Mesh range, power and mode

[Damien Berger]

Hi, 

In this topic, i was asking about how to increase range in mesh networks. I did some outdoors tests (clear field), changing power (-40, +4, +8 dBm) and mode (1MB and Long Range). Here are my results using light switch example:

nRF52832 / 1MB / +4dBm : ~420 m

nRF52840 / 1MB / +8dBm : ~420 m                                       nRF52840 / 1MB / -40dBm : ~1.3 m

nRF52840 / LR / +8dBm    : ~422 m                                       nRF52840 / LR / -40dBm    : ~1.5 m

It seems that my implementation of the "+8dBm" (which is not defined in Nordic mesh examples) is not correct since i got the same range. I just added a line in the typedef enum "radio_tx_power_t" in radio_config.h. Do i have to do something more to make it work ? What did i miss ? 

/**
 * @brief The tx power configuration values.
 *
 * @note NRF52 does not have -30DBm setting, instead it can be configured to -40DBm,
 * the value of -40DBm and -30DBm are kept the same for code compatibility.
 */
typedef enum
{
    RADIO_POWER_NRF_POS8DBM  = 0x08,
    RADIO_POWER_NRF_POS4DBM  = 0x04,
#ifdef NRF52_SERIES
    RADIO_POWER_NRF_POS3DBM  = 0x03,
#endif
    RADIO_POWER_NRF_0DBM     = 0x00,
    RADIO_POWER_NRF_NEG4DBM  = 0xFC,
    RADIO_POWER_NRF_NEG8DBM  = 0xF8,
    RADIO_POWER_NRF_NEG12DBM = 0xF4,
    RADIO_POWER_NRF_NEG16DBM = 0xF0,
    RADIO_POWER_NRF_NEG20DBM = 0xEC,
    RADIO_POWER_NRF_NEG30DBM = 0xD8,
#ifdef NRF52_SERIES
    RADIO_POWER_NRF_NEG40DBM = 0xD8
#endif
} radio_tx_power_t;

Then, about Long Range, i followed all the steps defined in this thread and the increase in range is really small (about 2 meters out of 420m in +8dBm and 0,20m at -40dBm). 
Could we not expect better results with long range ? Would the difference be greater in indoors environment (I will test this soon, but I think I am missing something when I see my results...).

[Susheel Nuguru]

Hi Damien,

Which softdevice version does the Mesh SDK use? 

The long range feature was tested to work well on S140v6.0.0. I am not sure about the versions before. Also the change in radio tx power should very much influence the range. 

 

[Damien Berger]

I used The SDK for Mesh V2.0.1 which use S140 V6.0.0 (for nRF52840) :

http://infocenter.nordicsemi.com/index.jsp?topic=%2Fcom.nordic.infocenter.meshsdk.v2.0.1%2Findex.html&cp=4_1_0

[Susheel Nuguru]

How are you testing this. Is the devices under test surrounded by many other devices. I mean to ask if the environment is very radio noisy? That could affect the range as there could be a lot of packet collisions in air causing corruption and hence lot of packet drops.

[Damien Berger]

I was only using two devices during each test, environment is not radio noisy, as you can see in the following picture, it is quite clear:

[Susheel Nuguru]

That seems like a good spot to test long range with 8dBm. It  kind of looks the same as the testing place I used to test long range here. But looking at your results it does not look that the radio has changed TX power. Can you be absolutely sure by sniffing the radio registers that the NRF_RADIO-> TXPOWER register is set to the right value. Because the below results cannot be correct with the correct configuration

nRF52832 / 1MB / +4dBm : ~420 m

nRF52840 / 1MB / +8dBm : ~420 m 

nRF52840 / LR / +8dBm    : ~422 m 

 Also, which nRF52840 Rev IC are you testing with?

There is one thing I need to check regarding adding range delays to the mesh code for long range. I will check this an come back to you soon.