited, so maybe anyone of you have any ideas?